KR20200005409A - Fundus image management device and method for determining suitability of fundus image - Google Patents

Abstract

The present invention provides a method for determining the quality of a fundus image which comprises the steps of: comparing a pixel value of at least one pixel included in a target area of the fundus image with a reference pixel value to detect a first pixel related to the quality of the fundus image; comparing a ratio of the first pixel of the total pixels of the fundus image with a reference ratio; and acquiring quality information of the fundus image based on a comparison result obtained by comparing a ratio of the first pixel of all pixels of the fundus image with the reference ratio.

Description

FUNDUS IMAGE MANAGEMENT DEVICE AND METHOD FOR DETERMINING SUITABILITY OF FUNDUS IMAGE}

The present invention relates to a fundus image management apparatus and a method for determining the suitability of a fundus image, and to a fundus image management apparatus and a method for determining suitability of a fundus image for determining suitability for a target disease of a fundus image.

Fundus examination is a diagnostic aid that is frequently used in ophthalmology because it can observe the abnormalities of the retina, optic nerve, and macular area. Recently, the use of the fundus examination is increasing the use of the non-invasive method to observe the degree of vascular damage caused by chronic diseases such as hypertension, diabetes, as well as eye diseases.

On the other hand, with the rapid development of the recent deep learning technology, the development of diagnostic artificial intelligence in the field of medical diagnostics, especially image-based diagnostics is being actively made. Global companies such as Google and IBM are also investing in the development of artificial intelligence for analyzing various medical image data, such as inputting large-scale data in collaboration with the medical field, and some companies produce artificial intelligence diagnostics that output excellent diagnostic results. He also succeeded in developing the tool.

However, when the AI model is trained using the fundus image, or when the diagnosis is based on the fundus image using the AI model, it is difficult to expect accurate learning or diagnosis results when the fundus image contains defects. there is a problem. Accordingly, there is a need for a method for determining the quality of the fundus image.

One object of the present invention is to provide a method for determining suitability for a target disease of an ocular fundus image and an apparatus for performing the same.

One object of the present invention is to provide a method for determining suitability for a target region of a fundus image and an apparatus for performing the same.

The problem to be solved by the present invention is not limited to the above-described problem, the objects that are not mentioned will be clearly understood by those skilled in the art from the present specification and the accompanying drawings. .

According to an aspect of the present invention, in the method for determining the suitability of the fundus image in relation to the diagnosis of the target disease based on the fundus image, the quality criteria corresponding to the target disease with respect to the step of acquiring the fundus image and the fundus image Determining a subject fitness of the fundus image using the subject disease comprising at least one of a first disease and a second disease; Including, but determining the object suitability, if the target disease is the first disease determines whether the fundus image meets the first quality criteria, if the target disease is the second disease whether the fundus image meets the second quality standards There may be provided a method for determining suitability of a fundus image comprising determining.

According to another aspect of the present invention, in the method of determining the suitability of the fundus image in relation to the diagnosis of the target disease based on the fundus image, acquiring the fundus image, determining whether the fundus image satisfies the quality criteria And determining, based on determining whether the ocular fundus image satisfies a quality criterion, the subject suitability of the ocular fundus image for a target disease, wherein determining whether the ocular fundus image satisfies a quality criterion comprises: a method associated with the first region; Determining whether the first quality criterion is satisfied, and whether the fundus image satisfies a second quality criterion associated with a second region that is at least partially distinguished from the first region, and determining the object suitability, If the target disease is a first disease and the fundus image meets a first quality criterion, the fundus image is of first suitability. It determined to be satisfied, and there is a target disease of the second disease and fundus image can be provided in conformity judgment process of the fundus image, comprising: determining that satisfy the fundus image of the second compliance if they meet the second quality criterion.

According to another aspect of the present invention, in the fundus image management apparatus for determining the suitability of the fundus image in relation to the diagnosis of the target disease based on the fundus image, with respect to the fundus image acquisition unit and the fundus image for acquiring the fundus image A suitability determination unit configured to determine a target suitability of the fundus image using a quality standard corresponding to the target disease, wherein the target disease includes at least one of a first disease and a second disease; Including, but, the suitability determination unit, the fundus to determine whether the fundus image satisfies the first quality criteria when the target disease is the first disease, and the fundus image to determine whether the fundus image meets the second quality criteria when the target disease is the second disease An image management apparatus may be provided.

According to another aspect of the present invention, in the fundus image management apparatus for determining the suitability of the fundus image in connection with the diagnosis of the target disease based on the fundus image, the fundus image acquisition unit for acquiring the fundus image, the quality of the fundus image And a quality determining unit determining a target suitability for the target disease of the fundus image based on determining whether the ocular fundus image satisfies a quality criterion, wherein the quality determining unit comprises: a first determination associated with the first area; Determine whether the quality criterion is satisfied, and whether the fundus image satisfies a second quality criterion associated with a second region that is at least partially distinguished from the first region, and the suitability determination unit determines that the target disease is the first disease If the fundus image satisfies the first quality criterion, it is determined that the fundus image satisfies the first suitability. An ocular fundus image management apparatus may be provided that determines that the ocular fundus image satisfies the second suitability when the ocular disease is a second disease and the ocular fundus image satisfies the second quality criterion.

Means for solving the problems of the present invention are not limited to the above-described solutions, and the solutions not mentioned will be clearly understood by those skilled in the art from the present specification and the accompanying drawings. Could be.

According to the present invention, the quality of the fundus image can be quickly determined.

According to the present invention, a more accurate neural network model can be obtained by effectively filtering out a low quality fundus image from the fundus image learning data.

According to the present invention, the diagnosis based on the bad image can be prevented, and the accuracy of the diagnosis result can be improved.

According to the present invention, by applying the quality determination process determined in consideration of the target lesion, accurate quality determination results can be obtained even when targeting various lesions.

The effects of the present invention are not limited to the above-described effects, and effects that are not mentioned will be clearly understood by those skilled in the art from the present specification and the accompanying drawings.

1 illustrates a diagnostic assistance system according to an embodiment of the present invention.
2 is a block diagram illustrating a learning apparatus according to an embodiment of the present invention.
3 is a block diagram illustrating in more detail a learning apparatus according to another embodiment of the present invention.
4 is a block diagram illustrating a diagnostic apparatus according to an embodiment of the present invention.
5 is a view for explaining a diagnostic apparatus according to another embodiment of the present invention.
6 illustrates a diagnostic assistance system according to an embodiment of the present invention.
7 is a block diagram illustrating a client device according to an embodiment of the present invention.
8 is a view for explaining a diagnostic assistance process according to an embodiment of the present invention.
9 is a view for explaining the configuration of the learning unit according to an embodiment of the present invention.
10 is a conceptual diagram illustrating an image data set according to an embodiment of the present invention.
11 is a view for explaining image resizing according to an embodiment of the present invention.
12 is a diagram for describing extension of an image data set according to an embodiment of the present invention.
13 is a block diagram illustrating a learning process of a neural network model according to an embodiment of the present invention.
14 is a block diagram illustrating a learning process of a neural network model according to an embodiment of the present invention.
15 is a diagram for describing a method of controlling a learning apparatus, according to an exemplary embodiment.
16 is a diagram for describing a method of controlling a learning apparatus, according to an exemplary embodiment.
17 is a diagram for describing a method of controlling a learning apparatus, according to an exemplary embodiment.
18 is a view for explaining the configuration of the diagnostic unit according to an embodiment of the present invention.
19 is a diagram for describing diagnosis target data according to an embodiment of the present invention.
20 is a diagram for describing a diagnosis process according to an exemplary embodiment.
21 is a diagram illustrating a parallel diagnostic assistance system according to some embodiments of the present invention.
22 is a diagram for describing a parallel diagnosis assistance system according to some embodiments of the present disclosure.
FIG. 23 is a diagram for describing a configuration of a learning apparatus including a plurality of learning units according to an exemplary embodiment.
24 is a diagram for describing a parallel learning process according to an embodiment of the present invention.
25 illustrates a parallel learning process according to another embodiment of the present invention.
26 is a block diagram illustrating a diagnosis unit according to an embodiment of the present invention.
27 is a diagram for describing a diagnostic assistance process according to one embodiment of the present invention.
28 is a diagram for describing a diagnosis assistance system according to an embodiment of the present invention.
29 is a diagram for describing a graphic user interface according to one embodiment of the present invention.
30 is a diagram for describing a graphic user interface according to one embodiment of the present invention.
FIG. 31 is a diagram illustrating region division of a fundus image according to an embodiment of the present invention. FIG.
32 is a diagram illustrating region division of a fundus image according to an embodiment of the present invention.
FIG. 33 is a diagram illustrating region division of a fundus image according to an embodiment of the present invention. FIG.
34 is a view for explaining an apparatus according to an embodiment of the present invention.
35 is a view for explaining an apparatus according to an embodiment of the present invention.
36 illustrates some examples of possible abnormalities in the fundus image.
37 is a flowchart illustrating a method of determining an image quality according to an embodiment of the present invention.
38 is a diagram for describing a graphic user interface according to one embodiment of the present invention.
39 is a flowchart for describing a quality determination of an image according to an exemplary embodiment.
40 is a flowchart illustrating a method of determining an image quality according to an embodiment of the present invention.
41 is a flowchart illustrating a method of determining an image quality according to an embodiment of the present invention.
42 is a flowchart illustrating a method of determining an image quality according to an embodiment of the present invention.
43 is a view for explaining a method of determining image quality according to an embodiment of the present invention.
44 is a diagram for describing a method of determining image quality according to an embodiment of the present invention.
45 is a diagram for describing a method of determining image quality according to an embodiment of the present invention.
46 is a view for explaining a method of determining image quality according to an embodiment of the present invention.
47 is a diagram for describing a plurality of fundus images according to an embodiment of the present invention.
48 is a diagram for describing selecting a target image according to an embodiment of the present invention.
49 is a diagram for describing generation of a target image according to an embodiment of the present invention.
50 is a diagram illustrating a method for diagnosing lesions according to some embodiments of the present invention.
51 is a diagram for describing a method of determining suitability of an image according to an embodiment of the present invention.
52 is a view for explaining a method of determining suitability of an image according to an embodiment of the present invention.
53 is a view for explaining a method of determining suitability of an image according to an embodiment of the present invention.
54 is a diagram for describing a method of determining suitability of an image according to an embodiment of the present invention.
55 is a view to explain a method of determining suitability of a fundus image according to an embodiment of the present invention.
56 is a view to explain a method of determining suitability of a fundus image according to an embodiment of the present invention.
57 is a view to explain a method of determining suitability of a fundus image according to an embodiment of the present invention.
58 is a view for explaining a method of determining the suitability of a fundus image according to an embodiment of the present invention.

The above objects, features and advantages of the present invention will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. However, the present invention may be modified in various ways and may have various embodiments. Hereinafter, specific embodiments will be illustrated in the drawings and described in detail.

In the drawings, the thicknesses of layers and regions are exaggerated for clarity, and the elements or layers are referred to as "on" or "on" of other components or layers. This includes both intervening other layers or other components in the middle as well as directly above other components or layers. Like numbers refer to like elements throughout. In addition, the component with the same function within the range of the same idea shown by the figure of each embodiment is demonstrated using the same reference numeral.

If it is determined that the detailed description of the known function or configuration related to the present invention may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted. In addition, numbers (eg, first, second, etc.) used in the description of the present specification are merely identification symbols for distinguishing one component from another component.

In addition, the suffixes "module" and "unit" for the components used in the following description are given or mixed in consideration of ease of specification, and do not have distinct meanings or roles from each other.

The method according to the embodiment may be embodied in the form of program instructions that can be executed by various computer means and recorded in a computer readable medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the media may be those specially designed and constructed for the purposes of the embodiments, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks. Magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like. The hardware device described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

One. Diagnostic aid using fundus image

1.1 Diagnostic Assistant Systems and Processes

1.1.1 Purpose and definition

Hereinafter, a diagnosis assisting system and method for assisting the determination of the presence or absence of a disease or a disease or an abnormality that is the basis of the determination, etc. will be described based on the fundus image. In particular, a neural network model for diagnosing a disease using a deep learning technique will be described, and a diagnostic assistance system or method for assisting the detection of the presence or absence of a disease using the constructed model will be described.

According to an embodiment of the present invention, a diagnostic assistance system or method for acquiring diagnosis information related to the presence or absence of a disease or finding information used for diagnosis of a disease, and the like, and assisting diagnosis using the fundus image is provided. Can be.

According to one embodiment of the present invention, a diagnostic assistance system or method may be provided to assist in the diagnosis of ocular disease based on the fundus image. For example, a diagnostic assistant system or method may be provided to assist diagnosis by acquiring diagnosis information related to the presence or absence of glaucoma, cataracts, macular degeneration, and prematurity retinopathy of the test subject.

According to another embodiment of the present invention, a diagnostic assistance system or method may be provided to assist in the diagnosis of other diseases (eg, systemic diseases or chronic diseases) that are not eye diseases. For example, a diagnostic assistant system or method may be provided that assists in diagnosis by acquiring diagnosis information of systemic diseases such as hypertension, diabetes, Alzheimer's, cytomegalovirus, stroke, heart disease, atherosclerosis, and the like.

According to another embodiment of the present invention, a diagnostic assistant system or method for detecting abnormal fundus findings that can be used for the diagnosis of eye diseases or other diseases can be provided. For example, color abnormalities across the fundus, lens opacities, cup-to-disc ratios, macular abnormalities (e.g., macular holes), blood vessel diameter, running A diagnostic assistant system or method for acquiring findings such as abnormalities of the back, abnormal diameters of the retinal arteries, bleeding of the retina, occurrence of exudate, drusen, and the like may be provided.

In the present specification, the diagnostic assistance information may be understood to encompass diagnostic information or findings based on the diagnosis according to the presence or absence of a disease.

1.1.2 Diagnostic Assist System Configuration

According to one embodiment of the invention, a diagnostic assistance system may be provided.

1 illustrates a diagnostic assistance system 10 according to an embodiment of the present invention. Referring to FIG. 1, the diagnostic assistance system 10 includes a learning apparatus 1000 for training a diagnostic model, a diagnostic apparatus 2000 for performing a diagnosis using the diagnostic model, and a client device 3000 for obtaining a diagnostic request. It may include. The diagnostic assistance system 10 may include a plurality of learning devices, a plurality of diagnostic devices, or a plurality of client devices.

The learning apparatus 1000 may include a learning unit 100. The learning unit 100 may perform training of the neural network model. For example, the learner 100 may acquire a fundus image data set and perform training of a neural network model for detecting a disease or abnormal finding from the fundus image.

The diagnosis apparatus 2000 may include a diagnosis unit 200. The diagnosis unit 200 may perform an acquisition of supplementary information used for diagnosing or diagnosing a disease using a neural network model. For example, the diagnosis unit 200 may acquire the diagnosis assistance information by using the diagnosis model trained by the learner.

The client device 3000 may include an imaging unit 300. The imaging unit 300 may capture an eye fundus image. The client device may be an ophthalmic fundus imaging device. Alternatively, the client device 3000 may be a handheld device, such as a smartphone or tablet PC.

In the diagnosis assistance system 10 according to the present embodiment, the learning apparatus 1000 determines a neural network model for use in diagnosis assistance by acquiring a data set and learning a neural network model, and the diagnostic apparatus requests information from a client. When the information is obtained, the diagnostic assistance information according to the diagnosis target image may be obtained using the determined neural network model, and the client device may request information from the diagnostic apparatus and acquire the diagnostic assistance information transmitted in response thereto.

The diagnostic assistance system according to another embodiment may include a diagnostic device and a client device for learning the diagnostic model and performing the diagnosis using the same. According to another embodiment, the diagnostic assistance system may include a diagnostic apparatus for learning a diagnostic model, obtaining a diagnostic request, and performing a diagnosis. The diagnostic assistance system according to another embodiment may include a learning device for learning a diagnostic model and a diagnostic device for obtaining a diagnosis request and performing a diagnosis.

The diagnostic assistance system disclosed herein is not limited to the above-described embodiments, but includes a learning unit that performs model learning, a diagnostic unit that acquires diagnostic assistance information according to the learned model, and an imaging unit that acquires a diagnosis target image. It may be implemented in any form, including.

In the following, some embodiments of each device constituting the system will be described.

1.1.2.1 Learning device

The learning apparatus according to an embodiment of the present invention may perform training of a neural network model to assist diagnosis.

2 is a block diagram illustrating a learning apparatus 1000 according to an embodiment of the present invention. 2, the learning apparatus 1000 may include a controller 1200 and a memory unit 1100.

The learning apparatus 1000 may include a controller 1200. The controller 1200 may control an operation of the learning apparatus 1000.

The controller 1200 may include one or more of a central processing unit (CPU), a random access memory (RAM), a graphic processing unit (GPU), one or more microprocessors, and other electronic components capable of processing input data according to predetermined logic. It may include.

The controller 1200 may read a system program stored in the memory unit 1100 and various processing programs. For example, the controller 1200 may develop a data processing process, a diagnostic process, and the like for performing the diagnostic assistant to be described later on the RAM, and perform various processes according to the deployed program. The controller 1200 may learn a neural network model to be described later.

The learning apparatus 1000 may include a memory unit 1100. The memory unit 1100 may store data for learning and a learning model.

The memory unit 1100 may include a nonvolatile semiconductor memory, a hard disk, a flash memory, a RAM, a read only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), or a tangible nonvolatile recording medium. It can be implemented as.

The memory unit 1100 may store various processing programs, parameters for performing processing of the programs, or such processing result data. For example, the memory unit 1100 may include a data processing process program, a diagnostic process program, a parameter for performing each program, and data obtained by performing such a program (for example, processed data or diagnostics) for performing diagnostic assistant, which will be described later. Result value).

The learning apparatus 1000 may include a separate learning unit (or learning module). The learning unit may perform training of the neural network model. The performance of learning is described in more detail in Table 2 below.

The learner may be included in the controller 1200 described above. The learning unit may be stored in the memory unit 1100 described above. The learning unit may be implemented by some components of the controller 1200 and the memory unit 1100 described above. For example, the learning unit may be stored in the memory unit 1100 and driven by the controller 1200.

The learning apparatus 1000 may further include a communication unit 1300. The communication unit 1300 may communicate with an external device. For example, the communication unit 1300 may communicate with a diagnostic device, a server device, or a client device described later. The communication unit 1300 may perform wired or wireless communication. The communication unit 1300 may perform bi-directional or unidirectional communication.

3 is a block diagram illustrating the learning apparatus 1000 according to another embodiment of the present invention in more detail. Referring to FIG. 3, the learning apparatus 1000 may include a processor 1050, a volatile memory 1010, a nonvolatile memory 1030, a mass storage device 1070, and a communication interface 1090.

The processor 1050 of the learning apparatus 1000 may include a data processing module 1051 and a learning module 1053. The processor 1050 may process a data set stored in a mass storage device or a nonvolatile memory through the data processing module 1051. The processor 1050 may perform training of the diagnostic assistance neural network model through the learning module 1053. Processor 1050 may include local memory. The communication interface 1090 may be connected with the network 1110.

However, the learning apparatus 1000 illustrated in FIG. 3 is merely an example, and the configuration of the learning apparatus 1000 according to the present invention is not limited thereto. In particular, the data processing module or learning module may be provided at a location different from that shown in FIG. 3.

1.1.2.2 Diagnostic device

The diagnostic apparatus may obtain diagnostic assistance information using a neural network model.

4 is a block diagram illustrating a diagnostic apparatus 2000 according to an embodiment of the present invention. Referring to FIG. 4, the diagnostic apparatus 2000 may include a controller 2200 and a memory 2100.

The controller 2200 may generate diagnostic assistance information using the diagnostic assistance neural network model. The controller 2200 may acquire diagnostic data for diagnosis (eg, fundus data of a subject) and acquire diagnostic assistance information predicted by the diagnostic data using the trained diagnostic assistance neural network model.

The memory unit 2100 may store the learned diagnostic assistance neural network model. The memory unit 2100 may store parameters, variables, and the like of the diagnostic assistance neural network model.

The diagnostic apparatus 2000 may further include a communication unit 2300. The communicator 2300 may communicate with a learning device and / or a client device. For example, the diagnosis apparatus 2000 may be provided in the form of a server communicating with the client device. In this regard, it will be described in more detail below.

5 is a diagram for describing a diagnostic apparatus 2000 according to another exemplary embodiment. Referring to FIG. 5, the diagnostic apparatus 2000 according to an embodiment of the present invention may include a processor 2050, a volatile memory 2030, a nonvolatile memory 2010, a mass storage device 2070, and a communication interface 2090. It may include.

The processor 2050 of the diagnostic apparatus may include a data processing module 2051 and a diagnostic module 2053. The processor 2050 may process the diagnostic data through the data processing module 2051, and may obtain diagnostic assistance information according to the diagnostic data through the diagnostic module 2053.

1.1.2.3 Server devices

According to an embodiment of the present invention, the diagnostic assistance system may include a server device. The diagnostic assistance system according to an embodiment of the present invention may include a plurality of server devices.

The server device may store and / or drive a neural network model. The server device may store weight values constituting the trained neural network model. The server device may collect or store data used for diagnostic assistance.

The server device may output the result of the diagnostic assistance process using the neural network model to the client device. The server device may obtain feedback from the client device. The server device may operate similarly to the diagnostic device described above.

6 illustrates a diagnostic assistance system 20 according to an embodiment of the present invention. Referring to FIG. 6, the diagnostic assistance system 20 according to an embodiment of the present invention may include a diagnostic server 4000, a learning device, and a client device.

The diagnostic server 4000, that is, the server device may communicate with a plurality of learning devices or a plurality of diagnostic devices. Referring to FIG. 6, the diagnostic server 4000 may communicate with the first learning apparatus 1000a and the second learning apparatus 1000b. Referring to FIG. 6, the diagnostic server 4000 may communicate with the first client device 3000a and the second client device 3000b.

For example, the diagnostic server 4000 trains the first learning apparatus 1000a for learning the first diagnostic assistance neural network model for obtaining the first diagnostic assistance information and the second diagnostic assistance neural network model for obtaining the second diagnostic assistance information. The second learning apparatus 1000b may communicate.

The diagnostic server 4000 stores the first diagnostic auxiliary neural network model for obtaining the first diagnostic assistance information and the second diagnostic auxiliary neural network model for obtaining the second diagnostic assistance information, and the first client device 3000a or the second client device. The diagnostic assistance information may be obtained in response to the request for obtaining the diagnostic assistance information from 3000b, and the obtained diagnostic assistance information may be transmitted to the first client device 3000a or the second client device 3000b.

Alternatively, the diagnostic server 4000 may communicate with the first client device 3000a requesting the first diagnostic assistance information and the second client device 3000b requesting the second diagnostic assistance information.

1.1.2.4 Client device

The client device may request diagnostic assistance information from the diagnostic device or the server device. The client device may acquire data necessary for diagnosis and transmit the acquired data to the diagnostic device.

The client device may include a data acquirer. The data acquirer may acquire data necessary for assistance of diagnosis. The data acquisition unit may be an imaging unit that acquires an image used for the diagnostic assistance model.

7 is a block diagram illustrating a client device 3000 according to an embodiment of the present invention. Referring to FIG. 7, the client device 3000 according to an exemplary embodiment may include an imaging unit 3100, a controller 3200, and a communication unit 3300.

The imaging unit 3100 may acquire image or image data. The imaging unit 3100 may acquire a fundus image. However, the client device 3000 may be replaced with a data acquisition unit other than the imaging unit 3100.

The communicator 3300 may communicate with an external device, such as a diagnostic device or a server device. The communicator 3300 may perform wired or wireless communication.

The controller 3200 may control the imaging unit 3100 to acquire an image or data. The controller 3200 may control the image capturer 3100 to acquire a fundus image. The controller 3200 may transmit the acquired fundus image to the diagnosis apparatus. The controller may transmit the image acquired through the imaging unit 3100 to the server device through the communication unit 3300, and acquire the diagnostic assistance information generated based on the image.

Although not shown, the client device may further include an output unit. The output unit may include a display that outputs an image or an image or a speaker that outputs an audio. The output unit may output the image or the image data obtained by the obtained imaging unit. The output unit may output diagnostic assistance information obtained from the diagnostic apparatus.

Although not shown, the client device may further include an input unit. The input unit may obtain a user input. For example, the input unit may obtain a user input for requesting the diagnostic assistance information. The input unit may obtain user information for evaluating the diagnostic assistance information obtained from the diagnostic apparatus.

In addition, although not shown, the client device may further include a memory unit. The memory unit may store an image acquired by the imaging unit.

1.1.3 Diagnostic Assistant Process Overview

The diagnostic assistant process may be performed by the diagnostic assistant system or the diagnostic assistant apparatus disclosed herein. The diagnostic assistant process can be considered as divided into a training process for learning a diagnostic assistant model used for diagnosis assistance and a diagnostic process using the diagnostic assistant model.

8 is a view for explaining a diagnostic assistance process according to an embodiment of the present invention. Referring to FIG. 8, the diagnostic assistance process according to an embodiment of the present disclosure acquires and processes data (S110), learns a neural network model (S130), and obtains variables of the learned neural network model (S150). It may include a diagnostic assistant process to acquire the learning process and the diagnostic target data (S210) and to obtain diagnostic assistance information (S250) using the trained neural network model (S230) based on the diagnostic target data.

More specifically, the training process may include a data processing process of processing the input training image data and processing the model into a state that can be used for training of the model, and a training process of training the model using the processed data. The training process may be performed by the learning apparatus described above.

The diagnostic process may include a data processing process of processing the input test target image data to a state capable of performing a diagnosis using a neural network model, and a diagnostic process of performing diagnosis using the processed data. The diagnostic process may be performed by the above-described diagnostic apparatus or server apparatus.

Below, each process is demonstrated.

1.2 Training process

According to one embodiment of the invention, a process for training a neural network model may be provided. As a specific example, a process of training a neural network model to perform or assist with a diagnosis based on the fundus image can be disclosed.

The training process described below may be performed by the learning apparatus described above.

1.2.1 Learning Department

According to an embodiment of the present invention, the training process may be performed by the learning unit. The learning unit may be provided in the above-described learning apparatus.

9 is a view for explaining the configuration of the learning unit 100 according to an embodiment of the present invention. Referring to FIG. 9, the learner 100 may include a data processing module 110, a queue module 130, a learning module 150, and a learning result obtaining module 170. Each module may perform separate steps of the data processing process and the learning process as described below. However, not all of the components described in FIG. 9 and the functions performed by each element are essential, and some elements may be added or some elements may be omitted depending on a learning form.

1.2.2 Data processing process

1.2.2.1 Image data acquisition

According to one embodiment of the invention, a data set may be obtained. According to one embodiment of the invention, the data processing module may acquire a data set.

The data set may be an image data set. Specifically, it may be a fundus image data set. The fundus image data set may be acquired using a general non-dongdong fundus camera or the like. The fundus image may be a panoramic image. The fundus image may be a red-free image. The fundus image may be an infrared photographed image. The fundus image may be an autofluorescence image. The image data may be obtained in any one of JPG, PNG, DCM (DICOM), BMP, GIF, and TIFF.

The data set may comprise a training data set. The data set may comprise a test data set. The data set may comprise a validation data set. In other words, the data set may be allocated to at least one data set of the training data set, the test data set, and the verification data set.

The data set may be determined in consideration of diagnostic assistance information to be obtained using a neural network model trained through the data set. For example, in order to train a neural network model for obtaining diagnostic assistance information related to cataracts, the data set obtained may be determined as an infrared fundus image data set. Alternatively, when the neural network model for acquiring diagnostic assistance information related to macular degeneration is to be trained, the acquired data set may be a self-fluoresced fundus image data set.

Individual data included in the data set may include a label. The label may be plural. In other words, individual data included in the data set may be labeled for at least one feature. As an example, the data set is a fundus image data set including a plurality of fundus image data, each fundus image data being a diagnostic information label (eg, the presence or absence of a specific disease) and / or finding information (eg, specific) according to the image. Abnormalities of the site) may include a label.

As another example, the data set is a fundus image data set and each fundus image data may include a perimeter information label for that image. For example, each fundus image data may include left and right eye information on whether the corresponding fundus image is an image of the left eye or right eye, gender information on whether the fundus image is a female or a male fundus image, and a subject who photographed the fundus image. Peripheral information labels, including age information about the age of may include.

10 is a conceptual diagram illustrating an image data set DS according to an embodiment of the present invention. Referring to FIG. 10, an image data set DS according to an embodiment of the present invention may include a plurality of image data IDs. Each image data ID may include an image I and a label L assigned to the image. Referring to FIG. 10, the image data set DS may include first image data ID1 and second image data ID2. The first image data ID1 may include a first image I1 and a first label L1 corresponding to the first image.

In FIG. 10, one image data includes one label. However, as described above, one image data may include a plurality of labels.

1.2.2.2 Image resizing

According to an embodiment of the present invention, the size of the acquired image data may be adjusted. That is, the images can be resized. According to an embodiment of the present invention, image resizing may be performed by the data processing module described above.

The size or aspect ratio of the image can be adjusted. The obtained plurality of images may be sized to have a constant size. Alternatively, the images can be scaled to have a constant aspect ratio. Resizing an image may be applying an image conversion filter to the image.

If the size or capacity of the individual images obtained is excessively large or small, the size or capacity of the image may be adjusted to convert to an appropriate size. Alternatively, if the size or capacity of the individual images vary, resizing may unify the size or capacity.

According to one embodiment, the capacity of the image may be adjusted. For example, when the capacity of an image exceeds an appropriate range, downsampling can reduce the image. Alternatively, when the capacity of the image falls short of an appropriate range, the image may be enlarged through upsampling or interpolating.

According to another exemplary embodiment, the size or aspect ratio of the image may be adjusted by cropping the image or adding pixels to the acquired image. For example, if an image contains an unnecessary part for learning, a part of the image may be cropped to remove it. Alternatively, if a portion of the image is cut off and the aspect ratio is not correct, a column or row may be added to adjust the image aspect ratio. In other words, the aspect ratio can be adjusted by adding margins or padding to the image.

According to another embodiment, the capacity and size or aspect ratio of the image may be adjusted together. For example, when the size of an image is large, the image size may be reduced by downsampling the image, and unnecessary portions included in the reduced image may be cropped and converted into appropriate image data.

In addition, according to another embodiment of the present invention, the orientation of the image data may be changed.

As a specific example, when the fundus image data set is used as the data set, each fundus image may be adjusted in size or resized. Cropping to remove the marginal portions other than the fundus portion of the fundus image or padding to adjust the aspect ratio by replenishing the cropped portion of the fundus image may be performed.

11 is a view for explaining image resizing according to an embodiment of the present invention. Referring to FIG. 11, an acquired fundus image may be resized by an image resizing process according to an embodiment of the present invention.

Specifically, the original fundus image (a) may be cropped (b) the margins unnecessary for obtaining diagnostic information (b) and may be reduced in size (c) to improve learning efficiency.

1.2.2.3 Image preprocessing

According to one embodiment of the invention, preprocessing of the image may be performed. If the image is used as it is input for learning, overfitting may occur as a result of learning about unnecessary characteristics, and the learning efficiency may also be reduced.

To prevent this, it is possible to improve learning efficiency and performance by appropriately pre-processing and using the image data to meet the purpose of learning. For example, pretreatment may be performed on the fundus image to facilitate highlighting abnormalities of the ocular disease or pretreatment to highlight retinal vessel to blood flow changes.

The preprocessing of the image may be performed by the data processing module of the learning unit described above. The data processing module may acquire the resized image and perform preprocessing required for learning.

The preprocessing of the image may be performed on the image in which the above-mentioned resizing process is completed. However, the contents of the present invention disclosed herein are not limited thereto, and the resizing process may be omitted and preprocessing may be performed on the image. Applying the preprocessing of the image may be to apply a preprocessing filter to the image.

According to an embodiment, a blur filter may be applied to the image. Gaussian filters may be applied to the image. Gaussian blur filters may be applied to the image. Alternatively, a deblur filter for sharpening the image may be applied to the image.

According to another embodiment, a filter for adjusting or modulating the color of the image may be applied. For example, a filter may be applied to change the value of some of the RGB values constituting the image, or to binarize the image.

According to another embodiment, a filter may be applied to emphasize specific elements in an image. For example, for fundus image data, preprocessing may be performed to highlight the vascular element from each image. At this time, the pretreatment for emphasizing the vascular element may be to apply one or more filters sequentially or in combination.

According to an embodiment of the present invention, the preprocessing of the image may be performed in consideration of the characteristics of the diagnostic assistance information to be obtained. For example, when it is desired to acquire diagnostic assistance information related to findings such as bleeding, drusen, microvascular flow, and exudate of the retina, pretreatment may be performed to convert the acquired fundus image into a red-free fundus image.

1.2.2.4 Image augmentation

According to one embodiment of the invention, the image may be augmented or expanded. Augmentation of the image may be performed by the data processing module of the learning unit described above.

Augmented images can be used to improve training performance of the neural network model. For example, when the amount of data for training of the neural network model is insufficient, by modulating existing training image data to expand the number of data for training, and using the modulated (or changed) image together with the original image You can increase the number of training image data. As a result, overfitting can be suppressed, a layer of a model can be formed deeper, and prediction accuracy can be improved.

For example, the expansion of the image data may be performed by inverting the left and right sides of the image, clipping out a part of the image, correcting a color value of the image, or adding artificial noise. As a specific example, clipping out a portion of an image may be performed by clipping some regions of an element constituting the image, or randomly clipping some regions. For example, the image data can be extended by inverting left and right, upside down, rotating, scaling ratio resizing, cropping, padding, color adjustment, or brightness adjustment.

As an example, the augmentation or extension of the image data described above may generally be applied to a training data set. However, the present invention may be applied to other data sets, for example, test data sets, that is, data sets for testing the verified model using training and validation data using training data.

As a specific example, when the fundus image data set is used as the data set, one or more processes of inverting, cropping, adding noise, or changing color in order to increase the number of data are randomly applied. An augmented fundus image data set may be obtained.

12 is a diagram for describing extension of an image data set according to an embodiment of the present invention. Referring to FIG. 12, an image according to embodiments of the present disclosure may be modified to improve prediction accuracy of a neural network model.

Specifically, referring to FIG. 12, an image according to one embodiment of the present invention may include a portion of which is dropped out (a), inverted from side to side (b), or a center point is moved (c, d) of the region. Color may be modulated (e).

1.2.2.5 Image serialization

According to one embodiment of the invention, the image data may be linearized. The image may be serialized by the data processing module described above. The serialization module can serialize the preprocessed image data and pass it to the queue module.

If the image data is used as it is for training, decoding is necessary because the image data has an image file type such as JPG, PNB, or DCM. If the training is performed after decoding each time, the performance of model training may be degraded. . Accordingly, the learning can be performed by serializing the image file without using it for learning. Therefore, serialization of image data can be performed to improve learning performance and speed. The image data to be serialized may be image data to which at least one of the above-described image resizing and image preprocessing is applied, or both of the image data may not be processed.

Each image data included in the image data set may be converted into a string form. Image data may be converted into binary data form. In particular, the image data may be transformed into a data form suitable for use in training the neural network model. For example, the image data may be converted into a TFRecord form for use in neural network model training using tensorflow.

As a specific example, when the fundus image set is used as the data set, the acquired fundus image set may be converted into a TFRecord form and used for learning a neural network model.

1.2.2.6 Queue

Queues can be used to eliminate data bottlenecks. The queue module of the learning unit may store image data in a queue and transmit the image data to the learning model module.

In particular, when using the CPU (Central Processing Unit) and the GPU (Graphic Processing Unit) together for the learning process, the use of queues minimizes the bottleneck between the CPU and the GPU, facilitates access to the database, and memory. It can improve the use efficiency.

The queue may store data used to train the neural network model. The cue can store image data. The image data stored in the queue may be image data processed or at least one of the above-described data processing processes (ie, resizing, preprocessing, and augmentation) is processed image data.

The queue may store image data, preferably serialized image data as described above. The queue may store image data and supply image data to the neural network model. The queue may deliver image data in a batch size unit to a neural network model.

The queue may provide image data. The queue may provide data to a learning module described later. As data is extracted from the learning module, the number of data accumulated in the queue may be reduced.

If the number of data stored in the queue decreases below the reference level as the neural network model is trained, the queue may request replenishment of the data. The queue may request replenishment of certain kinds of data. When the learning unit is requested to supplement the data, the queue may supplement the data in the queue.

The queue may be provided in the system memory of the learning device. For example, the queue may be formed in a random access memory (RAM) of a central processing unit (CPU). In this case, the size of the queue, that is, the capacity may be determined according to the RAM capacity of the CPU. The queue may be a first in first out (FIFO) queue, a primary queue, or a random queue.

1.2.3 Learning process

According to one embodiment of the invention, the learning process of the neural network model can be initiated.

According to an embodiment of the present invention, the training of the neural network model may be performed by the above-described learning apparatus. The learning process may be performed by the controller of the learning apparatus. The learning process may be performed by the learning module described above.

13 is a block diagram illustrating a learning process of a neural network model according to an embodiment of the present invention. Referring to FIG. 13, the training process of the neural network model according to an embodiment of the present invention obtains data (S1010), learns a neural network model (S1030), verifies a trained model (S1050), and learns a model. This may be performed by acquiring a variable of S1070.

Hereinafter, referring to FIG. 13, some examples of the learning process of the neural network model will be described.

1.2.3.1 Data entry

A data set for training the diagnostic assistance neural network model can be obtained.

The data obtained may be an image data set processed by the data processing process described above. As an example, the data set may include serialized fundus image data after it has been sized, preprocessed filters applied, and data is augmented.

In the training phase of the neural network model, a training data set may be obtained and used. In the verification phase of the neural network model, a verification data set may be obtained and used. In the testing phase of the neural network model, a test data set may be obtained and used. Each data set may include a fundus image and a label.

The data set can be obtained from the queue. The data set may be obtained in batch size units from the queue. For example, if 60 are specified as the batch size, the data set may be extracted from the queue in 60 units. The size of the batch size may be limited by the RAM capacity of the GPU.

The data set can be randomly obtained from the queue into the learning module. The data set may be obtained in the order accumulated in the queue.

The learning module may specify and extract the configuration of the data set obtained from the queue. For example, the learning module may extract the fundus image having the left eye label and the fundus image data having the right eye label of a specific subject to be used together for learning.

The learning module may obtain a data set of a specific label from the queue. For example, the learning module may obtain a fundus image data set having an abnormal diagnostic information label from a queue. The learning module may obtain a data set by specifying a ratio of the number of data according to the label from the queue. For example, the learning module may acquire a fundus image data set from the queue such that the number of fundus image data having an abnormal diagnostic information label and the number of fundus image data having a normal diagnostic information label are one to one.

1.2.3.2 Model design

The neural network model may be a diagnostic assistance model that outputs diagnostic assistance information based on the image data. The structure of the diagnostic assistance neural network model for obtaining the diagnostic assistance information may have a predetermined form. The neural network model may include a plurality of layers or layers.

The neural network model may be implemented in the form of a classifier that generates diagnostic assistance information. The classifier may perform double classification or multiple classification. For example, the neural network model may be a binary classification model that classifies input data into a normal or abnormal class for target diagnosis assistance information such as a specific disease or abnormality. Alternatively, the neural network model may be a multiple classification model that classifies input data into a plurality of class classes for particular characteristics (eg, degree of disease progression). Alternatively, the neural network model may be implemented as a regression type model that outputs a specific value related to a specific disease.

The neural network model may include a convolutional neural network (CNN). Can be. As the CNN structure, at least one of AlexNet, LENET, NIN, VGGNet, ResNet, WideResnet, GoogleNet, FractaNet, DenseNet, FitNet, RitResNet, HighwayNet, MobileNet, and DeeplySupervisedNet may be used. The neural network model may be implemented using a plurality of CNN structures.

As an example, the neural network model may be implemented to include a plurality of VGGNet blocks. More specifically, the neural network model includes a CNN layer having 64 filters of 3x3 size, a batch normalization (BN) layer and a ReLu layer, and a CNN layer having a 128x filter of 3x3 size, a ReLu layer, The second block, in which the BN layers are sequentially combined, may be provided in combination.

The neural network model includes each CNN block, followed by a max pooling layer, and at the end may include a global average pooling layer, a fully connected layer, and an activation layer (eg, sigmoid, softmax, etc.). have.

1.2.3.3 Model learning

Neural network models can be trained using training data sets.

Neural network models can be trained using labeled data sets. However, the learning process of the diagnostic assisted neural network model described herein is not limited thereto, and the neural network model may be trained in an unsupervised form using unlabeled data.

The training of the neural network model, based on the training image data, obtains a result using a neural network model to which arbitrary weight values are assigned, compares the obtained result with a label value of the training data and back propagates according to the error. Can be performed by optimizing the weight values. In addition, the training of the neural network model may be influenced from feedback from the verification results, test results and / or diagnostic steps of the models described below.

Training of the neural network model described above may be performed using TensorFlow. However, the present invention is not limited thereto, and frameworks such as Theano, Keras, Caffe, Torch, and CNTK (Microsoft Cognitive Toolkit) are used for learning neural network models. May be

1.2.3.4 Validation of the model

The neural network model can be verified using the validation data set. The verification of the neural network model may be performed by obtaining a result value for the verification data set from the neural network model in which the learning is performed, and comparing the result value with a label of the verification data set. Verification can be performed by measuring the accuracy of the results. Depending on the verification result, parameters of the neural network model (eg, weights and / or biases) or hyper parameters (eg, learning rates) may be adjusted.

For example, the learning apparatus according to an embodiment of the present disclosure may learn a neural network model predicting diagnosis assistance information based on the fundus image, and correspond to the diagnosis fundus image with the diagnosis assistance information of the verified fundus image of the trained model. Validation of the diagnostic assistance neural network model may be performed by comparison with the validation label.

For verification of the neural network model, a separate verification set, i.e., a data set having distinct factors not included in the training data set, may be used. For example, the separate verification set may be a data set in which the training data set and the factors such as race, environment, age, and gender are distinguished.

1.2.3.5 Model testing

Neural network models can be tested using test data sets.

Although not shown in FIG. 13, according to an embodiment of the present disclosure, the neural network model may be tested using a test data set that is distinguished from a training data set and a verification data set. Depending on the test results, the parameters (eg, weights and / or biases) or hyperparameters (eg, learning rates) of the neural network model may be adjusted.

For example, the learning apparatus according to an embodiment of the present invention obtains a result value of inputting test fundus image data that is not used for training and verification, from a neural network model trained to predict diagnostic assistance information based on the fundus image. Thus, a test of the trained and verified diagnostic auxiliary neural network model may be performed.

For the testing of the neural network model, a separately prepared external data set, i.e., a data set having factors distinct from training and / or validation data, may be used.

1.2.3.6 Output of the result

As a result of training the neural network model, parameter values of the optimized model may be obtained. As described above, as the model is repeatedly trained using the test data set, more appropriate parameter (or variable) values may be obtained. If the learning proceeds sufficiently, an optimized value of weight and / or bias can be obtained.

According to an embodiment of the present invention, the learned neural network model and / or parameters or variables of the learned neural network model may be stored in the learning apparatus and / or the diagnostic apparatus (or the server). The trained neural network model may be used for prediction of diagnostic assistance information by a diagnostic device and / or a client device. In addition, the parameters or variables of the learned neural network model may be updated by feedback obtained from the diagnostic apparatus or the client apparatus.

1.2.3.7 Model ensemble

According to an embodiment of the present invention, a plurality of sub-models may be learned at the same time in the course of learning one diagnostic auxiliary neural network model. The plurality of submodels may have different hierarchical structures.

In this case, the diagnostic assisted neural network model according to an embodiment of the present invention may be implemented by combining a plurality of sub-neural network models. In other words, the neural network model may be trained using an ensemble technique that combines a plurality of sub-neural networks.

When a diagnostic auxiliary neural network model is formed by forming an ensemble, prediction may be performed by combining the results predicted from various types of sub-neural network models, thereby improving accuracy of the result prediction.

14 is a block diagram illustrating a learning process of a neural network model according to an embodiment of the present invention. Referring to FIG. 14, in the training process of the neural network model according to an embodiment of the present invention, a data set is obtained (S1011), and a first model (ie, a first neural network model) and a second are obtained using the acquired data. The model (ie, the second neural network model) is trained (S1031, S1033), the trained first neural network model and the second neural network model are verified (S1051), the final neural network model is determined, and its parameters or variables are obtained (S10721). )can do.

Hereinafter, with reference to FIG. 14, some embodiments of the neural network model learning process will be described.

According to an embodiment of the present invention, the plurality of sub-neural network models may acquire the same training data set and generate output values separately. In this case, an ensemble of the plurality of sub-neural network models may be determined as a final neural network model, and parameter values for each of the plurality of sub-neural network models may be obtained as a learning result. The output value of the final neural network model may be determined as an average value of the output values by each sub-neural network model. Alternatively, the output value of the final neural network model may be determined as a weighted average value for the output value of each sub-neural network model in consideration of the accuracy obtained as a result of verification for each sub-neural network model.

More specifically, when the neural network model includes the first sub-neural network model and the second sub-neural network model, the optimized parameter value for the first sub-neural network model and the optimized parameter value of the second sub-neural network model by machine learning Can be obtained. In this case, an average value of output values (for example, probability values for specific diagnostic assistance information) obtained from the first sub-neural network model and the second sub-neural network model may be determined as the output value of the final neural network model.

According to another embodiment of the present invention, the accuracy of the individual sub-neural network models may be evaluated based on output values of each of the plurality of sub-neural network models. At this time, one of the plurality of sub-neural network models may be selected and determined as the final sub-neural network model based on the accuracy. The structure of the determined sub neural network model and the parameter values of the determined sub neural network model obtained as a result of the training may be stored.

As a more specific example, when the neural network model includes a first sub-neural network model and a second sub-neural network model, accuracy is obtained according to each of the first and second sub-neural network models, and a more accurate sub-neural network model is obtained. This can be determined by a neural network model.

According to another embodiment of the present invention, combining at least one sub-neural network among a plurality of neural network models, forming an ensemble of the combined at least one sub-neural network model, evaluating each ensemble, The combination of sub-neural network models forming a highly accurate ensemble can be determined as the final neural network model. In this case, the ensemble may be performed for all possible cases of selecting at least one of the plurality of sub-neural network models, and the sub-neural network combination evaluated as the highest accuracy may be determined as the final neural network model.

As a more specific example, when the neural network model includes a first sub-neural network model and a second sub-neural network model, the accuracy of the first sub-neural network model, the accuracy of the second sub-neural network model and the ensemble of the first and second sub-neural network models By comparing the accuracy by and the sub-neural network model configuration of the most accurate case can be determined as the final neural network model.

1.2.4 Example 1-Method of Control of Learning Device.

15 is a diagram for describing a method of controlling a learning apparatus, according to an exemplary embodiment.

Referring to FIG. 15, in the method of controlling a learning apparatus according to an embodiment of the present disclosure, performing preprocessing of the first fundus image (S110), serializing the preprocessed first fundus image (S130), and first Training of the neural network model (S150) may be included.

A control method of a learning apparatus according to an embodiment of the present invention obtains a first training data set including a plurality of fundus images, processes the fundus image included in the first training data set, and the first training. A training apparatus for training a first neural network model, a target fundus image for acquiring diagnosis assistance information is obtained using a data set, and the diagnosis assistance is based on the target fundus image using the trained first neural network model. The method may be a control method of a learning apparatus included in a system including a diagnostic apparatus for obtaining information.

Performing preprocessing of the first fundus image (S110) may perform preprocessing of the first fundus image such that the first fundus image included in the first training data set is converted into a form suitable for learning of the first neural network model. It may further include doing.

The control method of the learning apparatus according to an embodiment of the present invention may include serializing the preprocessed first fundus image (S130). The first fundus image may be serialized in a form that is easy for learning of the neural network model.

In this case, the training of the first neural network model (S150) further includes training the first neural network model classifying the target fundus image into a first label or a second label by using the serialized first fundus image. can do.

The learning apparatus may obtain a second training data set including a plurality of fundus images and at least partially different from the first training data set, and train the second neural network model using the second training data set.

According to an embodiment of the present invention, a method of controlling a learning apparatus may include performing preprocessing of a second fundus image such that a second fundus image included in a second training set is suitable for learning a second neural network model. Serializing the two fundus images and using the serialized second fundus image, training a second neural network model that classifies the subject fundus image into a third label or a fourth label.

16 is a diagram for describing a method of controlling a learning apparatus, according to an exemplary embodiment. Referring to FIG. 16, in the method of controlling a learning apparatus according to an embodiment of the present disclosure, performing a preprocessing of a second fundus image (S210), serializing a preprocessed second fundus image (S230), and a second method Training of the neural network model (S250) may be included.

In FIG. 16, for convenience of description, preprocessing of the first fundus image, serialization of the first fundus image, and learning using the first fundus image, followed by preprocessing of the first fundus image, serialization of the second fundus image, and first fundus image Although described as being performed following learning with, the content of the present invention is not limited thereto.

Preprocessing of the second fundus image included in the second training set, and serializing the second fundus image and learning using the second fundus image, include the above-described preprocessing of the first fundus image, serialization of the first fundus image, and first fundus. It can be performed independently of learning using images. Preprocessing of the second fundus image included in the second training set, and serializing the second fundus image and learning using the second fundus image, include the above-described preprocessing of the first fundus image, serialization of the first fundus image, and first fundus. It can be performed in parallel with learning using images. In other words, the preprocessing of the second fundus image included in the second training set, and the serialization of the second fundus image and the learning using the second fundus image must be the preprocessing of the first fundus image described above and the serialization of the first fundus image. And after the learning using the first fundus image, or not previously performed, the processing on the first fundus image and the processing on the second fundus image may be performed without mutual dependence.

The first preprocessing performed on the fundus image included in the first training data set may be distinguished from the second preprocessing performed on the fundus image included in the second training data set. For example, the first pretreatment may be vascular emphasis pretreatment, and the second pretreatment may be color modulation pretreatment. Each preprocessing may be determined in consideration of diagnostic assistance information to be obtained through each neural network model.

According to an embodiment of the present invention, a method of controlling a learning apparatus may evaluate an accuracy of the trained first neural network model by using a first verification data set that is at least partially distinguished from a first training data set. Validating the first neural network model and evaluating the accuracy of the trained first neural network model using a second verification data set that is at least partially distinct from a second training data set, thereby evaluating the second neural network model It may further comprise the step of verifying. At this time, the verification of the first neural network model and the second neural network model may be performed independently.

The serialized first fundus image is sequentially stored in a first queue, and the serialized fundus image stored in the first queue is used for learning the first neural network model in a predetermined capacity unit from the first queue, and serialized. A second fundus image is sequentially stored in a second cue that is distinct from the first cue, and the serialized fundus image stored in the second cue is subjected to learning of the first neural network model in a predetermined capacity unit from the second cue. Can be used.

The first neural network model may include a first sub neural network model and a second sub neural network model. In this case, classifying the target fundus image into the first label or the second label may be performed by considering a first prediction value predicted by the first sub-neural network model and a second prediction value predicted by the second sub-neural network model. Can be performed.

The second neural network model may include a third sub neural network model and a fourth sub neural network model. In this case, classifying an object fundus image into the third label or the fourth label may be performed by considering a third prediction value predicted by the third sub-neural network model and a fourth prediction value predicted by the fourth sub-neural network model. Can be performed.

The first training data set may include at least a portion of the fundus image labeled with the first label, and the second training data set may include at least a portion of the fundus image labeled with the third label. In this case, the fundus image labeled with the first label and the at least some fundus image labeled with the third label may be partially common.

The first label is a normal label indicating that the subject corresponding to the subject fundus image is normal with respect to the first finding, and the second label is an abnormal label indicating that the subject is abnormal with respect to the second finding. Can be.

Performing preprocessing of the first fundus image may include cropping the first fundus image to satisfy a reference aspect ratio and changing the size of the first fundus image.

The pre-processing of the first fundus image may further include applying a blood vessel emphasis filter to the fundus image such that the processing unit highlights the blood vessels included in the first fundus image.

The serialized first fundus image may be sequentially stored in a queue, and the serialized first fundus image stored in the queue may be used for learning the first neural network model in a predetermined capacity unit from the queue. The cue may request replenishment of the serialized first fundus image if the serialized first fundus image not used for the first learning is reduced below a reference dose.

The first finding may be any one of retinal bleeding findings, retinal exudation findings, lens clouding findings, and diabetic retinopathy findings.

17 is a diagram for describing a method of controlling a learning apparatus, according to an exemplary embodiment.

Referring to FIG. 17, the control method of the learning apparatus according to an embodiment of the present invention may further include verifying the first neural network model (S170) and updating the first neural network model (S190).

The step S170 of validating the first neural network model may be performed by evaluating the accuracy of the trained first neural network model by using the first verification data set that is at least partially distinguished from the first training data set. The method may further include verifying.

The method may further include updating the first neural network model by reflecting the verification result obtained from updating the first neural network model (S190) and validating the first neural network model (S170).

Meanwhile, the first neural network model may include a first sub neural network model and a second sub neural network model. In this case, the training of the first neural network model may include verifying the first sub-neural network model using a first verification data set to obtain an accuracy of the first sub-neural network model, and using the first verification data set. The second sub-neural network model is verified to obtain an accuracy of the second sub-neural network model, and the accuracy of the first sub-neural network model and the accuracy of the second sub-neural network model are compared to make a more accurate sub-neural network model as the final neural network model. May include determining.

1.3 Diagnostic Assistant Process

According to an embodiment of the present invention, a diagnostic assistance process (or diagnostic process) for acquiring diagnostic assistance information using a neural network model may be provided. As a specific example, by the diagnostic assistant process, diagnostic assistant information (eg, diagnostic information or findings information) may be predicted using a fundus image and trained diagnostic assistant neural network model.

The diagnostic assistance process described below may be performed by a diagnostic apparatus.

1.3.1 Diagnostic department

According to an embodiment of the present invention, the diagnostic process may be performed by the diagnosis unit 200. The diagnosis unit 200 may be provided in the above-described diagnosis apparatus.

18 is a view for explaining the configuration of the diagnostic unit 200 according to an embodiment of the present invention. Referring to FIG. 18, the diagnosis unit 200 may include a diagnosis request obtaining module 210, a data processing module 230, a diagnosis module 250, and an output module 270.

Each module may perform separate steps of the data processing process and the learning process as described below. However, not all of the components described in FIG. 16 and the functions performed by each of the elements are essential, and some elements may be added or some elements may be omitted depending on the aspect of diagnosis.

1.3.2 Data Acquisition and Diagnostic Requests

The diagnostic apparatus according to an embodiment of the present invention may obtain diagnostic target data and obtain diagnostic assistance information based on the diagnostic target data. The diagnosis target data may be image data. The data acquisition and the acquisition of the diagnosis request may be performed by the above-described diagnosis request acquisition module of the diagnosis unit.

19 is a diagram for describing diagnosis target data TD according to an exemplary embodiment of the present invention. Referring to FIG. 19, the diagnosis target data TD may include a diagnosis target image TI and diagnosis object information PI.

The diagnosis target image TI may be an image for obtaining diagnosis assistance information on the diagnosis target object. For example, the diagnosis target image may be a fundus image. The diagnostic target (TI) JPG, PNG, DCM (DICOM), BMP, GIF, can be any one of the formats of TIFF.

The diagnostic object information PI may be information for identifying a diagnosis object. Alternatively, the diagnostic object information PI may be characteristic information of a diagnostic object or an image. For example, the diagnostic object information PI may include information such as a date and time of imaging the diagnosis target image, photographing equipment, an identification number of the subject to be diagnosed, an ID, a name, a gender, an age, or a weight. When the diagnosis target image is a fundus image, the diagnostic object information PI may further include eye related information such as binocular information indicating whether the left eye is the right eye.

The diagnostic device may obtain a diagnostic request. The diagnostic apparatus may acquire the diagnosis target data together with the diagnosis request. When the diagnostic request is obtained, the diagnostic apparatus may acquire diagnostic assistance information using the trained diagnostic assistance neural network model. The diagnostic device may obtain a diagnostic request from the client device. Alternatively, the diagnostic apparatus may obtain a diagnostic request from a user through an input means provided separately.

1.3.3 Data processing process

The obtained data can be processed. Processing of the data may be performed by the data processing module of the above-described diagnostic unit.

The data processing process may generally be performed similarly to the data processing process in the aforementioned learning process. Hereinafter, the data processing process in a diagnostic process is demonstrated centering on the difference with the data processing process in a learning process.

In the diagnostic process, the diagnostic device may acquire data as in the learning process. In this case, the obtained data may be in the same format as the data obtained in the learning process. For example, in the learning process, when the learning apparatus trains the diagnostic assistance neural network model using the image data in DCM format, the diagnostic apparatus may acquire the DCM image and obtain the diagnosis assistance information using the trained neural network model. .

In the diagnostic process, the acquired diagnosis subject image may be resized, similar to the image data used in the learning process. The image to be diagnosed may be shaped to have an appropriate dose, size, and / or aspect ratio in order to efficiently perform diagnostic assistance information prediction through the trained diagnostic assistance neural network model.

For example, when the diagnosis target image is a fundus image, resizing such as cropping an unnecessary portion of the image or reducing its size may be performed to predict diagnosis information based on the fundus image.

In the diagnosis process, a preprocessing filter may be applied to the obtained diagnosis target image, similar to the image data used in the learning process. An appropriate filter may be applied to the diagnosis target image so that the accuracy of the prediction of the diagnosis assistance information through the trained diagnosis assistance neural network model is further improved.

For example, if the image to be diagnosed is a fundus image, preprocessing for facilitating prediction of positive diagnostic information, such as image preprocessing for highlighting blood vessels or image preprocessing for highlighting or weakening a specific color, may be applied to the image to be diagnosed. have.

In the diagnostic process, the obtained diagnostic subject image may be serialized, similar to the image data used in the learning process. The diagnostic target image may be converted or serialized into a form that facilitates driving a diagnostic model in a specific work frame.

Serialization of the diagnosis target image may be omitted. This may be because, unlike in the learning stage, the processor does not have a large number of data processed at a time in the diagnostic stage, so that the burden on the data processing speed is relatively small.

In the diagnosis process, the acquired diagnosis subject image may be stored in a queue, similar to the image data used in the learning process. However, in the diagnostic process, since the number of processed data is smaller than that of the learning process, the step of storing the data in the queue may be omitted.

On the other hand, in the diagnostic process, since an increase in the number of data is not required, it may be desirable not to use the data enhancement or image enhancement procedure unlike the learning process in order to obtain accurate diagnosis assistance information.

1.3.4 Diagnostic process

According to one embodiment of the invention, a diagnostic process using a trained neural network model can be initiated. The diagnostic process can be performed in the above-described diagnostic apparatus. The diagnostic process can be performed at the aforementioned diagnostic server. The diagnostic process may be performed by the controller of the above-described diagnostic apparatus. The diagnostic process may be performed by the diagnostic module of the above-described diagnostic unit.

20 is a diagram for describing a diagnosis process according to an exemplary embodiment. Referring to FIG. 20, the diagnosis process may be performed by obtaining data to be diagnosed (S2010), by using a learned neural network model (S2030), and by obtaining a result corresponding to the obtained diagnosis target data (S2050). However, processing of the data may be selectively performed.

Hereinafter, each step of the diagnostic process will be described with reference to FIG. 20.

1.3.4.1 Data entry

According to an embodiment of the present invention, the diagnostic module may acquire diagnostic target data. The obtained data may be processed data as described above. For example, the acquired data may be fundus image data of a subject to whom the pre-processing to adjust the size and emphasize the blood vessel is applied. According to an embodiment of the present invention, the left eye image and the right eye image of one subject may be input together as diagnosis target data.

1.3.4.2 Data classification

The diagnostic assistance neural network model provided in the classifier form may classify the input diagnosis target image into a positive or negative class for a predetermined label.

The trained diagnostic auxiliary neural network model may receive diagnostic target data and output a predicted label. The trained diagnostic assistance neural network model may output a predicted value of the diagnostic assistance information. Diagnostic assistance information may be obtained using the trained diagnostic assistance neural network model. Diagnostic assistance information may be determined based on the predicted label.

For example, the diagnostic assisted neural network model may predict diagnostic information (ie, information about the presence of a disease) or finding information (ie, information about the presence or absence of abnormalities) of a subject's eye disease or systemic disease. In this case, the diagnostic information or findings may be output in a probability form. For example, the probability that the subject has a specific disease or the probability that there is a specific abnormality of the fundus image of the subject may be output. In the case of using the diagnostic assistance neural network model provided in the classifier form, the predicted label may be determined in consideration of whether the output probability value (or prediction score) exceeds the threshold value.

As a specific example, the diagnosis-assisted neural network model may output the presence or absence of diabetic retinopathy of a subject as a probability value by using the fundus photograph of the subject as a diagnosis target image. In the case of using the diagnostic assistive neural network model of the classifier type having 1 as normal, the fundus photograph of the subject is input to the diagnostic assistive neural network model, and whether the subject has diabetic retinopathy or not has a normal: abnormal probability value of 0.74: 0.26. It can be obtained in the form of.

Herein, the data is classified based on a classifier-type diagnostic auxiliary neural network model. However, the present invention is not limited thereto, and specific diagnostic auxiliary values may be defined using the diagnostic auxiliary neural network model implemented as a regression model. For example, blood pressure).

According to another embodiment of the present invention, suitability information of an image may be obtained. The suitability information may indicate whether the diagnosis target image is suitable for obtaining diagnostic assistance information using the diagnostic assistance neural network model.

The suitability information of the image may be quality information of the image. The quality information or suitability information may indicate whether the diagnosis target image reaches the reference level.

For example, when the diagnosis target image has a defect due to a defect of the photographing equipment or the influence of illumination when the image is taken, a non-conformance result may be output as the suitability information on the diagnosis target image. If the diagnosis target image contains noise more than a certain level, the diagnosis target image may be determined to be inappropriate.

The suitability information may be a value predicted using a neural network model. Alternatively, the suitability information may be information obtained through a separate image analysis process.

According to an embodiment, even when an image is classified as inappropriate, diagnostic assistance information obtained based on the inappropriate image may be obtained.

According to one embodiment, the image classified as inappropriate may be reviewed by a diagnostic aid neural network model.

In this case, the diagnostic assistance neural network model for performing the review may be different from the diagnostic assistance neural network model for performing the initial review. For example, the diagnostic apparatus may store the first diagnostic auxiliary neural network model and the second diagnostic auxiliary neural network model, and an image classified as inappropriate through the first diagnostic auxiliary neural network model may be reviewed through the second diagnostic auxiliary neural network model. .

According to another embodiment of the present invention, a class activation map (CAM) may be obtained from a trained neural network model. Diagnostic assistant information may include CAM. The CAM can be obtained along with other diagnostic assistance information.

In the case of a CAM, it may optionally be obtained. For example, in the case of CAM, when the diagnostic information or finding information obtained by the diagnostic assistance model is classified into an abnormal class, the CAM may be extracted and / or output.

1.3.5 Output of diagnostic assistant information

Diagnostic assistance information may be determined based on labels predicted from the diagnostic assistance neural network model.

The output of the diagnostic assistance information may be performed by the above-described output module of the diagnostic unit. The diagnostic assistance information may be output from the diagnostic apparatus to the client apparatus. The diagnostic assistance information may be output from the diagnostic apparatus to the server apparatus. The diagnostic assistance information may be stored in a diagnostic device or a diagnostic server. The diagnostic assistance information may be stored in a server device provided separately.

Diagnostic assistant information can be managed in a database. For example, the acquired diagnosis assistance information may be stored and managed together with the diagnosis target image of the subject according to the identification number of the subject. In this case, the diagnosis target image and the diagnosis assistance information of the subject may be managed in a time sequence. By managing the diagnosis assistant information and the diagnosis target image in time series, tracking and history management of individual diagnosis information may be facilitated.

Diagnostic assistance information may be provided to the user. The diagnostic assistance information may be provided to the user through the output means of the diagnostic apparatus or the client apparatus. The diagnostic assistance information may be output for the user to recognize through visual or audio output means provided in the diagnostic apparatus or the client apparatus.

According to an embodiment of the present invention, an interface for effectively providing diagnostic assistance information to a user may be provided. This user interface will be described in more detail later in 5. User Interface.

When the CAM is obtained by the neural network model, an image of the CAM may be provided together. In the case of a CAM image, it may optionally be provided. For example, when the diagnostic information acquired through the diagnosis assisted neural network model is normal finding information or normal diagnostic information, the CAM image is not provided, and when the obtained diagnostic information is abnormal finding information or abnormal diagnostic information, for more accurate clinical diagnosis. CAM images can be provided together.

If the image is classified as inappropriate, the suitability information of the image may be provided together. For example, when the image is classified as inappropriate, the diagnostic assistance information and the non-conformance determination information acquired according to the image may be provided together.

The diagnosis subject image determined to be inappropriate may be classified as a retake subject image. In this case, the reshooting guide for the target object of the image classified as the reshooting target may be provided together with the suitability information.

Meanwhile, in response to providing the diagnostic assistance information obtained through the neural network model, feedback related to the learning of the neural network model may be obtained. For example, feedback may be obtained for adjusting parameters or hyperparameters related to training of the neural network model. Feedback may be obtained through a user input provided in the diagnostic apparatus or the client apparatus.

According to an embodiment of the present invention, the diagnostic assistance information corresponding to the diagnosis target image may include grade information. The grade information may be selected from among a plurality of grades. The grade information may be determined based on diagnostic information and / or finding information obtained through the neural network model. The grade information may be determined in consideration of suitability information or quality information of the diagnosis target image. If the neural network model is a classifier model that performs multiple classifications, the rating information may be determined in consideration of the class of the diagnosis target image classified by the neural network model. If the neural network model is a regression model that outputs a value associated with a particular disease, the grade information may be determined in consideration of the output value.

For example, the diagnostic assistance information acquired corresponding to the diagnosis target image may include any one grade information selected from the first grade information or the second grade information. The grade information may be selected as the first grade information when abnormal finding information or abnormal diagnosis information is obtained through the neural network model. The grade information may be selected as second grade information when abnormal finding information or abnormal diagnosis information is not obtained through the neural network model. Alternatively, the grade information may be selected as the first grade information when the value obtained through the neural network model exceeds the reference value, and may be selected as the second grade information when the value obtained is less than the reference value. The first grade information may indicate that strong abnormality information exists in the diagnosis target image, compared to the second grade information.

Meanwhile, the grade information may be selected as the third grade information when it is determined that the quality of the diagnosis target image is lower than a reference by using an image analysis or a neural network model. Alternatively, the diagnostic assistant information may include third grade information together with the first or second grade information.

When the diagnostic assistance information includes the first grade information, the first user guide may be output through the output means. The first user guide may indicate that a more precise examination is required for the subject (ie, the patient) corresponding to the diagnosis assistance information. For example, the first user guidance may indicate that a secondary diagnosis (eg, a separate medical institution or a power supply procedure) for the subject is required. Alternatively, the first user guide may instruct treatment required for the subject. As a specific example, when abnormal information about macular degeneration of a subject is obtained by the diagnostic assistant information, the first user guide may provide instructions on injection prescription and power supply procedure for the subject (eg, List).

When the diagnostic assistance information includes the second grade information, the second user guide may be output through the output means. The second user guide may include a later management method for the subject corresponding to the diagnosis assistance information. For example, the second user guide may indicate a next medical treatment time, a next medical treatment subject, or the like.

When the diagnosis target information includes the third grade information, the third user guide may be output through the output means. The third user guide may indicate that re-photographing is required for the diagnosis target image. The third user guide may include information about the quality of the diagnosis target image. For example, the third user guide may include information of defects (eg, whether bright artifacts or dark artifacts or the like) present in the diagnosis target image.

1.4 Multiple Labels for Diagnostic Aid System

According to an embodiment of the present invention, a diagnostic assistance system for predicting a plurality of labels (eg, a plurality of diagnostic assistance information) may be provided. To this end, the diagnostic assistance neural network of the aforementioned diagnostic assistance system may be designed to predict a plurality of labels.

Alternatively, in the above-described diagnostic assistance system, a plurality of diagnostic assistance neural networks that predict different labels may be used in parallel. Hereinafter, such a parallel diagnostic assistance system will be described.

1.4.1 Parallel Diagnostic Assist System Configuration

According to an embodiment of the present invention, a parallel diagnostic assistance system for acquiring a plurality of diagnostic assistance information may be provided. The parallel diagnostic assistance system may learn a plurality of neural network models for acquiring a plurality of diagnostic assistance information, and obtain a plurality of diagnostic assistance information using the trained plurality of neural network models.

For example, the parallel diagnostic assistance system acquires a first neural network model for obtaining first diagnostic assistance information related to the presence or absence of an eye disease in a subject and second diagnostic assistance information related to the presence or absence of a systemic disease in the subject based on the fundus image. The second neural network model may be trained, and diagnostic assistance information regarding the presence or absence of eye diseases and systemic diseases of the subject may be output using the learned first neural network model and the second neural network model.

21 and 22 are diagrams for describing a parallel diagnostic assistance system according to some embodiments of the present invention. 21 and 22, the parallel diagnosis assistance system may include a plurality of learning units.

Referring to FIG. 21, the parallel diagnosis assistance system 30 according to an embodiment of the present invention may include a learning apparatus 1000, a diagnosis apparatus 2000, and a client apparatus 3000. In this case, the learning apparatus 1000 may include a plurality of learning units. For example, the learning apparatus 1000 may include a first learner 100a and a second learner 100b.

Referring to FIG. 22, the parallel diagnosis assistance system 40 according to an embodiment of the present invention may include a first learning device 1000a, a second learning device 1000b, a diagnosis device 2000, and a client device 3000. It may include. The first learning apparatus 1000a may include a first learning unit 100a. The second learning apparatus 1000b may include a second learning unit 100b.

21 and 22, the first learner 100a may obtain a first data set and output a first parameter set of the first neural network model obtained as a result of learning with respect to the first neural network model. The second learner 100b may obtain a second data set and output a second parameter set of the second neural network model obtained as a result of learning with respect to the second neural network model.

The diagnosis apparatus 2000 may include a diagnosis unit 200. The description described with reference to FIG. 1 may be similarly applied to the diagnosis apparatus 2000 and the diagnosis unit 200. The diagnosis unit 200 obtains the first diagnosis assistance information and the second diagnosis assistance information by using the first neural network model and the second neural network model trained from the first learner 100a and the second learner 100b, respectively. can do. The diagnosis unit 2000 may store the parameters of the trained first neural network model acquired from the first learner 100a and the second learner 100b and the parameters of the trained second neural network model.

The client device 3000 may include a data acquirer, for example, an imaging unit 300. However, the imaging unit 300 may be replaced with a means for acquiring data used to acquire other diagnostic assistance information. The client device may transmit a diagnosis request and diagnosis target data (eg, a fundus image obtained from the image pickup unit) to the diagnostic device. In response to transmitting the diagnostic request, the client device 3000 may obtain a plurality of diagnostic assistance information according to the transmitted diagnostic target data from the diagnostic apparatus.

Meanwhile, in FIGS. 21 and 22, the diagnosis assistance system 40 has been described based on the case in which the diagnostic assistance system 40 includes the first learner 100a and the second learner 100b. However, the present disclosure is not limited thereto. According to another embodiment of the present invention, the learning apparatus may include a learning unit for acquiring three or more different diagnostic assistance information. Alternatively, the diagnostic assistance system may include a plurality of learning apparatuses for obtaining different diagnostic assistance information.

More specific operations of the learning apparatus, the diagnostic apparatus, and the client apparatus will be described in more detail below.

1.4.2 Parallel training process

According to an embodiment of the present invention, a plurality of neural network models may be learned. The training process for learning each neural network model can be performed in parallel.

1.4.2.1 Parallel learning

The training process may be performed by a plurality of learning units. Each training process may be performed independently of each other. The plurality of learning units may be provided in one learning device or in each of the plurality of learning devices.

FIG. 23 is a diagram for describing a configuration of a learning apparatus including a plurality of learning units according to an exemplary embodiment. Configuration and operation of each of the first learner 100a and the second learner 100b may be implemented similarly to those described above with reference to FIG. 9.

Referring to FIG. 23, a process of a neural network model according to an embodiment of the present invention may include obtaining a first data processing module 110a, a first queue module 130a, a first learning module 150a, and a first learning result. The first learning unit 100a and the second data processing module 110b including the module 170a, the second cue module 130b, the second learning module 150b, and the second learning result obtaining module 170b may be used. It may be performed by the learning apparatus 1000 including the second learning unit 100b.

Referring to FIG. 23, the training process of the neural network model according to an embodiment of the present invention may be performed by the first learner 100a and the second learner 100b, respectively. The first learner 100a and the second learner 100b may independently learn the first neural network model and the second neural network model. Referring to FIG. 23, the first learner 100a and the second learner 100b may be provided in the above-described learning apparatus. Alternatively, the first learner and the second learner may be provided in different learning devices.

1.4.2.2 Parallel data acquisition

According to an embodiment of the present invention, the plurality of learning units may acquire data. The plurality of learners may acquire different data sets. Alternatively, the plurality of learners may obtain the same data set. In some cases, the plurality of learners may acquire a data set, some of which are common. The data set may be a fundus image data set.

The first learner may obtain a first data set, and the second learner may obtain a second data set. The first data set and the second data set can be distinguished. The first data set and the second data set may be some common. The first data set and the second data set may be labeled fundus image data sets.

The first data set may include data labeled as normal for the first feature and data labeled as abnormal for the first feature. For example, the first data set may include a fundus image labeled as normal and a fundus image labeled as abnormal with respect to lens clouding.

The second data set may include data normally labeled for the second feature (distinguish from the first feature) and data labeled abnormally for the second feature. For example, the second data set may include a fundus image labeled as normal and a fundus image labeled as abnormal with respect to diabetic retinopathy.

The data set labeled normally for the first feature included in each of the first data set and the second data set and the data labeled normally for the second feature may be common. For example, the first data set includes a fundus image labeled as normal with respect to lens cloudiness and a fundus image labeled as abnormal, and the second data set includes a fundus image and abnormal labeled as normal with diabetes retinopathy. The fundus image labeled as being included, but the fundus image labeled as normal with respect to the lens clouding included in the first data set and the fundus image labeled as normal with respect to diabetic retinopathy included in the second data set will be common. Can be.

Alternatively, data abnormally labeled with respect to the first feature included in each of the first data set and the second data set and data abnormally labeled with respect to the second feature may be common. That is, data labeled for multiple features may be used for training neural network models for multiple features.

Meanwhile, the first data set may be the fundus image data set photographed by the first method, and the second data set may be the fundus image data set photographed by the second method. The first method and the second method may be any one selected from red-free photographing, panorama photographing, autofluorescence photographing, infrared photographing, and the like.

The data set used in each learning unit may be determined in consideration of diagnostic assistance information obtained by the neural network to be learned. For example, when the first learner trains a first neural network model to acquire diagnostic assistance information related to retinal abnormalities (eg, microvascular perfusion, exudate, etc.), the red fundographed first fundus image data You can obtain a set. Alternatively, when the second learner learns a second neural network model for acquiring diagnosis assistance information related to macular degeneration, the second learner may acquire a second fundus image data set by autofluorescence.

1.4.2.3 Parallel data processing

The plurality of learners may process the obtained data, respectively. As described in the 2.2 data processing process described above, each learner may process the data by applying one or more of an image resizing, a preprocessing filter, an image enhancement, and an image serialization process. The first data processing module of the first learner may process the first data set, and the second data processing module of the second learner may process the second data set.

The first learner and the second learner included in the plurality of learners may differently process the obtained data set in consideration of diagnostic assistance information obtained from the neural network model learned by the learner. For example, the first learner may perform preprocessing to cause blood vessels to be emphasized with respect to the fundus images included in the first fundus image data set in order to train the first neural network model for obtaining first diagnostic assistance information related to hypertension. Can be done. Alternatively, the second learner may learn about the fundus images included in the second fundus image data set in order to train the second neural network model for acquiring second diagnostic assistance information related to abnormal findings such as exudates of the retina and microvascular vessels. You can also perform preprocessing to convert to a red-free image.

1.4.2.4 Parallel queue

The plurality of learners may store data in a queue. As described in 2.2.6 Queues above, each learner can store processed data in a queue and deliver it to the learning module. For example, the first learner may store the first data set in the first cue module and provide the first data set sequentially or randomly. The second learning module may store the second data set in the second cue module and provide the second learning module sequentially or randomly.

1.4.2.5 Parallel learning process

The plurality of learners may train the neural network model. Each learning module may independently train a diagnostic assisted neural network model that predicts for different labels using a training data set. The first learning module of the first learning unit trains the first neural network model, and the second learning module of the second learning unit trains the second neural network module.

The plurality of diagnostic assistance neural network models can be trained in parallel and / or independently. By learning the model to predict different labels through the plurality of neural network models as described above, the prediction accuracy for each label can be improved and the efficiency of the prediction operation can be increased.

Each diagnostic assisted neural network model can be prepared similar to that described in 2.3.2 Model Design. Each sub-learning process may be performed similar to that described above in 2.3.1 to 2.3.5.

The parallel learning process according to an embodiment of the present invention may include training a diagnostic assisted neural network model predicting different labels. The first learner may train the first diagnostic assistance neural network model predicting the first label. The second learner may train a second diagnostic assistance neural network model that predicts the second label.

The first learner may train a first diagnostic assisted neural network model that obtains a first data set and predicts a first label. For example, the first learner may train a first diagnostic assisted neural network model predicting macular degeneration of a subject from an ocular fundus image by using a fundus image training data set labeled for macular degeneration.

The second learner may train a first diagnostic assisted neural network model that obtains a second data set and predicts a second label. For example, the second learner may train a second diagnostic assistive neural network model for predicting whether a subject is diabetic retinopathy from the fundus image, using the fundus image training data set labeled for diabetic retinopathy.

The learning process of the neural network model will be described in more detail with reference to FIGS. 24 and 25 below.

24 is a diagram for describing a parallel learning process according to an embodiment of the present invention. The parallel learning process may be applied to both the case where the parallel diagnosis assistance system is implemented as shown in FIG. 21, the case where it is implemented as shown in FIG. 22, and other forms. However, for convenience of description, the following description will be made based on the parallel diagnosis assistance system implemented as shown in FIG. 21.

Referring to FIG. 24, the parallel learning process may include a plurality of sub-learning processes each learning a plurality of diagnostic assistance neural network models predicting different labels. The parallel learning process may include a first sub learning process for learning the first neural network model and a second sub learning process for learning the second neural network model.

For example, the first sub-learning process acquires first data (S1010a), uses a first neural network model (S1030a), verifies a first model (ie, a first diagnostic auxiliary neural network model) (S1050a) This may be performed by obtaining parameters of the neural network model (S1070a). The second sub-learning process acquires second data (S1010b), uses the second neural network model (S1030b), verifies the second neural network model (ie, the second diagnostic auxiliary neural network model) (S1050b), and the second neural network. This may be performed by obtaining parameters of the model (S1070b).

The sub-learning process may include training the neural network model by inputting training data into the sub-neural network model, verifying the model by comparing the label value obtained as the output with the input training data, and reflecting the verification result back to the sub-neural network model. have.

Each sub-learning process obtains a result by using a neural network model to which arbitrary weight values are assigned, compares the obtained result with a label value of training data, and performs backpropagation according to the error. May include optimizing them.

In each sub-learning process, the diagnostic assistance neural network model may be verified through a validation data set that is distinct from the training data set. Validation data sets for verifying the first neural network model and the second neural network model may be distinguished.

The plurality of learning units may acquire a learning result. Each learning result obtaining module may obtain information about the learned neural network module from the learning module. Each learning result obtaining module may obtain parameter values of the neural network module learned from the learning unit. The first learning result obtaining module of the first learning unit may obtain a first parameter set of the first neural network model learned from the first learning module. The second learning result obtaining module of the second learning unit may obtain a second parameter set of the second neural network model learned from the second learning module.

By each sub-learning process, optimized parameter values, i.e., a set of parameters, of the learned neural network model can be obtained. As learning proceeds with more training data sets, more appropriate parameter values can be obtained.

A first set of parameters of the first diagnostic assistance neural network model learned by the first sub learning process may be obtained. A second parameter set of the second diagnostic assistance neural network model learned by the second sub learning process may be obtained. As the learning proceeds sufficiently, optimized values of the weight and / or bias of the first diagnostic auxiliary neural network model and the second diagnostic auxiliary neural network model can be obtained.

The parameter set of each obtained neural network model may be stored in a learning device and / or a diagnostic device (or server). The first parameter set of the first diagnostic assistance neural network model and the second parameter set of the second diagnostic assistance neural network model may be stored together or separately. The parameter set of each neural network model learned may be updated by feedback obtained from a diagnostic device or a client device.

1.4.2.6 Parallel Ensemble Learning Process

Even when training a plurality of neural network models in parallel, the above-described ensemble type model learning may be used. Each sub learning process may comprise training a plurality of sub neural network models. The plurality of submodels may have different hierarchical structures. In the following, the contents described in 2.3.7 may be similarly applied unless otherwise stated.

When a plurality of diagnostic assistive neural network models are trained in parallel, some of the sub-learning processes that train each diagnostic assistive neural network model train a single model, and some sub-learning processes work together a plurality of submodels. It may be implemented in the form of learning.

As the model is trained using the ensemble in each sub-learning process, the shape of a more optimized neural network model is obtained in each sub-process, and the error of prediction may be reduced.

25 illustrates a parallel learning process according to another embodiment of the present invention. Referring to FIG. 25, each learning process may include training a plurality of sub-neural network models.

Referring to FIG. 25, the first sub-learning process obtains first data (S1011a), uses a 1-1 neural network model and a 1-2 neural network model (S1031a, S1033a) and a 1-1 neural network model. The second neural network model may be verified (S1051a), and the final shape and the parameters of the first neural network model may be determined (S1071a). The second sub-learning process acquires second data (S1011b), uses the 2-1 neural network model and the 2-2 neural network model (S1031b, S1033b), the 2-1 neural network model and the 2-2 neural network model. In operation S1051b, the final form of the first model (ie, the first diagnostic assistance neural network model) and its parameters may be determined (S1071b).

The first neural network learned in the first sub-learning process may include a 1-1 neural network model and a 1-2 neural network model. The 1-1 neural network model and the 1-2 neural network model may be provided in different hierarchical structures. The first-first neural network model and the first-two neural network model may obtain the first data set and output the predicted labels, respectively. Alternatively, the predicted label by the ensemble of the 1-1 neural network model and the 1-2 neural network model may be determined as the final predictive label.

In this case, the 1-1 neural network model and the 1-2 neural network model may be verified using the verification data set, and a highly accurate neural network model may be determined as the final neural network model. Alternatively, the ensemble of the 1-1 neural network model, the 1-2 neural network model, the 1-1 neural network model, and the 1-2 neural network model is verified, and the neural network model form in the case of high accuracy is used as the final first neural network. You can also decide by model.

Similarly with respect to the second sub-learning process, a highly accurate neural network of the ensemble of the 2-1 neural network model, the 2-2 neural network model, the 2-1 neural network model, and the 2-2 neural network model is final. Model (ie, a second diagnostic assistance neural network model).

Meanwhile, in FIG. 25, for convenience, each sub-learning process includes two sub-models, but this is merely an example, and the present invention is not limited thereto. The neural network model trained in each sub-learning process may include only one neural network model or three or more submodels.

1.4.3 Parallel diagnostic process

According to an embodiment of the present invention, a diagnostic process for acquiring a plurality of diagnostic assistance information may be provided. The diagnostic process for acquiring the plurality of diagnostic assistance information may be implemented in the form of a parallel diagnostic assistance process including a plurality of diagnostic processes independent of each other.

1.4.3.1 Parallel diagnostics

According to an embodiment of the present invention, the diagnostic assistance process may be performed by a plurality of diagnostic modules. Each diagnostic assistance process can be performed independently.

26 is a block diagram illustrating a diagnosis unit 200 according to an embodiment of the present invention.

Referring to FIG. 26, the diagnostic unit 200 according to an embodiment of the present invention may include a diagnostic request obtaining module 211, a data processing module 231, a first diagnostic module 251, and a second diagnostic module 253. And an output module 271. Each module of the diagnostic unit 200 may operate similarly to the diagnostic module of the diagnostic unit illustrated in FIG. 18 unless otherwise noted.

In FIG. 26, even when the diagnosis unit 200 includes a plurality of diagnostic modules, the diagnostic request obtaining module 211, the data processing module 231, and the output module 271 are illustrated as being common. It is not limited to this configuration, a plurality of diagnostic request acquisition module, data processing module and / or output module may also be provided. A plurality of diagnostic request acquisition modules, data processing modules and / or output modules may also operate in parallel.

For example, the diagnosis unit 200 may include a first data processing module that performs a first processing on an input diagnosis target image and a second processing module that performs second data processing on a diagnosis target image. The first diagnostic module may acquire first diagnostic assistance information based on the first processed diagnosis target image, and the second diagnostic module may acquire second diagnostic assistance information based on the second processed diagnosis target image. The first and / or second processing may be any one selected from image resizing, color modulation of the image, blur filter application, blood vessel highlighting, red-free transformation, partial region cropping, and some element extraction.

The plurality of diagnostic modules may acquire different diagnostic assistance information. The plurality of diagnostic modules may obtain diagnostic assistance information using different diagnostic assistance neural network models. For example, the first diagnostic module acquires first diagnostic assistance information related to the subject's eye disease using a first neural network model that predicts whether the subject has eye disease, and the second diagnostic module detects the eye disease. By using a second neural network model that predicts whether a subject is a systemic disease, second diagnostic assistance information related to a subject's systemic disease may be obtained.

In more specific example, the first diagnostic module acquires first diagnostic assistance information on whether the subject is diabetic retinopathy by using a first diagnostic assistive neural network model that predicts whether the subject is diabetic retinopathy based on the fundus image. The second diagnostic module may acquire second diagnostic assistance information related to hypertension of the subject using a second diagnostic assistive neural network model predicting whether the subject is hypertension based on the fundus image.

1.4.3.2 Parallel diagnostic process

The diagnostic assistance process according to an embodiment of the present invention may include a plurality of subdiagnostic processes. Each sub-diagnostic process may be performed using a different diagnostic assistance neural network model. Each sub-diagnostic process may be performed in a different diagnostic module. For example, the first diagnostic module may perform a first sub-diagnosis process of acquiring first diagnostic assistance information through the first diagnostic assistance neural network model. Alternatively, the second diagnostic module may perform a second sub-diagnosis process of obtaining second diagnostic assistance information through the second diagnostic assistance neural network model.

The trained plurality of neural network models may input diagnostic target data and output predicted labels or probabilities. Each neural network model is provided in the form of a classifier, and the input diagnosis target data may be classified with respect to a predetermined label. In this case, the plurality of neural network models may be provided in the form of a classifier learned about different characteristics. Each neural network model can classify the data to be diagnosed as described in 3.4.2.

On the other hand, from each diagnostic assistance neural network model, a CAM can be obtained, and the CAM can be obtained selectively. The CAM can be extracted if a predetermined condition is satisfied. For example, when the first diagnostic assistance information indicates that the subject is abnormal for the first characteristic, the first CAM may be obtained from the first diagnostic assistance neural network model.

27 is a diagram for describing a diagnostic assistance process according to one embodiment of the present invention.

Referring to FIG. 27, the diagnostic assistance process according to an embodiment of the present invention acquires data to be diagnosed (S2011), and diagnoses using the first diagnostic assistance neural network model and the second diagnostic assistance neural network model (S2031a and S2031b). It may include acquiring (S2051) diagnostic assistance information according to the target data. The data to be diagnosed may be processed data.

According to an embodiment of the present invention, the diagnostic assistance process may include obtaining first diagnostic assistance information through a trained first diagnostic assistance neural network model, and obtaining second diagnostic assistance information through a trained second diagnostic assistance neural network model. It may include. The first diagnostic assistance neural network model and the second diagnostic assistance neural network model may obtain first diagnostic assistance information and second diagnostic assistance information based on the same diagnosis target data.

For example, the first diagnosis aid neural network model and the second diagnosis aid neural network model are based on the diagnosis target fundus image, the first diagnosis aid information on whether the subject is macular degeneration and the second diagnosis on whether the subject is diabetic retinopathy 2 diagnostic assistance information can be obtained respectively.

In addition, unless otherwise indicated, the diagnostic assistance process described with reference to FIG. 27 may be implemented similarly to the diagnostic assistance process described above with reference to FIG. 20.

1.4.3.3 Output of diagnostic assistant information

According to an embodiment of the present invention, diagnostic assistance information obtained by a parallel diagnostic assistance process may be obtained. The obtained diagnostic assistance information may be stored in the diagnostic apparatus, the server apparatus, and / or the client apparatus. The obtained diagnostic assistance information may be delivered to an external device.

The plurality of diagnostic assistance information may indicate a plurality of labels predicted by the plurality of diagnostic assistance neural network models, respectively. The plurality of diagnostic assistance information may correspond to a plurality of labels predicted by the plurality of diagnostic assistance neural network models, respectively. Alternatively, the diagnostic assistance information may be information determined based on a plurality of labels predicted by the plurality of diagnostic assistance neural network models. The diagnostic assistance information may correspond to a plurality of labels predicted by the plurality of diagnostic assistance neural network models.

In other words, the first diagnostic assistance information may be diagnostic assistance information corresponding to the first label predicted through the first diagnostic assistance neural network model. Alternatively, the first diagnostic assistance information may be diagnostic assistance information determined by considering the first label predicted through the first diagnostic assistance neural network model and the second label predicted through the second diagnostic assistance neural network model.

Meanwhile, an image of the CAM obtained from the plurality of diagnostic assistance neural network models may be output. The CAM image may be output when a predetermined condition is satisfied. For example, when the first diagnostic assistance information indicates that the subject is abnormal for the first characteristic or when the second diagnostic assistance information indicates that the subject is abnormal for the second characteristic, The CAM image obtained from the diagnostic assistance neural network model outputting the diagnostic assistance information indicated as abnormal may be output.

The plurality of diagnostic assistance information and / or CAM images may be provided to the user. The plurality of diagnostic assistance information and the like may be provided to the user through output means of the diagnostic apparatus or the client apparatus. Diagnostic assistance information may be visually output. In this regard, it is described in more detail in 5. User interface.

According to an embodiment of the present invention, the diagnostic assistance information corresponding to the diagnosis target image may include grade information. The grade information may be selected from among a plurality of grades. The grade information may be determined based on the plurality of diagnostic information and / or finding information obtained through the neural network model. The grade information may be determined in consideration of suitability information or quality information of the diagnosis target image. The grade information may be determined in consideration of a class in which a diagnosis target image is classified by a plurality of neural network models. The grade information may be determined in consideration of numerical values output from the plurality of neural network models.

For example, the diagnostic assistance information acquired corresponding to the diagnosis target image may include any one grade information selected from the first grade information or the second grade information. The grade information may be selected as first grade information when at least one abnormality finding information or abnormal diagnosis information is obtained among the diagnostic information obtained through the neural network model. The grade information may be selected as the second grade information when abnormal finding information or abnormal diagnosis information is not obtained from the diagnosis information acquired through the neural network model.

The grade information is selected as the first grade information when at least one of the values acquired through the neural network model exceeds the reference value, and the grade information is selected as the second grade information when all of the acquired values fall below the reference value. May be The first grade information may indicate that strong abnormality information exists in the diagnosis target image, compared to the second grade information.

The grade information may be selected as the third grade information when it is determined that the quality of the diagnosis target image is equal to or lower than a reference value by using an image analysis or a neural network model. Alternatively, the diagnostic assistant information may include third grade information together with the first or second grade information.

When the diagnostic assistance information includes the first grade information, the first user guide may be output through the output means. The first user guide may include at least one abnormality finding information included in the diagnosis assistance information or a matter corresponding to the abnormal diagnosis information. For example, the first user guide may indicate that a more precise examination is required for a subject (ie, a patient) corresponding to abnormal information included in the diagnosis assistance information. For example, the first user guidance may indicate that a secondary diagnosis (eg, a separate medical institution or a power supply procedure) for the subject is required. Alternatively, the first user guide may instruct treatment required for the subject. As a specific example, when abnormal information about macular degeneration of a subject is obtained by the diagnostic assistant information, the first user guide may provide instructions on injection prescription and power supply procedure for the subject (eg, List).

When the diagnostic assistance information includes the second grade information, the second user guide may be output through the output means. The second user guide may include a later management method for the subject corresponding to the diagnosis assistance information. For example, the second user guide may indicate a next medical treatment time, a next medical treatment subject, or the like.

When the diagnosis target information includes the third grade information, the third user guide may be output through the output means. The third user guide may indicate that re-photographing is required for the diagnosis target image. The third user guide may include information about the quality of the diagnosis target image. For example, the third user guide may include information of defects (eg, whether bright artifacts or dark artifacts or the like) present in the diagnosis target image.

The first to third grade information may be output by an output unit of the client device or the diagnostic device. Specifically, it may be output through a user interface described later.

1.4.4 Example 2-Diagnostic Assistance Systems

The diagnostic assistance system according to an embodiment of the present invention may include a fundus image obtaining unit, a first processing unit, a second processing unit, a third processing unit, and a diagnostic information output unit.

According to an embodiment of the present invention, the diagnostic assistance system may include a diagnostic device. The diagnostic apparatus may include a fundus image acquirer, a first processor, a second processor, a third processor, and / or a diagnostic information output unit. However, the present invention is not limited thereto, and each unit included in the diagnosis assistance system may be located at appropriate locations on the learning device, the diagnosis device, the learning diagnosis server, and / or the client device. Hereinafter, for convenience, a diagnosis apparatus of the diagnosis assistance system will be described based on a case where the fundus image obtaining unit, the first processing unit, the second processing unit, the third processing unit, and the diagnostic information output unit are included.

28 is a diagram for describing a diagnosis assistance system according to an embodiment of the present invention. Referring to FIG. 28, the diagnostic assistance system may include a diagnostic apparatus, and the diagnostic apparatus may include a fundus image obtaining unit, a first processing unit, a second processing unit, a third processing unit, and a diagnostic information output unit.

According to an embodiment of the present invention, a diagnosis assisting system for assisting diagnosis of a plurality of diseases based on the fundus image may include a fundus image obtaining unit configured to acquire an object fundus image, which is a basis for obtaining diagnosis assistance information about a subject. And obtaining a first result associated with a first finding on the subject using a first neural network model for the subject fundus image, wherein the first neural network model is machine learned based on the first set of fundus images. The subject using a first processing unit, a second neural network model with respect to the target fundus image, wherein the second neural network model is machine learned based on a second set of fundus images different from at least a portion of the first fundus image set; A second processing unit for obtaining a second result related to a second finding about, based on the first result and the second result, for the subject The processing unit may comprise 3, and the user parts of the diagnostic information output for providing the determined diagnostic information for determining the diagnostic information. In this case, the first finding and the second finding may be used for diagnosis of different diseases.

The first neural network model is trained to classify the input fundus image into any one of a normal label and an abnormal label in relation to the first finding, and the first processing unit uses the first neural network model to classify the target fundus image into the normal label. Alternatively, the first result may be obtained by classifying into one of the abnormal labels.

The third processor may determine whether the diagnostic information based on the target fundus image is normal information or abnormal information by considering the first result and the second result together.

The third processing unit may determine the diagnostic information on the subject by giving priority to the abnormal label so that diagnostic accuracy is improved.

 The third processing unit determines that the diagnostic information is normal when the first label is a normal label for the first finding, and the second label is a normal label for the second finding, and the first label is the first label. If the label is not a normal label for the finding or the second label is not a label for the second finding, the diagnostic information may be abnormally determined.

The first finding is associated with an ocular disease, and the first result may indicate whether the subject is normal for the ocular disease. The second finding is associated with systemic disease, and the second result may indicate whether the subject is normal for the systemic disease.

A first eye is associated with a first eye disease, the first result indicates whether the subject is normal for the first eye disease, and the second eye is a second eye that is distinct from the first eye disease Associated with a disease, the second result may indicate whether the subject is normal for the second eye disease.

 The first finding is a finding for diagnosing a first eye disease, the first result indicates whether the subject is normal to the first eye disease, and the second finding is for diagnosing the first eye disease. The findings are distinguished from the first findings, and the second result may indicate whether the subject is normal to the second eye disease.

The first neural network model includes a first sub neural network model and a second sub neural network model, and the first result is a first prediction value predicted by the first sub neural network model and a second predicted by the second sub neural network model. It may be determined by considering the prediction value together.

The first processor may acquire a Class Activation Map (CAM) associated with the first label through the first neural network model, and the diagnostic information output unit may output an image of the CAM.

The diagnostic information output unit may output an image of the CAM when the diagnostic information acquired by the third processing unit is abnormal diagnostic information.

The diagnosis assistance system may further include a fourth processor configured to acquire quality information of the target fundus image, and the diagnostic information output unit may output quality information of the target fundus image obtained by the fourth processor.

 When it is determined by the fourth processor that the quality information of the target fundus image is equal to or less than a predetermined quality level, the diagnostic information output unit is further configured to provide the user with the determined diagnostic information. Information indicating that the following can be provided together.

1.5 User interface

According to an embodiment of the present invention, the aforementioned client device or diagnostic device may have a display unit for providing the diagnostic assistance information to the user. In this case, the display unit may be provided to clearly transmit the diagnostic assistance information to the user, and to easily obtain feedback from the user.

As an example of the display unit, a display for providing visual information to a user may be provided. In this case, a graphic user interface for visually delivering the diagnostic assistance information to the user may be used. For example, in the fundus diagnosis assistance system for acquiring the diagnosis assistance information based on the fundus image, a graphic user interface for effectively displaying the acquired diagnosis assistance information and helping the user to understand may be provided.

29 and 30 are diagrams for describing a graphical user interface for providing diagnostic information to a user, according to some embodiments of the present disclosure. Hereinafter, with reference to FIGS. 29 and 30, some embodiments of the user interface that can be used in the fundus diagnosis assistance system will be described.

Referring to FIG. 29, a user interface according to an embodiment of the present invention may display identification information of a subject corresponding to a diagnosis fundus image. The user interface may include a target image identification information display unit 401 that displays identification information of a subject (ie, a patient) and / or photographing information (eg, a photographing date) of a diagnosis target fundus image.

The user interface according to an embodiment of the present invention may include a fundus image display unit 405 for displaying a fundus image of a left eye and a right eye of a same subject, respectively. The fundus image display unit 405 may display a CAM image.

The user interface according to an embodiment of the present invention indicates that the image of the left eye or the right eye for each of the left eye fundus image and the right eye fundus image, and a diagnostic information indicator indicating whether the diagnosis information and the user confirmation of each image are displayed. The diagnostic information indicating unit 403 may be included.

The color of the diagnostic information indicator may be determined in consideration of diagnostic assistance information obtained based on the target fundus image. The diagnostic information indicator may be displayed in a first color or a second color according to the diagnostic assistance information. For example, when the first to third diagnostic assistance information is acquired in one target fundus image, when even one diagnostic assistance information includes abnormal (that is, abnormal) information, the diagnostic information indicator is displayed in red. In the case where all the diagnostic assistance information includes normal (ie, no abnormality) information, the diagnostic information indicator may be displayed in green.

The type of diagnostic information indicator may be determined according to whether the user confirms it. The diagnostic information indicator may be displayed in a first form or a second form depending on whether the user confirms the information. For example, referring to FIG. 25, the diagnostic information indicator corresponding to the target fundus image reviewed by the user is indicated by a filled circle, and the diagnostic information indicator corresponding to the target fundus image unreviewed by the user is filled with a semicircle. Can be displayed.

The user interface according to an embodiment of the present invention may include a diagnostic information indicating unit 407 for indicating the diagnostic assistance information. The diagnostic assistance information indicator may be located in the left eye and right eye images, respectively. The diagnostic assistance information indicating unit may indicate a plurality of findings or diagnostic information.

The diagnostic assistance information indicator may include at least one diagnostic assistance information indicator. The diagnostic assistance information indicator may indicate corresponding diagnostic assistance information through color change.

For example, a second diagnosis aid indicating diagnosis of diabetic retinopathy through the first diagnosis aid information indicating the presence of lens turbidity through the first diagnosis aid neural network model and the second diagnosis aid neural network model with respect to the diagnosis fundus image. Information and third diagnostic assistance information indicating whether there is a retinal abnormality finding through the third diagnostic assistance neural network model, the diagnostic information indicating unit may include the first diagnostic assistance information, the second diagnostic assistance, and the third diagnostic assistance information, respectively. It may include a first to third diagnostic assistant information indicator indicating.

More specifically, referring to FIG. 29, in relation to the left eye fundus image, the diagnostic information indicating unit 407 may include first diagnostic assistance information indicating that the diagnosis aid information acquired based on the left eye fundus image of the subject is abnormal lens turbidity. , The first diagnostic assistant information having the first color when the second diagnostic assistant information indicating diabetic retinopathy normal (no abnormal findings) and the third diagnostic assistant information indicating the retinal abnormality (with abnormal findings) are obtained. The indicator, the first diagnostic assistant information indicator having a second color, and the third diagnostic assistant information indicator having a first color may be displayed.

The user interface according to an embodiment of the present invention may obtain a user comment on the diagnosis target fundus image from the user. The user interface may include a user comment object 409 and display a user input window in response to a user selection for the user comment object. The comments obtained from the user may be used to update the diagnostic assistance neural network model. For example, the user input window displayed in response to the selection of the user comment object may obtain a user evaluation of the diagnostic assistance information through the neural network, and the obtained user evaluation may be used to update the neural network model.

The user interface according to an embodiment of the present invention may include a review instruction object 411 indicating whether the user reviews the respective fundus image. The review instruction object may receive a user input indicating that the user review of each diagnosis target image is completed, and the display may be changed from the first state to the second state. For example, referring to FIGS. 29 and 30, the review instruction object may be changed to a second state indicating that the user input is confirmed in the first state displaying the confirmation request text.

The list 413 of the diagnosis target fundus images may be displayed. In the list, identification information of the subject, an image photographing date, and an indicator of whether to review the binocular image may be displayed together.

In the list 413 of the diagnosis target fundus images, a review completion indicator 415 indicating the reviewed diagnosis fundus image may be displayed. The review completion indicator 415 may be displayed when user selection has occurred for all of the review indicating object 411 for both eyes of the corresponding image.

Referring to FIG. 30, when it is determined that there is a quality abnormality in the diagnosis fundus image, the user graphic interface may include a low quality warning object 417 which indicates to the user that there is an abnormality in the quality of the target fundus image. have. The low quality warning object 417 is displayed when the diagnosis fundus image is judged by the diagnosis unit to be less than the quality (that is, the reference quality level) of the level at which the appropriate diagnosis aid information can be predicted from the diagnosis aid neural network model. Can be.

In addition, referring to FIG. 28, the low quality warning object 419 may be displayed in the list 413 of the diagnosis fundus images.

2. Judging the quality of the fundus image

2.1 Introduction

2.1.1 Background and purpose

In the present specification, a method and apparatus for detecting an image having a flaw in order to perform diagnosis of a lesion using an image will be described.

In order to perform a particular judgment based on the image, for example, diagnosis of a lesion, etc., it must be assumed that the image is properly captured and has a quality higher than a reference. Therefore, it is preferable to not use an image that does not have a quality suitable for performing a diagnosis due to a defect generated during an imaging process or an image processing process, so that the image prior to the diagnosis based on the image is not used. Can be preceded by a quality judgment.

For example, even when a diagnosis of a specific disease is to be performed on a subject based on the fundus image, it may be appropriate to be used as the basis of the diagnosis only for the fundus image whose quality is guaranteed to a certain level or more. In the case of the fundus image, artifacts may occur according to the angle of incidence of light or the distance to the fundus during the photographing process, and in the case of an image having such an artifact, it is difficult to use as a basis for diagnosis. It needs to be followed.

When diagnosing a disease based on an image having an artifact, it may be difficult to accurately diagnose the disease. Thus, a method such as an operator, a program manager, or the like can directly determine the quality of a captured image has been proposed and used.

However, if there are many images to be processed, it may be inefficient for a person to directly determine the quality of each image. In particular, when a large number of images are used, such as learning artificial intelligence based on the images, a person's direct determination of the quality of a large amount of images, respectively, can lead to a serious waste of human resources.

In addition, the judgment standard of a person is subjective, and may vary according to the person who performs the judgment, and even a judgment by the same person may be difficult to expect a consistent judgment standard for all images.

In the present specification, an image sorting method and apparatus for quickly and accurately determining an image quality and providing a certain quality criterion will be described with reference to some embodiments. Hereinafter, the method and apparatus for selecting the fundus image for convenience of description will be described, but the content of the invention described herein is not limited thereto. The invention described herein may be applied or modified in various cases where image quality determination is required.

2.1.2 Fundus images

In the present specification, the fundus image may refer to an image of the inside of the eye that is viewed through the pupil. The fundus image may mean an image in which at least some of the retina, retinal blood vessel, optic nerve papilla, and choroid are photographed.

The fundus image may be a panoramic image. The fundus image may be a red-free image. The fundus image may be an infrared photographed image. The fundus image may be an autofluorescence image. The image data may be obtained in any one of JPG, PNG, DCM (DICOM), BMP, GIF, and TIFF.

FIG. 31 is a diagram illustrating region division of the fundus image FI according to an embodiment of the present invention. According to an embodiment of the present invention, the fundus image FI may include a plurality of regions associated with an imaging frame or an imaging condition of the fundus image.

Referring to FIG. 31, the fundus image FI may include a fundus region FA in which the fundus is imaged and a non-fungus region NFA in which the fundus is not included. In some cases, the fundus image FI may not include the non- fundus region NFA. The fundus image FI may include a fundus region and a non-fungus region NFA separated by the structure of the imaging device or the camera control application.

FIG. 32 is a diagram illustrating region division of a fundus image according to another embodiment of the present invention. FIG. Referring to FIG. 32, the fundus region FA of the fundus image FI may include an inner region IA and an outer region OA.

The inner region IA and the outer region OA may be divided based on the boundary BO. The boundary BO may be concentric with the periphery of the fundus region. The boundary BO can be determined circularly. The boundary BO may be determined as a circle having the center of the fundus region as its center and having a predetermined radius R. The outer area OA may be determined as an area spaced apart from the center of the fundus area by a predetermined distance R or more. The inner region IA may be determined as a region where the distance from the center of the fundus region is R or less.

The boundary BO may be determined such that the ratio of the width of the inner area IA to the width of the outer area OA has a predetermined value. The boundary BO may be determined such that the ratio of the width of the inner region IA to the width of the fundus region has a predetermined value. For example, the boundary BO may be determined such that the width of the inner region IA defined by the boundary BO corresponds to 80% of the fundus region.

Herein, the description has been made based on the case where the inner region IA and the outer region OA are divided based on the boundary BO, but this is not a limitation of the present invention, and the inner region IA and the outer region OA are described. May be determined to overlap some.

FIG. 33 is a diagram illustrating region division of a fundus image according to another embodiment of the present invention. FIG. According to an embodiment of the present invention, the fundus image may include a plurality of regions determined in consideration of positions of anatomical elements included in the fundus image.

Referring to FIG. 33, the fundus area FA may include the macular part MA, the optic nerve head part ODA, and the periphery part. The fundus image may be divided to include the macula, optic nerve head and periphery based on the location of the macula and / or optic nerve head detected from the fundus image.

The macular portion MA may be determined to include a region in which the macular of the fundus region is distributed. The optic nerve head (ODA) may be determined to include a region in which the optic nerve papilla of the fundus region is distributed. The periphery may be determined as an area excluding the macular part MA and the optic nerve head part ODA from the fundus area.

The boundaries of the macular (MA), optic nerve papilla (ODA), and periphery may be determined based on the location of the macular and / or optic nerve papilla. The boundary of the macular portion MA may be determined to surround the area where the macular is distributed. The boundary of the optic nerve head (ODA) may be determined to surround an area where the optic nerve head is distributed.

In some cases, the periphery may be divided into a plurality of detailed areas. Referring to FIG. 33, an upper part SNA, a lower part INA, an upper part STA, a lower part ITA, and a side TA may be included. The upper part SNA, the lower part INA, the upper part STA, the lower part ITA and the side part TA may be distinguished based on the positions of the macula and / or optic nerve papilla.

The boundaries of the upper part SNA, the lower part INA, the upper part STA, the lower part ITA and the side part TA may be distinguished based on the positions of the macular and / or optic nerve papilla. For example, the upper abdomen is considered in consideration of a first straight line connecting the position of the macula (eg, the approximate center point) and the position of the optic nerve head (eg, the approximate center point) and a second straight line perpendicular to the first straight line and passing through the position of the optic nerve head. (SNA), hobby INA, upper part STA, lower part ITA and side TA may be determined. Further, for example, the side portion TA may be determined in consideration of the third straight line and the fourth straight line parallel to the first straight line and in contact with the macular portion MA.

32 and 33 illustrate the case where each area is divided by a boundary indicated by a dotted line, but according to an embodiment, each area may be defined to partially overlap.

The areas and / or boundaries described above with respect to FIGS. 31-33 may be used consistently for different images. Alternatively, the regions and / or boundaries described above may be appropriately determined and used for each image. The above-described areas and / or boundaries may be determined and used by user input. The division of the fundus image described with reference to FIGS. 31 to 33 is merely an example, and a boundary or a detail region of the fundus image for separating the fundus image may be implemented in various forms. For example, the detail area of the fundus image may be divided by the user as needed.

The fundus image can be used to diagnose ocular disease. The fundus image can be used to diagnose ocular diseases such as glaucoma, cataracts, macular degeneration and prematurity retinopathy. Fundus images include hemorrhages, exudate, cotton patches, drusen & drusenoid deposits, turbidity, degeneration, proliferative tissue, vascular abnormalities, retinal pigmentary changes, macular abnormalities, macular perforation (macular hole), choroidal atrophy (chroioretinal atrophy) can be used for the diagnosis of lesions.

The fundus image can be used to diagnose other diseases (eg, systemic or chronic diseases) that are not eye diseases. For example, fundus images can be used to diagnose diseases such as diabetes (or diabetic retinopathy), Alzheimer's, cytomegalovirus, stroke, and the like. In another example, the fundus image can be used to diagnose cardiovascular disease (heart disease, coronary artery disease, etc.). The fundus image may be used for diagnosis of hypertension, hypotension, atherosclerosis, angina pectoris, heart failure, and the like.

From the fundus image, findings for diagnosing ocular disease or other disease can be identified. For example, from an ocular fundus image, color abnormalities across the fundus, opacity of the lens, abnormality of the C / D ratio (cup to disc ratio), macular abnormality (eg, macular hole), blood vessels Abnormalities such as diameter, running, abnormalities of retinal arteries, retinal hemorrhage, exudation, drusen, and the like can be identified.

2.2  Image Management System and Devices

2.2.1 System and operation

According to an embodiment of the present invention, an image management system and / or apparatus for determining the quality of a fundus image and managing the fundus image may be provided. The image management system and / or the device may perform the image quality determining method described herein. The method of determining the quality of the fundus image disclosed herein may be performed by the above-described diagnostic assistance system or a device configuring the diagnostic assistance system.

The image management system and / or device, image quality management system and / or device, etc., disclosed herein may be implemented similarly to the diagnostic assistance system and / or device described above with respect to FIGS. In other words, the diagnostic assistance system described above with reference to FIGS. 1 to 9 may be implemented as an image management system that manages the quality of the fundus image.

For example, the image management system may include a learning device, a diagnostic device, and a client device, which may operate similar to the diagnostic assistance system of FIG. 1. The learning device, diagnostic device, and / or client device may perform the quality judgment of the fundus image and / or the suitability determination of the fundus image described herein. The diagnostic device (or image management device) included in the image management system may operate similarly to the diagnostic device described with reference to FIG. 4 or 5.

As another example, the image management system may include a diagnostic device and a client device. The diagnostic device may perform the functions of the learning device and / or the server device. The diagnostic device and / or the client device may perform the quality determination of the fundus image and / or the suitability determination of the fundus image described herein.

As another example, the image management system may further include a server device. The server device may operate similar to that described above with respect to FIG. 6. The server device may perform the quality determination of the fundus image and / or the suitability determination of the fundus image described herein.

As another example, the image management system may include a mobile device. The mobile device may perform all or part of the operations of the learning device, diagnostic device and / or client device described above. The mobile device may communicate with a server device. The mobile device may perform quality determination of the fundus image and / or suitability determination of the fundus image described herein.

Although the above has been described with reference to some examples of the image management system, even if not described above, the configuration of the image management system may be similarly constructed from the description related to the diagnostic assistance system described with reference to FIGS. 1 to 9. It can be obvious.

2.2.2 Device and operation

The diagnostic assistance systems 10 and 20 and the devices included in the system described above with reference to FIGS. 1 to 9 may perform the quality determination of the fundus image.

According to an embodiment, the learning apparatus 1000 may include a quality determiner that determines the quality of the fundus image. The learning apparatus 1000 may include a suitability determination unit that determines suitability of the fundus image. The learning apparatus 1000 may determine the quality of the fundus data for learning by using the quality determining unit, and may learn the neural network model using the selected image as a result of the quality determining. The learning apparatus 1000 may determine suitability for the target disease of the fundus image by using the suitability determination unit, and may learn a neural network model by using the selected image as a result of the suitability determination.

 The quality determining unit or suitability determining unit of the learning apparatus 1000 may be included in the control unit 1200 of the learning apparatus 1000. The controller 1200 of the learning apparatus 1000 may perform an operation of the quality determining unit or the suitability determining unit of the learning apparatus 1000. The quality determining unit or suitability determining unit of the learning apparatus 1000 may be included in the processor 1050 of the learning apparatus 1000. The processor 1050 of the learning apparatus 1000 may perform an operation of the quality determining unit or the suitability determining unit of the learning apparatus 1000.

According to an embodiment of the present disclosure, the diagnostic apparatus 2000 may include a quality determination unit that determines the quality of the fundus image. The diagnosis apparatus 2000 may include a suitability determination unit that determines suitability of the fundus image. The diagnosis apparatus 2000 may acquire the fundus image of the subject and determine the quality of the acquired fundus image by using the quality determining unit. The diagnostic apparatus 2000 may acquire a fundus image of a subject and determine the quality of the acquired fundus image by using a suitability determination unit. The diagnosis apparatus 2000 may determine the quality of the fundus image by using the quality determination unit, and obtain diagnostic assistance information from the neural network model using the selected image. The diagnosis apparatus 2000 may determine suitability of the fundus image by using the suitability determination unit, and may obtain diagnostic assistance information from the neural network model using the selected image.

The quality determining unit or suitability determining unit of the diagnostic apparatus 2000 may be included in the control unit 2200 of the diagnostic apparatus 2000. The controller 2200 of the diagnostic apparatus 2000 may perform an operation of the quality determination unit or the suitability determination unit of the diagnostic apparatus 2000. The quality determining unit or suitability determining unit of the diagnostic apparatus 2000 may be included in the processor 2050 of the diagnostic apparatus 2000. The processor 2050 of the diagnostic apparatus 2000 may perform an operation of the quality determining unit or the conformity determining unit of the diagnostic apparatus 2000.

According to an embodiment of the present disclosure, the client device 3000 may include a quality determiner that determines the quality of the fundus image or a suitability determiner that determines the suitability of the fundus image. The client device 3000 may acquire the fundus image of the subject and determine the quality of the acquired fundus image by using the quality determination unit. The client device 3000 may acquire the fundus image of the subject and determine the quality of the acquired fundus image by using the suitability determination unit. The client device 3000 may determine the quality of the fundus image and transmit the selected image to an external device (eg, a diagnostic device or a server device) based on a quality determination result obtained using the quality determination unit.

The quality determining unit or suitability determining unit of the client device 3000 may be included in the controller 3300 or the processor of the client device 3000. The controller 3200 or the processor of the client device 3000 may perform an operation of the quality determination unit or the suitability determination unit of the client device 3000.

According to an embodiment, the server device 4000 may include a quality determination unit determining the quality of the fundus image or a suitability determination unit determining the suitability of the fundus image. The server device 4000 may determine the quality of the image acquired from the learning apparatus or the client apparatus using the quality determination unit. The server device 4000 may determine the quality of an image acquired from an external device (eg, a client device) using the quality determination unit, and perform diagnosis on the selected image as a result of the quality determination. Alternatively, the quality determining unit of the server device 4000 may determine the quality of the fundus image included in the fundus image data set, and transmit the selected image to an external device (eg, a client device or a diagnostic device) as a result of the quality determination. The server device 4000 may determine a quality of an image obtained from an external device (eg, a client device) using the suitability determination unit, perform a diagnosis on the selected image as a result of the suitability determination, or perform a diagnosis on the selected image, or transmit the selected image to an external device (eg, a client). Device or diagnostic device).

The quality determination unit or suitability determination unit of the server device 4000 may be included in a control unit or a processor of the server device. The controller or processor of the server device may perform an operation of the quality determination unit or the suitability determination unit of the server device 4000.

According to an embodiment, the mobile device may include a quality determining unit determining the quality of the fundus image or a suitability determining unit determining the suitability of the fundus image. The mobile device may include an imaging unit, and determine the quality of an image acquired through the imaging unit. For example, the quality determining unit of the mobile device may determine the quality of the captured image and perform diagnosis on the selected image as a result of the quality determination.

The quality determining unit or suitability determining unit of the mobile device may be included in the controller or processor of the mobile device. The controller or processor of the mobile device may perform an operation of the quality determination unit or the suitability determination unit of the mobile device.

34 and 35 are diagrams for describing an apparatus for performing quality judgment of a fundus image or suitability determination of a fundus image disclosed herein. 34 and 35 may be implemented in the form of the device described above with reference to FIGS.

34 is a diagram for describing an ocular fundus image management apparatus 5000 according to an exemplary embodiment. Referring to FIG. 34, the fundus image management apparatus 5000 according to an exemplary embodiment may include a fundus image acquirer 5010 and a quality determiner 5030. The fundus image acquirer 5010 and / or the quality determiner 5030 may be provided in a controller or a processor of the apparatus.

The fundus image acquirer 5010 may capture or acquire a fundus image. The fundus image acquirer 5010 may acquire a fundus image through a camera or may acquire a fundus image from an external device. The fundus image may be normalized to facilitate quality determination of the format, size, and the like of the acquired fundus image.

The quality determining unit 5030 may determine the quality of the fundus image. The quality determining unit 5030 may perform the quality determination of the fundus image described herein. The quality determination unit 5030 may perform quality determination in consideration of the area. The quality determination unit 5030 may perform quality determination in consideration of artifacts. The quality determination unit 5030 may perform quality determination in consideration of the pixel value.

Referring to FIG. 34, the fundus image management apparatus 5000 according to an exemplary embodiment may further include a quality information manager 5050. The quality information manager 5050 may be provided in a controller, a processor, or a memory of the apparatus.

The quality information manager 5050 may manage quality information of the fundus image. The quality information manager 5050 may assign the quality information obtained as a result of the quality determination of the fundus image to the fundus image. The quality information manager 5050 may manage the fundus image data set based on the quality information matched to the fundus image. For example, the quality information manager 5050 may classify and store the fundus image data based on the quality information.

According to an embodiment of the present invention, a fundus image management apparatus for detecting an artifact related to the quality of the fundus image from the fundus image photographing the fundus and managing the quality of the fundus image may include a fundus image acquisition unit for acquiring the fundus image; The apparatus may include a quality determiner configured to detect a first pixel from a target region of the fundus image and determine the quality of the fundus image in consideration of the number of the first pixels.

The artifact may be either a bright artifact due to excessive reflection of light occurring in the fundus region of the fundus image or a dark artifact due to a shadow occurring in the fundus region of the fundus image.

The quality determiner may include: a first pixel detector configured to detect a first pixel related to the quality of the fundus image by comparing pixel values of at least one pixel included in a target area of the fundus image with a reference pixel value; A quality of acquiring quality information of the fundus image based on a comparison result obtained by comparing the ratio of the first pixel to the reference ratio and the comparison ratio obtained by comparing the ratio of the first pixel to the reference ratio among the total pixels of the fundus image. It may include an information acquisition unit.

The target area may include a first area spaced at least a first distance from the center of the area corresponding to the fundus of the fundus image, a second area within a first distance from the center of the area corresponding to the fundus, and an optic nerve papilla among the areas corresponding to the fundus. It may include at least one selected from the third region corresponding to the region in which is distributed and the fourth region corresponding to the region in which the macular is distributed among the regions corresponding to the fundus.

If the artifact is a first artifact, the target region is determined to be a first region spaced at least a first distance from the center of the region corresponding to the fundus of the fundus image, and if the artifact is a second artifact, the target region is the fundus of the fundus image May be determined as a second region in which is distributed. In this case, the first distance may be determined such that an area within a first distance from the center of the area corresponding to the fundus is equal to or greater than a first ratio with respect to the entire fundus image area.

If the artifact is a first artifact, the reference pixel value is a first pixel value, and the step of detecting the first pixel by comparison with the reference pixel value includes a first pixel included in the target area and having a pixel value that is greater than the reference pixel value. Detecting, the reference ratio is a first reference ratio, and comparing with the reference ratio further includes determining whether a ratio of the first pixel of all pixels of the fundus image is greater than the first reference ratio, and the comparison result The acquiring of the quality information may further include determining that the fundus image is a bad image when the ratio of the first pixel to the entire pixel of the fundus image is greater than the first reference ratio.

If the artifact is a second artifact, the reference pixel value is a second pixel value, and the step of detecting the first pixel by comparison with the reference pixel value includes a first pixel that is included in the target area and has a pixel value that is less than the reference pixel value. Detecting, the reference ratio is a second reference ratio, and comparing with the reference ratio further includes determining whether a ratio of the first pixel of all pixels of the fundus image is less than the second reference ratio, and comparing The acquiring of the quality information based on the result may further include determining that the fundus image is a bad image when the ratio of the first pixel of the total pixels of the fundus image is smaller than the second reference ratio.

The fundus image management apparatus may further include a quality information management unit that adds a quality information label to the fundus image based on the quality information and manages the fundus image according to the quality information label.

According to another embodiment of the present invention, a fundus image management apparatus for detecting an artifact related to the quality of the fundus image from the fundus image photographing the fundus and determining the quality of the fundus image, the fundus image acquisition unit acquiring the fundus image And a quality determination unit that detects at least one artifact pixel from a target region of the fundus image and determines the quality of the fundus image in consideration of the artifact pixel.

The quality determiner may include a target region determiner that determines at least a portion of the fundus image as a target region for detecting artifacts, and an artifact pixel detector that detects at least one artifact pixel corresponding to the artifact from the target region.

If the artifact is the first artifact, the quality determining unit determines the target region as the first region that is spaced a predetermined distance away from the center of the fundus image corresponding to the fundus, and when the artifact is the second artifact, the target region is included in the fundus image. It may be determined as the fundus area, which is the area where the fundus is distributed.

The first artifact may be a bright artifact due to excessive reflection of light generated in the fundus region of the fundus image, and the second artifact may be a dark artifact due to shadow occurring in the fundus region of the fundus image.

The fundus image management apparatus may further include a converting unit configured to obtain a transformed fundus image by grayscale converting the fundus image. The quality determining unit may determine the quality of the fundus image based on the converted fundus image.

The artifact pixel detector includes a comparison unit that compares pixel values of a plurality of pixels included in the target region with a reference pixel value range, and at least one pixel having a pixel value included in the target region and included in the target pixel value range. The determination unit may further include determining.

The comparison unit compares pixel values of the plurality of pixels included in the target area with the first reference pixel value when the artifact is the first artifact, and pixels of the plurality of pixels included in the target area when the artifact is the second artifact. The value may be compared with the second reference pixel value.

If the artifact is the first artifact, the determiner determines at least one pixel included in the target area and having a pixel value greater than the first reference pixel value as the first artifact pixel, and when the artifact is the second artifact, At least one pixel included in the area and having a pixel value smaller than the second reference pixel value may be determined as the second artifact pixel.

The quality determining unit may determine the quality of the fundus image in consideration of whether the amount of the detected artifact exceeds the reference amount.

The quality determining unit determines whether the number of detected artifact pixels exceeds a reference ratio with respect to the total number of pixels of the fundus image, and when the number of artifact pixels exceeds the reference ratio with respect to the total number of pixels of the fundus image, the fundus image is an artifact. It may be determined to include.

The fundus image management apparatus may further include a quality information manager configured to add a quality information label generated based on the quality information to the fundus image and manage the fundus image according to the quality information label.

35 is a diagram for describing the fundus image management apparatus 6000 according to an exemplary embodiment. Referring to FIG. 35, the fundus image management apparatus 6000 may include a fundus image acquirer 6010 and a suitability determiner 6030. The fundus image acquirer 6010 and / or the suitability determiner 6030 may be provided in a controller or a processor of the apparatus.

The fundus image acquirer 6010 may acquire a fundus image through a camera or may acquire a fundus image from an external device. The fundus image may be normalized to facilitate quality determination of the format, size, and the like of the acquired fundus image.

The suitability determination unit 6030 may determine suitability for the target disease of the fundus image. The suitability determination unit 6030 may perform suitability determination on the image described herein. The suitability determination unit 6030 may determine the suitability of the fundus image in consideration of the target region of the fundus image. The suitability determination unit may determine suitability of the fundus image in consideration of the presence or absence of a target artifact. The suitability determination unit may determine suitability of the fundus image in consideration of a pixel value, a distribution, and the like.

Referring to FIG. 35, the fundus image management apparatus 6000 according to an exemplary embodiment may further include a suitability information manager 6050. The suitability information manager 6050 may be provided in a controller, a processor, or a memory of the apparatus.

The suitability information manager 6050 may manage suitability information of the fundus image. The suitability information manager 6050 may assign suitability information obtained as a result of the suitability determination of the fundus image to the fundus image. The suitability information manager 6050 may manage the fundus image data set based on the suitability information matched to the fundus image. For example, the suitability information manager 6050 may classify and store the fundus image based on the suitability information. The suitability information manager 6050 may store and manage the fundus images having the same target suitability information together.

In connection with the diagnosis of the target disease based on the fundus image according to an embodiment of the present invention, the fundus image management apparatus for determining the suitability of the fundus image, the target disease for the fundus image acquisition unit and the fundus image to acquire the fundus image It may include a suitability determination unit for determining the target suitability of the fundus image by using a quality standard corresponding to. The subject disease may include at least one of a first disease and a second disease.

The suitability determination unit may determine whether the fundus image satisfies the first quality criterion when the target disease is the first disease, and determine whether the fundus image satisfies the second quality standard when the target disease is the second disease.

The quality criterion may include a quality criterion for the target area. The determination of the suitability of the fundus image by the suitability determining unit determines whether the first region of the fundus image satisfies the first quality criterion when the target disease is the first disease, and when the target disease is the second disease, The method may further include determining whether the region satisfies the second quality criterion.

Quality criteria may include quality criteria for the presence or absence of artifacts. The suitability determination unit, when the target disease is the first disease, determining the first quality criterion includes determining whether the image includes a first artifact corresponding to the first disease, and when the target disease is the second disease, Determining the second quality criterion can further include determining whether the fundus image includes a second artifact corresponding to the second disease.

The first artifact corresponding to the first disease may be detected from a first region associated with the first disease, and the second artifact corresponding to the second disease may be detected from a second region associated with the second disease. In other words, the first artifact having a corresponding relationship with the first disease may correspond to the first region having a corresponding relationship with the first disease. The matching relationship between the disease and the region, the disease and the artifact, the artifact and the region may be previously stored and used in the suitability determination apparatus or the quality determination apparatus. Alternatively, the above-described matching relationship may be specified by the user. The above-described matching relationship may be learned and generated according to the neural network model.

According to an embodiment, the ocular fundus image management apparatus may further include a diagnostic information acquisition unit configured to acquire diagnostic information regarding a target disease based on the ocular fundus image.

The diagnosis information acquiring unit acquires first diagnostic assistance information for the first disease when the fundus image satisfies the first quality criterion when the target disease is the first disease, and when the target disease is the second disease, the fundus image is reconstructed. If the second quality criterion is satisfied, the second diagnostic assistance information for the second disease may be obtained.

According to an embodiment, the fundus image management apparatus may further include a suitability information acquisition unit that acquires suitability information of the fundus image based on the determination result.

The fitness information acquiring unit acquires fitness information indicating that the fundus image is suitable for the first disease when the target disease is the first disease and the fundus image satisfies the first quality criterion, and the target disease is the second disease. When the fundus image meets the second quality criterion, suitability information indicating that the fundus image is suitable for the second disease may be obtained.

 The subject disease may include a first disease and a second disease. The suitability determination unit may determine whether the fundus image satisfies the first quality criterion and the second quality criterion.

The suitability information acquiring unit, when it is determined that the fundus image satisfies the first quality criterion but does not satisfy the second quality criterion, the suitability information indicating that the fundus image is suitable for the first disease and not suitable for the second disease. Can be obtained.

According to another embodiment of the present invention, in relation to the diagnosis of the target disease based on the fundus image, the fundus image management apparatus for determining the suitability of the fundus image, the fundus image acquisition unit for acquiring the fundus image, the quality of the fundus image Based on the quality determining unit to determine and whether the fundus image satisfies the quality criteria, it may include a suitability determination unit for determining the target suitability for the target disease of the fundus image.

The quality determiner may determine whether the first quality criterion associated with the first region is satisfied and whether the fundus image satisfies the second quality criterion associated with the second region that is at least partially distinguished from the first region. have.

The quality determiner may determine the presence or absence of the first artifact in consideration of the first area, and determine whether or not the second artifact is in consideration of the second area.

The quality determining unit determines whether the fundus image includes the first artifact by determining whether the fundus image includes the first artifact, and determines whether the fundus image satisfies the second quality standard by determining whether the fundus image includes the second artifact. Can be.

The first area may be an area spaced a predetermined distance or more from the center of the fundus area in which the fundus of the fundus image is distributed. The second area may be an area within a predetermined distance from the center of the fundus area.

In addition, the first region or the second region may be an outer region spaced at least a first distance from the center of the region corresponding to the fundus of the fundus image, an inner region within a first distance from the center of the region corresponding to the fundus, and the fundus. It may include at least one selected from the optic nerve papilla region corresponding to the region where the optic nerve papilla is distributed and the macular region corresponding to the region where the macula is distributed among the regions corresponding to the fundus. The first area or the second area may be an area that includes at least a part of the detail area of the fundus image described with reference to FIGS. 31 to 33 or excludes at least a part thereof.

The suitability determination unit determines that the fundus image satisfies the first suitability when the target disease is the first disease and the fundus image satisfies the first quality criterion, and the target disease is the second disease and the fundus image uses the second quality criterion. If satisfied, it may be determined that the fundus image satisfies the second suitability.

The suitability determination unit may have a first suitability but no second suitability for the fundus image that satisfies the first quality criterion but does not satisfy the second quality criterion when the target disease is the first disease and the second disease. You can judge. In this case, the fundus image determined to have the first suitability but not the second suitability may be suitable as a diagnosis basis of the first disease, but may be inappropriate as a disease basis of the second disease. The fundus image determined to have the first suitability but not the second suitability may be interpreted as being suitable for the first disease group to which the first disease belongs, but inappropriate to the second disease group to which the second disease belongs. .

According to an embodiment, the ocular fundus image management apparatus may further include a diagnostic information acquisition unit configured to acquire diagnostic information regarding a target disease based on the ocular fundus image.

If the target disease is the first disease, the diagnostic information acquisition unit acquires first diagnostic assistance information for the first disease when the fundus image satisfies the first quality criterion, and when the target disease is the second disease, the fundus image is acquired. If the second quality criterion is satisfied, the second diagnostic assistance information for the second disease may be obtained.

The subject disease may include a first disease and a second disease. In this case, when the fundus image satisfies the first quality criterion and does not satisfy the second quality criterion, the diagnostic information obtaining unit acquires first diagnostic assistance information for the first disease and does not acquire the second diagnostic assistance information. You can not. Alternatively, the diagnostic information acquisition unit may include the first diagnostic assistance information for the first disease and the second diagnostic assistance information for the second disease even when the fundus image satisfies the first quality standard and does not satisfy the second quality standard. Acquisition, but may be stored along with the second diagnostic assistance information dissatisfaction history of the second quality standard. The diagnostic assistance information stored together and the dissatisfaction details of the quality standard may be provided to the user.

Hereinafter, the method for determining the quality of the image, the method for determining suitability, and the like described in the present specification may be performed by the system or apparatus described with reference to FIGS. 1 to 9. Alternatively, the quality determination method, the suitability determination method, and the like described in this specification may be performed by the processor or the controller described with reference to FIGS. 1 to 9. The quality determination method, the suitability determination method, and the like described in the present specification may be performed by at least one of the system, the device, and the processor (or the controller) described with reference to FIGS. 1 to 9.

2.3 How to judge image quality

2.3.1 Normal

As described above, defects may occur in the fundus image depending on the photographing state. For example, an object may be detected that interferes with the observation of the fundus element from the fundus image. Alternatively, the brightness of the image may be higher or lower than the appropriate range. For example, artifacts may occur in the process of capturing an image.

An artifact may refer to a photographed state or an image in which such a state occurs so as to obstruct identification of a subject. For example, bright artifacts can occur when light is excessively reflected from the fundus. Or dark artifacts can occur when shadows are in the fundus. Alternatively, reflection artifacts caused by reflected light may occur. In addition, various types of artifacts may occur in the fundus image that make it difficult to diagnose the diagnosis or fundus element based on the image.

36 illustrates some examples of possible abnormalities in the fundus image. Referring to FIG. 63, a bright artifact (a), a dark artifact (b), or haze (c) may occur in the fundus image. Each abnormality may be slightly different from the occurrence position and the sun, and considering the difference, an image quality determining method for detecting each abnormality may be implemented differently.

Bright artifacts can be caused by optical reflections, bright borders, and the like. Bright artifacts may occur at locations where blood vessels or optic nerve papillas are distributed. Bright artifacts may be distributed close to the outside of the fundus image.

Dark artifacts may appear due to shadows, etc., generated during imaging of the fundus image. Dark artifacts may be distributed close to the center of the fundus image. Dark artifacts and bright artifacts may occur together in the same image.

Cloudiness can occur due to incorrect focusing or contamination on the optical path during fundus image capture. Turbidity can be distributed throughout the fundus region of the fundus image. Haze can appear due to clouding of the lens or vitreous. Cloudy artifacts can themselves be used as a basis for diagnosis of a particular disease or finding.

When the fundus image has artifacts, it may be difficult to identify elements included in the image, so it may be inappropriate to perform a fundus examination or obtain other information based on the fundus image having the artifacts.

In particular, in the case of training the artificial intelligence model to perform diagnosis based on the fundus image, it is essential to use a plurality of images as training data, but in the case of an image including an artifact as described above, element identification will be difficult. There is a possibility of negatively affecting the learning of the model. Therefore, separate management is required for images including artifacts.

In this regard, the following describes a method and apparatus for detecting artifacts occurring in an image, evaluating the quality of an image with or without an artifact, or selecting a defective image.

Hereinafter, descriptions will be made based on artifacts occurring in the fundus image, in particular, bright artifacts or dark artifacts, but it is obvious that the contents of the invention described herein are not limited thereto, and the contents of the invention described herein. May be applied in various cases to determine an inappropriate image or an image that is inappropriate to obtain information.

2.3.2 Judgment method

According to an embodiment of the present invention, the quality of the image may be determined using at least one pixel value included in the image. Determining the quality of an image may mean various operations of evaluating whether an image satisfies a predetermined criterion, such as evaluating an image, selecting a bad image, or detecting a defect of the image. Determining the quality of the image may include determining whether the image includes the various types of artifacts described above.

37 is a flowchart illustrating a method of determining an image quality according to an embodiment of the present invention. Referring to FIG. 37, in the method of determining an image quality according to an embodiment of the present invention, the method may include obtaining a fundus image (S11), converting a fundus image (S13), and obtaining quality information of the fundus image ( S15) may be included.

Determination of the quality of an image according to an embodiment of the present invention may include acquiring a fundus image (S11). The fundus image may be a diagnosis target image or a learning target image.

The fundus image may be obtained through a fundus image pickup device. The fundus image may be obtained using a non-dong pupil fundus camera or the like generally used in ophthalmology. The fundus image may be obtained via a portable fundus imaging device. The fundus image may be obtained using a camera embedded in the mobile device. In place of or in conjunction with the fundus image, a fundus video may be acquired and / or used. According to an embodiment, the fundus image may be extracted from the fundus video. In this regard, a plurality of fundus image items to be described later will be described in more detail.

The fundus image may be obtained from an external server or a storage medium. Alternatively, the fundus image may be stored in advance. For example, the method for determining the quality of an image described herein may be performed by a predetermined information processing apparatus, and the at least one fundus image may be pre-stored in a memory of a predetermined information processing apparatus. The fundus image may be obtained via wired or wireless communication.

Determination of the quality of the image according to an embodiment of the present invention may include converting the fundus image (S13). In this case, the pixel value used for quality determination, which will be described later, may be obtained from the transformed fundus image.

Determination of the quality of the fundus image may include converting the fundus image to grayscale. The at least one pixel value included in the fundus image may be obtained from the grayscale fundus image after the obtained fundus image is grayscaled (grayscale).

Determination of the quality of the fundus image may include converting the fundus image having a plurality of channels to have a single channel. At least one pixel value may be obtained from a fundus image transformed to have a single channel. For example, the fundus image may comprise an RGB channel, and pixel values may be obtained from the R channel image.

Determination of the quality of the fundus image may include binarizing the fundus image. For example, the fundus image may be converted to a bitmap image. When the fundus image is binarized based on a predetermined pixel value, at least one pixel value may be obtained from the binarized fundus image.

Determination of the quality of the fundus image may include obtaining and / or converting a histogram of the fundus image. When the fundus image is histogram transformed, at least one pixel value may be obtained from the histogram transformed fundus image. In determining the quality of an image, a pixel value of at least one pixel may be obtained from a histogram normalized or equalized image.

Meanwhile, the step of converting the fundus image may be selectively used. Determination of the quality of the image may convert the fundus image, obtain quality information using the converted fundus image, or obtain quality information based on the original image without converting the fundus image.

Determination of the quality of the image according to an embodiment of the present invention may include acquiring quality information of the fundus image (S30).

Acquiring quality information of the fundus image may include determining whether the fundus image satisfies a quality condition. According to an embodiment, acquiring the quality information may include determining the plurality of quality conditions and acquiring the quality information in consideration of the determination result of the plurality of quality conditions.

The quality information of the image may be obtained in consideration of the above-described pixel value or the target pixel. The quality information of the image may be performed in consideration of whether the image satisfies the above-described quality conditions.

The quality information of the image may be normal information indicating whether the image is a normal image or an abnormal image. The quality information of the image may be defect information indicating whether the image contains defects or not. The defect information may further include specific information on what defect the image contains.

The quality information of the image may include artifact information on whether or not the artifact is included. The quality information of the image may include identification information of the artifact indicating the type of the artifact, and artifact information indicating whether the corresponding fundus image includes the artifact or the intensity of inclusion of the artifact.

The quality information of the image may be provided in probability form. For example, the normal information of the image may include the probability that the image is normal. The defect information of the image may include a probability that the image includes a defect or a probability that the image includes a plurality of defects, respectively.

The quality information of the image may further include information about an area. For example, when the target pixel is detected from the target area using the predetermined area as the target area, or when the quality condition is determined for the target area, the quality information of the area may be included. The quality information may include identification information of the target area and quality information about the target area.

For example, the quality information of the image may include normal information or defect information on the first area of the fundus image. The quality information of the image may further include information on the second area as well as information on the first area of the fundus image.

The quality information of the image may include a quality value, a quality score, a quality degree, and the like. The quality score may be calculated in consideration of the number, ratio, and the like of pixels corresponding to the artifact. The quality score and the like may be calculated in consideration of the type of artifact and / or the degree of the artifact. The quality score and the like may be obtained through a neural network model.

The target area may be determined as a part of the fundus image or the fundus area. The target region may be selected from a plurality of regions constituting the fundus image or the fundus region. The target area may be selected from the areas described with reference to FIGS. 31 to 33.

For example, referring to FIG. 32, the fundus region may include an inner region IA and an outer region OA. The target area may be determined as the outer area OA or the inner area IA. For example, when the object to be detected for quality determination is a bright artifact, the object area may be determined as the outer area OA.

For example, referring to FIG. 33, the target area may be the macular area MA, the optic nerve head part ODA, or the periphery part. Alternatively, the target area may be an area including at least one of the upper part SNA, the lower part INA, the upper part STA, the lower part ITA, and the side part TA. Alternatively, the target region may be from the fundus image or the fundus region, from the macular region (MA), the optic nerve papilla (ODA), the periphery, the upper part (SNA), the lower part (INA), the upper part (STA), and the lower part (ITA) It may be an area except at least one of the TAs.

Here, the description will be made based on the case where the target area is selected from areas of a predetermined fundus image, but the present invention is not limited thereto. For example, the target area may be selected by user input. Alternatively, the target area may be determined as a region showing an unusual aspect in each image.

38 is a flowchart illustrating a quality determination of an image according to an embodiment of the present invention. Referring to FIG. 38, determining the quality information of the fundus image (S15) may include detecting a target pixel (S15a) and determining a quality condition (S15b).

The method of determining the quality of an image may include detecting a target pixel 15a.

The target pixel may be a pixel corresponding to a defect or an artifact to be detected. The target pixel may be a pixel corresponding to a normal region where no defects or artifacts occur. The subject pixel may be a pixel corresponding to an obstructive element that interferes with the identification of a disease element to be detected. The target pixel may be a pixel corresponding to an indicator element that is an indicator for determining whether an abnormality has occurred in the fundus image.

Detecting the target pixel may include determining a target pixel among pixels included in the fundus image that satisfies the target pixel requirement. Detecting the target pixel may be performed in consideration of the pixel value described above.

The target pixel requirement may include a requirement for a reference range or reference value for the pixel value.

The reference range or reference value may be a predetermined range or value. Alternatively, the reference range or reference value may be a range or value obtained as a result of training the neural network model. The reference range or reference value may be adaptively changed to improve the accuracy of the determination as the quality of the image is judged a plurality of times.

For example, the image quality determination may include determining, as a target pixel, a pixel having a pixel value belonging to or not belonging to a predetermined reference range. Alternatively, the image quality determination may include determining a pixel having a pixel value that exceeds or falls below a predetermined reference value as the target pixel.

As a specific example, the target pixel requirement may include that the intensity value of the pixel is equal to or greater than the reference intensity, and detecting the target pixel may include determining a pixel having the intensity value equal to or greater than the reference intensity as the target pixel.

According to an embodiment, the target pixel requirement may be differently specified for each region. Detecting the target pixel from the image includes determining whether the first target pixel requirement is satisfied for the pixels included in the first region, and determining whether the second target pixel requirement is satisfied for the pixels included in the second region. can do.

Detecting the target pixel may include detecting the target pixel included in a predetermined area included in the fundus image. Detecting the target pixel may include determining a pixel included in a predetermined area and satisfying the target pixel requirement as the target pixel.

Alternatively, the target pixel requirement may include a requirement for the area. Detecting the target pixel may include determining a pixel located in a predetermined target area as a target pixel that satisfies the target requirement.

As a specific example, the target pixel requirement may be included in a target area that is an area corresponding to the fundus of the fundus image, and detecting the target pixel may include detecting a target pixel included in the target area corresponding to the fundus.

The fundus image may include at least one area. The predetermined area (or target area) may be determined in consideration of the position of the anatomical components of the eye. The at least one area may be determined in consideration of the location of occurrence of the element that interferes with the identification. The target area may be selected from at least one area included in the fundus image.

For example, it may include a plurality of regions divided by at least one boundary that anatomically significantly divides the fundus image. The boundary for dividing the fundus image may be determined based on the location of at least one of the macular, optic nerve papilla, blood vessel, and ocular embryo included in the fundus image.

The method of determining the quality of an image may include determining whether a quality condition is satisfied (S15b).

Determining whether the fundus image satisfies the quality condition may be performed by considering pixel values of at least one pixel included in the fundus image. Determining whether the fundus image satisfies the quality condition may be performed by considering a pixel value of at least one pixel included in the target area in the fundus image. The pixel value may be obtained from the original fundus image or the transformed fundus image. Alternatively, the quality condition of the fundus image may be determined based on other numerical values or data obtained, for example, histogram, frequency curve, etc. of the fundus image.

The pixel value may be any one of an intensity value, an intensity value, a brightness value, a luminance value, a color value, a chroma value, a saturation value, and a brightness value of the pixel. The pixel value may be in an integer or vector form. For example, the pixel value may be provided in the form of an 8-bit integer.

The pixel value may include at least one item value extracted from the fundus image having values for the plurality of items. For example, when the fundus image has a brightness value, a color value, and a saturation value for each pixel, at least one pixel value may include a brightness value.

The pixel value may include a plurality of values for the plurality of channels. For example, the fundus image may include an RGB channel, and the pixel value may include a color value for each of R, G, and B channels.

Determining whether the ocular fundus image satisfies a quality condition may include detecting a target pixel (or abnormal pixel) included in the image and satisfying a predetermined criterion. The target pixel may be a plurality.

Determining whether a fundus image satisfies a quality condition is based on the distribution of pixel values, deviations of pixel values, maximum or minimum pixel values, and differences between pixel values (hereinafter, referred to as "machining values of pixel values"). Can be performed.

Determining whether a fundus image satisfies a quality condition is that if the pixel value or processing value of the pixel value falls within the reference range, exceeds the reference value, or falls below the reference value, the fundus image satisfies the quality condition. It can be determined.

Determining whether the fundus image satisfies the quality condition may include determining whether the fundus image satisfies the quality condition in consideration of the detected target pixel.

Determining whether the fundus image satisfies a quality condition may include determining whether the number of target pixels is greater than or equal to a reference number. For example, if the number of target pixels included in the fundus image exceeds the reference number, the image may be determined to be dissatisfied with the quality condition.

According to an embodiment, the plurality of quality conditions may be determined in relation to the reference number. For example, when the number of target pixels exceeds the first reference number and falls short of the second reference number, the fundus image determines that the first quality condition is not satisfied, and when the number of target pixels exceeds the second reference number, the corresponding fundus The image may be determined to be dissatisfied with the second quality condition.

Determining whether the fundus image satisfies a quality condition may include determining whether a ratio of the target pixel is equal to or greater than a reference ratio. For example, when the ratio of the number of target pixels to the total number of pixels of the fundus image is greater than or equal to the reference ratio, the image may be determined to satisfy the quality condition.

In an embodiment, whether the quality condition of the fundus image is satisfied may be determined based on the binarized fundus image. For example, the fundus image is binarized with respect to the reference pixel value, and based on the binarized fundus image, the number or ratio of the target pixels having a pixel value that exceeds or falls short of the reference pixel value is obtained, and the obtained number is obtained. Alternatively, the ratio may be compared with the reference number or the reference ratio to determine whether the fundus image satisfies the quality condition.

With regard to determining whether the fundus image satisfies a quality condition, the subject area may be considered together.

As an example, when the target pixel is detected in the target region, the image may be determined to be dissatisfied with the quality condition. Alternatively, when the number of target pixels detected in the target area exceeds or falls short of the reference number, the image may be determined to be dissatisfied or satisfying the quality condition. When the ratio of the number of target pixels to the total number of pixels of the target area exceeds the reference ratio, the image may be determined to be dissatisfied or satisfying the quality condition.

As another example, when the target pixel is detected in an area other than the target area, the image may be determined to be dissatisfied or satisfying the quality condition.

As another example, when the target pixel is detected in the first region, the image is determined to be unsatisfactory in the first quality condition, and when the target pixel is detected in the second region, the image satisfies the second quality condition or You can judge that you are not satisfied.

As a specific example, the quality condition of the image may be determined in consideration of the target pixel included in the first target area corresponding to the area in which the fundus of the fundus image is distributed. Alternatively, the quality condition of the image may be determined in consideration of pixel values of pixels corresponding to regions in which the macular of the fundus is distributed.

According to an embodiment, the quality condition may be determined differently for each region. Determining whether the image satisfies the quality requirements means that the target pixel detected from the first region exceeds the first reference ratio with respect to the first region, and the target pixel detected from the second region has a second reference ratio with respect to the second region. It may include determining whether exceeds.

39 is a flowchart for describing a quality determination of an image according to an exemplary embodiment. Referring to FIG. 39, in the method of determining the quality of the fundus image by detecting artifacts related to the quality of the fundus image from the fundus image according to an embodiment of the present invention, detecting the first pixel related to the quality of the fundus image (S110), comparing the ratio of the first pixel of the entire pixels of the fundus image to the reference ratio (S130), and obtaining the quality information based on the comparison result (S150).

Detecting a first pixel related to the quality of the fundus image (S110) may compare a pixel value of at least one pixel included in a target area of the fundus image with a reference pixel value, thereby comparing the first pixel related to the quality of the fundus image. Detecting.

The target area is determined according to the artifacts detected for quality determination, and the artifact is either a bright artifact due to excessive reflection of light occurring in the fundus region of the fundus image or a dark artifact due to a shadow occurring in the fundus region of the fundus image. Can be.

The target area may include a first area spaced at least a first distance from the center of the area corresponding to the fundus of the fundus image, a second area within a first distance from the center of the area corresponding to the fundus, and an optic nerve papilla among the areas corresponding to the fundus. It may include at least one selected from the third region corresponding to the region in which is distributed and the fourth region corresponding to the region in which the macular is distributed among the regions corresponding to the fundus.

Comparing the ratio of the first pixel of all the pixels of the fundus image to the reference ratio (S130) may include determining whether the ratio of the first pixel of all the pixels of the fundus image exceeds the reference ratio. Alternatively, comparing the ratio of the first pixel of all the pixels of the fundus image to the reference ratio (S130) may include determining whether the ratio of the first pixel of all the pixels of the fundus image is less than the reference ratio. .

Acquiring quality information based on the comparison result (S150) may include obtaining quality information of the fundus image based on a comparison result obtained by comparing a ratio of the first pixel of all the pixels of the fundus image with a reference ratio. It may include doing.

For example, the step of obtaining quality information based on the comparison result (S150) may include: when the ratio of the first pixel of all the pixels of the fundus image exceeds the reference ratio, the quality indicating the fundus image is the abnormal fundus image. May include obtaining information. Alternatively, the obtaining of the quality information based on the comparison result (S150) may include: quality information indicating that the fundus image is a normal fundus image when the ratio of the first pixel of all the pixels of the fundus image is less than the reference ratio. It may include acquiring.

According to an embodiment, the method of determining a quality of a fundus image may further include obtaining a first scale value with respect to a first scale with respect to a plurality of pixels included in the fundus image from the fundus image. In this case, the pixel value obtained for the quality determination may include a first scale value. The first scale may be a brightness scale of the plurality of pixels, and the first scale value may be a brightness scale value.

According to an embodiment, the method of determining the quality of the fundus image may further include binarizing the fundus image with respect to a reference pixel value. In this case, the detecting of the first pixel may further include detecting the first pixel based on the binarized fundus image.

40 is a flowchart illustrating a method of determining an image quality according to an embodiment of the present invention. Referring to FIG. 40, a method of determining an image quality according to an embodiment of the present invention includes determining a region of a fundus image as a target region (S210) and detecting an artifact pixel from the target region (S230). can do.

 According to an embodiment of the present invention, a method of determining the quality of the fundus image by detecting artifacts related to the quality of the fundus image from the fundus image of the subject's fundus image is performed by detecting artifacts in at least a portion of the fundus image. Determining the target region (S210) and detecting at least one artifact pixel corresponding to the artifact from the target region (S230).

When the artifact is the first artifact, the target region may be determined as the first region that is a region spaced a predetermined distance from the center of the region corresponding to the fundus of the eye fundus image. The first artifact may be a bright artifact due to excessive reflection of light occurring in the fundus area of the fundus image. The first artifact may be various kinds of defects that occur in the outer area OA of the fundus image. The first artifact may be a first type artifact described later.

When the artifact is the second artifact, the target area may be determined as a fundus area that is an area in which the fundus is distributed in the fundus image. The second artifact may be a dark artifact due to a shadow occurring in the fundus region of the fundus image. The second artifact may be various kinds of defects occurring in the fundus region FA or the inner region IA of the fundus image. The second artifact may be a second type artifact, described below.

According to an embodiment, the method of determining the quality of the fundus image may further include determining the quality of the fundus image based on the detected artifact pixel.

Determining the quality of the fundus image may further include determining the quality of the fundus image in consideration of whether the amount of detected artifacts exceeds a reference amount. The amount of the artifact may be determined based on the number of pixels corresponding to the artifact detected from the fundus region, the area of the region corresponding to the artifact detected from the fundus region, or the number of artifacts detected from the fundus region. The reference amount may be defined as a reference pixel number, a reference width, a number of reference artifacts, or the like.

Determining the quality of the fundus image includes determining whether the number of detected artifact pixels exceeds a reference ratio with respect to the total number of pixels in the fundus image, and the reference ratio with respect to the total number of pixels of the fundus image. The eye fundus image may comprise determining to include an artifact if exceeded.

According to an embodiment of the present disclosure, the method for determining the quality of a fundus image may further include obtaining a converted fundus image by grayscale converting the fundus image. Detecting the artifact pixel may further include detecting the artifact pixel from the transformed fundus image. The method for determining the quality of the fundus image may include obtaining a transformed fundus image obtained by binarizing the fundus image, changing the color tone, or applying other modifications.

41 is a flowchart illustrating a method of determining an image quality according to an embodiment of the present invention. Referring to FIG. 41, in the method of determining an image quality according to an embodiment of the present invention, determining a region of a fundus image as a target region (S210), wherein a pixel value of a pixel included in the target region is a reference pixel value range. Comparing with (S231) and determining a pixel having a pixel value included in the reference pixel value range as an artifact pixel (S233).

Detecting an artifact pixel from the target region (S230) may include comparing pixel values of a plurality of pixels included in the target region with a reference pixel value range (S231) and / or included in the target region and included in the reference pixel value range. In operation S233, at least one pixel having the pixel value may be determined as an artifact pixel. The pixel value may be a brightness value of the pixel.

The reference pixel value range may be determined as a range including a value larger than the reference pixel value or a range including a value smaller than the reference pixel value. Alternatively, the reference pixel value range may be defined as a range including a value larger than the reference lower limit value and smaller than the reference upper limit value.

42 is a flowchart illustrating a method of determining an image quality according to an embodiment of the present invention. Referring to FIG. 42, in the method of determining an image quality according to an embodiment of the present disclosure, determining a region of a fundus image as a target region (S210) and determining a pixel value of a pixel included in the target region as a first reference. Comparing with the pixel value (S235) and determining a pixel included in the target area and having a pixel value larger than the first reference pixel value as the first artifact pixel (S237).

When the artifact is the first artifact, the detecting of the artifact pixel (S230) may include comparing pixel values of a plurality of pixels included in the target region with a first reference pixel value (S235) and being included in the target region. In operation S237, at least one pixel having a pixel value greater than the first reference pixel value may be determined as the first artifact pixel.

For example, when the artifact is a bright artifact, the detecting of the artifact pixel (S230) may include comparing a brightness value of pixels included in the target area with a bright artifact reference pixel value and having a value brighter than the bright artifact reference pixel value. May be determined to be a bright artifact pixel.

Meanwhile, when the artifact is the second artifact, the detecting of the artifact pixel (S230) may include comparing pixel values of a plurality of pixels included in the target region with a second reference pixel value and including the second reference pixel value in the target region. The method may further include determining at least one pixel having a pixel value smaller than the reference pixel value as the second artifact pixel.

For example, when the artifact is a dark artifact, the detecting of the artifact pixel (S230) may include comparing the brightness value of the pixels included in the target area with a dark artifact reference pixel value and having a brightness value smaller than the dark artifact pixel reference value. Determining at least one pixel as a dark artifact pixel.

In the above description, for the sake of convenience, the quality of the image has been described as a reference, but this is only an example, which is a substitute for various operations having the purpose of selecting an image that is inappropriate to achieve the purpose throughout. It should be interpreted as possible.

2.3.3 Algorithm by Type of Detection Object

According to one embodiment of the present invention, determining the quality of an image may include detecting various defects or abnormalities occurring in the image. The above-described image determination method may be implemented differently depending on the type of defect to be determined.

Hereinafter, some embodiments will be described with reference to an image quality determination algorithm according to the above type for detecting to determine the quality of an image.

2.3.3.1 Type 1 Artifact (Bright Artifact)

According to an embodiment of the present invention, an image quality determining method for determining whether the fundus image includes a first type artifact (hereinafter, referred to as a first artifact) may be provided. Image quality determination may be performed in consideration of the degree to which the area corresponding to the first artifact is distributed with respect to the fundus image.

The first artifact may be a bright artifact. Bright artifacts can be caused by optical reflections, bright borders, and the like. Bright artifacts may be distributed close to the outside of the fundus image. Hereinafter, an image quality determination method will be described on the basis of a case where the first artifact is a bright artifact that occurs close to the outside of the fundus region of the fundus image due to an optical factor.

43 is a view for explaining a method of determining image quality according to an embodiment of the present invention.

Referring to FIG. 43A, an image quality determining method according to an embodiment of the present invention may include acquiring a fundus image (S21) and acquiring first quality information (S23). Referring to (b) of FIG. 43, in the image quality determining method according to an embodiment of the present invention, acquiring a fundus image (S21), detecting a first pixel (S23a), and determining a first quality condition It may include the step (S23b).

According to an embodiment of the present invention, an image quality determining method for determining whether a fundus image includes a first artifact comprises: acquiring a fundus image and first type quality information (hereinafter, first quality information) of the fundus image; ) May be obtained.

The image quality determining method for determining whether the fundus image includes the first artifact may further include converting the fundus image to facilitate acquisition of an intensity value of each pixel of the fundus image. Regarding image conversion, the above-described information may be similarly applied in 2.2 Determination.

Acquiring the first quality information may include detecting a first pixel from the fundus image and determining a first quality condition.

The first pixel may be a pixel corresponding to the first artifact. The first pixel may be a pixel corresponding to a region in which the first artifact is distributed. The first pixel may be a pixel used as an indicator for detecting the first artifact. As a specific example, the first pixel may be a bright pixel for detecting bright artifacts. For example, a bright pixel may be a pixel having a particularly bright brightness value or having a specific color value.

The detecting of the first pixel may include detecting a pixel having a pixel value greater than or equal to a reference value from the fundus image. For example, detecting the first pixel may include detecting a first pixel having a brightness value of the pixel having a reference brightness value or more.

The reference value may be a predetermined value. Alternatively, the reference value can be determined differently for the individual image. For example, the reference value may be determined as an average value of pixel values of all pixels included in the image. Alternatively, the reference value may be determined as an average value of pixel values of pixels included in the fundus region.

The first pixel can be detected from a specific area. For example, the first pixel can be detected from the fundus area. Alternatively, the first pixel may be detected from a target area determined as part of the fundus area. More specifically, the first pixel may be detected from a target area that is an area spaced apart from the center of the fundus area by a predetermined distance or more. For example, the first pixel may be detected from a target area that is an area outside of the boundary determined to include a predetermined portion of the fundus area from the center of the fundus area.

In consideration of the detected first pixel, it may be determined whether the fundus image satisfies the first quality condition. Alternatively, whether the fundus image includes the first artifact may be determined in consideration of the detected first pixel.

The determining of the first quality condition may include determining the first quality condition in consideration of the detected first pixel. The first quality condition may include a range of the first number of pixels, a ratio of the number of first pixels to the number of pixels of the fundus image, and the like.

Whether the fundus image satisfies the first quality condition may be determined in consideration of the first pixel included in the target area. In this case, the determining of the first quality condition may include determining whether the fundus image satisfies the first quality condition in consideration of a ratio of the first pixel detected from the target region in the fundus image.

For example, determining the first quality condition may include determining whether the number of first pixels exceeds a reference ratio of the number of pixels of the entire fundus image (or fundus area).

As another example, determining the first quality condition may include determining whether a ratio of the first pixel detected from the outer region OA in the fundus image (or fundus region) exceeds the reference ratio.

More specifically, when the number of first pixels having a brightness value greater than or equal to the reference brightness detected from the outer area OA exceeds 2% of the total number of pixels of the fundus image, it may be determined that the first quality condition is satisfied. Alternatively, it may be determined that the first artifact has occurred in the fundus image.

The reference ratio may be a predetermined value. Alternatively, the image may be determined for each image in consideration of variations in brightness values of the entire image. Or it may be a value obtained as a result of artificial intelligence learning. The reference ratio may be 5% or less.

44 is a diagram for describing a method of determining image quality according to an embodiment of the present invention.

Referring to (a) to (c) of Figure 44, the image quality determination method according to an embodiment of the present invention obtains a fundus image (a), converts the acquired fundus image (b), the transformed fundus (C) determining whether the image contains a bright artifact based on the image.

Referring to FIG. 44A, the acquired fundus image may include an artifact near an outer portion of the fundus region. For example, the fundus image may comprise bright artifacts.

Referring to FIG. 44B, converting the fundus image may include converting the fundus image to grayscale. Converting the fundus image may include histogram normalization of the fundus image.

Referring to FIG. 44C, determining whether the fundus image includes a bright artifact may include detecting a bright artifact pixel. Detecting bright artifact pixels may include the number of pixels, the position of the pixels, the ratio of pixels, and the like.

Detecting bright artifact pixels involves detecting bright artifact pixels (i.e., pixels having a brightness value greater than or equal to a reference brightness) whose values are greater than or equal to a threshold from an outer region of a boundary spaced a predetermined distance from the center of the fundus region. It may include. The boundary may be determined to partition the fundus region such that the region therein is 80% of the fundus region.

Determining whether the image contains bright artifacts (c) detects the bright artifact pixels, determines whether the proportion of the detected bright artifact pixels to the fundus image is greater than or equal to the reference ratio (e.g., greater than 2%). When the ratio of the bright artifact pixel to the fundus image is greater than or equal to the reference ratio, the fundus image may be determined to have a bright artifact.

The determining of the first quality condition may further include determining that the fundus image includes the first artifact when the first quality condition is satisfied. For example, if the fundus image satisfies the first quality condition, it may be determined that the fundus image includes bright artifacts.

If it is determined whether the quality condition is satisfied, the first quality information may be obtained.

The first quality information may indicate suitability of the fundus image. In particular, the first quality information may indicate suitability associated with the bright artifact of the fundus image. The first quality information may indicate suitability associated with the area where bright artifacts of the fundus image occur.

The first quality information may indicate whether a bright artifact is included in the fundus image. For example, when the fundus image satisfies the first quality condition, first quality information indicating that the fundus image includes a bright artifact may be obtained.

The first quality information may include bright artifact presence information on whether the fundus image includes bright artifacts, degree information of bright artifacts, location information of bright artifacts, and the like.

According to an embodiment of the present invention, a method of determining a quality of a fundus image by detecting a first type artifact from a fundus image of a subject's fundus image may include pixel values of at least one pixel included in a target area of the fundus image. Comparing the reference pixel value with a reference pixel value to detect a first pixel related to the quality of the fundus image, comparing the proportion of the first pixel of all the pixels of the fundus image to the reference ratio and the first of all pixels of the fundus image Obtaining quality information of the fundus image based on a comparison result obtained by comparing the ratio occupied by the pixel with a reference ratio, wherein the artifact is a first artifact, and the target region is an area of a region corresponding to the fundus of the fundus image. The first region may be spaced apart from the center by a first distance or more.

In this case, the first distance may be determined such that an area within a first distance from the center of the area corresponding to the fundus is equal to or greater than a first ratio with respect to the entire fundus image area.

The reference pixel value is a first pixel value, and the detecting of the first pixel in comparison with the reference pixel value may include detecting a first pixel included in the target area and having a pixel value larger than the reference pixel value.

The reference ratio is the first reference ratio, and the comparing with the reference ratio may further include determining whether a ratio of the first pixel of all the pixels of the fundus image is greater than the first reference ratio.

The acquiring of the quality information based on the comparison result may further include determining that the fundus image is a bad image when the ratio of the first pixel of the total pixels of the fundus image is greater than the first reference ratio.

2.3.3.2 Type 2 artifacts (dark artifacts)

According to an embodiment of the present invention, an image quality determination method for determining whether the fundus image includes a second type artifact (hereinafter, referred to as a second artifact) may be provided. Image quality determination may be performed in consideration of the degree to which the area corresponding to the second artifact is distributed with respect to the fundus image.

The second artifact may be a dark artifact. The dark artifact may occur when the distance between the fundus and the image pickup device or the amount of light incident on the fundus is picked up while the eye fundus image is captured. Dark artifacts can occur in any area. Dark artifacts can occur close to the center of the fundus region. Hereinafter, an image quality determination method will be described based on the case where the second artifact is a dark artifact caused by an optical factor.

45 is a diagram for describing a method of determining image quality according to an embodiment of the present invention. Referring to FIG. 45A, an image quality determining method according to an embodiment of the present invention may include acquiring a fundus image (S31) and acquiring second quality information (S33). Referring to FIG. 45B, in the image quality determining method according to an embodiment of the present disclosure, the method may include obtaining a fundus image (S31), detecting a second pixel (S33a), and determining a second quality condition. It may include the step (S33b).

According to an embodiment of the present invention, an image quality determining method for determining whether a fundus image includes a second artifact includes obtaining a fundus image and second type quality information of the fundus image (hereinafter, referred to as second quality information). ) May be obtained.

According to an embodiment of the present disclosure, the image quality determining method may further include an image converting step. The above description may be similarly applied to the conversion of the image in the 2.2 determining method.

Dark artifacts may appear due to shadows, etc., generated during imaging of the fundus image. Dark artifacts may be distributed close to the center of the fundus image. Dark artifacts and bright artifacts may occur together in the same image.

Acquiring the second quality information may include detecting a second pixel from the fundus image and determining a second quality condition.

The second pixel may be a pixel corresponding to the second artifact. The second pixel may be a pixel corresponding to a region in which the second artifact is distributed. The second pixel may be a pixel used as an indicator for detecting the second artifact. As a specific example, the second pixel may be a dark pixel for detecting dark artifacts. For example, the dark pixel may be a pixel having a particularly dark pixel value or having a specific color value.

The detecting of the second pixel may include detecting a pixel having a pixel value less than or equal to a reference value from the fundus image. For example, detecting the second pixel may include detecting a second pixel whose brightness value is equal to or less than a reference brightness value.

The second pixel can be detected from a particular area. The second pixel may be detected from a fundus area or a target area that is part of the fundus area. For example, the second pixel may be detected from a target area that is an area within a predetermined distance from the center of the fundus area. The second pixel may be detected from the inner region IA inside the boundary BO described in FIG. 32.

In consideration of the detected second pixel, it may be determined whether the fundus image satisfies the second quality condition. Alternatively, whether the fundus image includes the second artifact may be determined in consideration of the detected second pixel.

The determining of the second quality condition may include determining the second quality condition in consideration of the detected second pixel. The second quality condition may include a range of the second pixel number, a ratio of the second pixel number to the pixel number of the fundus image, and the like.

Whether the fundus image satisfies the second quality condition may be determined in consideration of the second pixel included in the target area. In this case, the determining of the second quality condition may include determining whether the fundus image satisfies the second quality condition in consideration of the proportion of the second pixel detected from the target region in the fundus image.

For example, determining the second quality condition may include determining whether the number of detected second pixels is less than a reference ratio with respect to the total number of pixels of the fundus image (or fundus area).

As another example, determining the second quality condition may include determining whether a ratio of the second pixel detected from the inner region IA or the fundus region FA of the fundus region to the fundus image (or the fundus region) exceeds the reference ratio. It may include doing.

More specifically, when the number of second pixels having a brightness value below the reference brightness detected from the inner area OA or the fundus area FA exceeds 2% of the total number of pixels of the fundus image, the second quality condition is satisfied. It can be judged. Alternatively, it may be determined that the second artifact has occurred in the fundus image.

46 is a view for explaining a method of determining image quality according to an embodiment of the present invention.

46 (a) to (c), the image quality determining method according to an embodiment of the present invention obtains a fundus image (a), converts the acquired fundus image (b), and converts the transformed fundus (C) determining whether the image contains a bright artifact based on the image.

Referring to FIG. 46A, the acquired fundus image may include an area photographed darker than the surrounding. The acquired fundus image may include dark artifacts.

Referring to FIG. 46B, transforming the fundus image may include grayscale transformation or histogram normalization of the fundus image.

Referring to FIG. 46C, determining (c) whether the image includes an artifact may include detecting a dark artifact pixel. Referring to FIG. 46C, detecting the dark artifact pixel may include detecting a pixel whose value is less than or equal to a threshold (that is, a pixel having a brightness value less than or equal to a reference brightness).

For example, detecting the dark artifact pixel may include detecting a pixel whose value is equal to or less than a threshold value from an inner region within a predetermined distance from the center of the fundus region. In this case, the boundary for determining the inner region OA of the fundus region may be determined such that the inner region occupies 80% of the fundus region.

Referring to FIG. 46, determining whether the image includes dark artifacts may include detecting a dark artifact pixel, determining whether the detected dark artifact pixel occupies the fundus image is greater than or equal to a reference ratio (2% or more). When the ratio of the dark artifact pixel to the fundus image is greater than or equal to the reference ratio, the fundus image may be determined to have a bright artifact.

The determining of the second quality condition may further include determining that the fundus image includes the second artifact when the second quality condition is satisfied. For example, when the fundus image meets the second quality condition, it may be determined that the fundus image includes dark artifacts.

If it is determined whether the quality condition is satisfied, the second quality information may be obtained.

The second quality information may indicate suitability of the fundus image. In particular, the second quality information may indicate suitability associated with the dark artifact of the fundus image. The second quality information may indicate suitability associated with the area in which dark artifacts of the fundus image occur.

The second quality information may indicate whether dark artifacts are included in the fundus image. For example, when the fundus image satisfies the second quality condition, second quality information indicating that the fundus image includes the bright artifact may be obtained.

The second quality information may include dark artifact presence information related to whether the fundus image includes dark artifacts, degree information of dark artifacts, location information of dark artifacts, and the like.

According to an embodiment of the present invention, a method of determining a quality of a fundus image by detecting a second type artifact from a fundus image of a subject's fundus image may include pixel values of at least one pixel included in a target area of the fundus image. Comparing the reference pixel value with a reference pixel value to detect a first pixel related to the quality of the fundus image, comparing the proportion of the first pixel of all the pixels of the fundus image to the reference ratio and the first of all pixels of the fundus image Obtaining quality information of the fundus image based on a comparison result obtained by comparing the ratio occupied by the pixel with a reference ratio, wherein the artifact is a first artifact, and the target region is an area of a region corresponding to the fundus of the fundus image. The first region may be spaced apart from the center by a first distance or more.

In this case, the artifact may be a second artifact, and the target region may be a second region in which the fundus of the fundus image is distributed. The reference pixel value is a second pixel value, and the detecting of the first pixel in comparison with the reference pixel value may include detecting a first pixel included in the target area and having a pixel value smaller than the reference pixel value.

The reference ratio is the second reference ratio, and the comparing with the reference ratio may further include determining whether a ratio of the first pixel of all the pixels of the fundus image is smaller than the second reference ratio.

The acquiring of the quality information based on the comparison result may further include determining that the fundus image is a defective image when the ratio of the first pixel of the total pixels of the fundus image is smaller than the second reference ratio.

2.3.3.3 Third Type Artifact

According to an embodiment of the present invention, an image quality determination method for determining whether the fundus image includes a third type artifact (hereinafter referred to as a third artifact) may be provided. Image quality determination may be performed in consideration of the degree of the third artifact included in the fundus image.

The third artifact may be an artifact in which the entire fundus region is blurred. For example, there may be a case in which the fundus region is clouded due to contamination on the lens or other obstacle on the optical path during imaging.

For example, the third artifact may be a disease-derived artifact due to eye diseases of the subject. If the subject has a certain type of eye disease, there may be cases where the fundus image cannot be used to diagnose other diseases, even if no artifacts from other external factors have occurred. For example, when opacity occurs in the eye of a subject, since the fundus image is mostly blurred, there is a problem that reliability is not guaranteed when diagnosing other diseases based on the fundus image. Therefore, it is necessary to separately detect such an artifact and notify the user.

In the following, an artifact that makes element identification difficult throughout the fundus area is defined as a third artifact, and a method for determining the quality of the fundus image in consideration of the third artifact will be described.

According to an embodiment of the present invention, an image quality determining method for determining whether a fundus image includes a third artifact comprises: acquiring a fundus image and third type quality information (hereinafter, third quality information) of the fundus image; ) May be obtained.

The image quality determining method may further include converting the fundus image, and the foregoing description may be similarly applied in the 2.2 determining method.

Acquiring the third quality information may include detecting a third pixel from the fundus image and determining a third quality condition.

The third pixel may be a pixel corresponding to the third artifact. The third pixel may be a pixel corresponding to a region in which the third artifact is distributed. The third pixel may be used as an indicator for detecting the third artifact. The third pixel may be a cloudy pixel. The third pixel may be a pixel having a pixel value within a predetermined range or having a pixel value difference within a predetermined ratio when compared with a neighboring pixel.

In consideration of the detected third pixel, it may be determined whether the fundus image satisfies the third quality condition. Alternatively, whether the fundus image includes the third artifact may be determined in consideration of the detected third pixel.

The determining of the third quality condition may further include determining the third quality condition in consideration of the number, range, and ratio of the third pixels included in the target area (eg, the fundus area).

The determining of the third quality condition may further include determining that the fundus image includes the third artifact when the third quality condition is satisfied.

 Acquiring the third quality information may include detecting the third artifact from the fundus image. Whether the fundus image includes the third artifact may be determined using the fundus region of the fundus image as the target region.

Detecting the third artifact may further include detecting the third artifact using a histogram of the fundus image. For example, the determination of the presence or absence of the third artifact may use the deviation of the histogram of the fundus image, and when the deviation is less than or equal to a predetermined value, the corresponding image may be determined to include the third artifact. In other words, determining whether the third artifact is present may include determining that the fundus image has the third artifact when the brightness distribution of the pixels included in the fundus region of the fundus image is equal to or greater than a reference value.

Alternatively, according to an embodiment of the present invention, whether the fundus image includes the third artifact may be determined by using a neural network model trained using the fundus image data set having the third artifact.

When it is determined whether the third quality condition is satisfied or the third artifact is detected from the image, the third quality information may be obtained. The third quality information may indicate suitability of the fundus image, eg, suitability associated with turbidity.

The third quality information may include third artifact information indicating whether a third artifact has occurred in the fundus image and / or a degree of the third artifact. The third quality information may include turbid artifact information indicating the presence or absence of turbid artifacts, the degree of turbid artifacts, and the like.

2.3.4 Multiple fundus images

According to an embodiment of the present invention, the imaging device (or client device) may acquire a plurality of fundus images. The plurality of fundus images may be a plurality of fundus images captured by the imaging device. The plurality of fundus images may be images continuously photographed by the imaging device. Alternatively, the plurality of fundus images may be a plurality of fundus images obtained from the fundus image obtained by the imaging device. For example, the plurality of fundus images included in the fundus image may be a plurality of fundus images extracted from the fundus image and corresponding to the plurality of frames of the fundus image.

47 is a diagram for describing a plurality of fundus images according to an embodiment of the present invention.

Referring to FIG. 47, according to an embodiment of the present disclosure, a plurality of fundus images a may be obtained. The plurality of fundus images a may be obtained in a time t order. The plurality of fundus images a may be images continuously photographed. The plurality of fundus images (a) may be images included in the fundus image.

The plurality of fundus images a may include a first fundus image b1, a second fundus image b2, and a third fundus image b3. Hereinafter, some embodiments of a method of determining a target fundus image from a plurality of fundus images will be described with reference to FIG. 46.

Determining the quality of the fundus image may include acquiring a plurality of fundus images and determining the quality of the plurality of fundus images according to the above-described quality determination method of the image. In particular, when the diagnosis target fundus image is determined in the diagnosis step, one fundus image of the plurality of fundus images may be determined as the target fundus image based on a quality determination result for each of the plurality of fundus images.

According to an embodiment of the present disclosure, an image selected from a plurality of fundus images may be determined as a diagnosis target fundus image.

As a specific example, a fundus image having a better quality among the plurality of fundus images may be determined as the target fundus image. For example, determining the quality of the fundus image may include performing, by the controller, quality determination on the first fundus image b1 and the second fundus image b2. If the first fundus image b1 has better quality information than the second fundus image b2, the controller may determine the first fundus image b1 as the target fundus image.

As another specific example, when the fundus images are sequentially obtained, the target fundus images may be replaced. For example, when the fundus image is acquired through continuous shooting or video shooting, the control unit may determine that the first fundus image b1 having a better quality among the first fundus image b1 and the second fundus image b2 is an object. In the state determined as the fundus image, the quality of the third fundus image b3 acquired later than the first fundus image b1 and the second fundus image b2 is determined, and the third fundus image b3 is determined as the first fundus. If the image has a better quality than the image b1, the target fundus image may be replaced with the third fundus image b3.

When the target fundus image is replaced, the information obtained based on the target fundus image may be replaced. For example, in a diagnosis apparatus that continuously acquires an ocular fundus image and simultaneously provides diagnostic assistance information, when the target fundus image is replaced by continuous acquisition of the fundus image, the provided diagnosis assistance information may be replaced.

As another specific example, when a fundus image is obtained that does not satisfy the quality criteria, the fundus image may be discarded. In determining the quality of the fundus image, the controller performs quality determination on the second fundus image b2 obtained after the first fundus image b1 and the first fundus image b1 are acquired, and the second If the fundus image b2 does not meet the quality criteria, it may include discarding the second fundus image b2. Alternatively, the controller may perform a quality determination on the first fundus image b1 and the second fundus image b2, and the second fundus image b2 may have a higher quality value or quality score than the first fundus image b1. If low, the second fundus image b2 may be discarded.

As another specific example, when the fundus image satisfying the quality criteria is obtained, the acquisition of the fundus image may be stopped. Determining a quality criterion of the fundus image includes the controller performing a quality determination on the first fundus image b1 and the third fundus image b3 obtained after the first fundus image b1 is obtained. can do. If the third fundus image b3 satisfies a quality criterion (eg, does not have an artifact), the acquisition of the fundus image may be stopped. For example, the controller may instruct the imaging device to stop acquiring the fundus image. The controller may determine an image satisfying the quality criteria as the target fundus image. In other words, when an image satisfying the quality criteria is obtained, the controller may stop acquiring the fundus image and determine an image satisfying the quality criteria as the target fundus image. When the target fundus image is determined, diagnostic assistance information or the like may be output based on the target fundus image. In addition, for example, if an image satisfying the quality criteria is acquired, the controller may start acquiring the image. In other words, the controller judges the quality of the fundus image sequentially photographed, but discards the fundus image that does not satisfy the quality standard, and acquires (or stores) the corresponding fundus image when the fundus image that satisfies the quality standard is captured. You can start

According to another embodiment of the present invention, one fundus image generated by merging or combining at least one fundus image selected from a plurality of fundus images may be determined as a diagnosis fundus image.

As a specific example, the controller may extract different regions of the fundus image from the plurality of fundus images, and determine the image generated by merging the extracted partial images as the target fundus image. For example, determining the quality of the fundus image may include determining, by the controller, the quality of the first fundus image b1 and the second fundus image b2. As a result of the quality determination, the fourth fundus image generated by merging the first region of the first fundus image b1 and the second region of the second fundus image b2 may be determined as the target fundus image. As a result of the quality determination, the fourth fundus image generated by merging the first region of the first fundus image b1 and the second region of the second fundus image b2 determined to satisfy the quality criteria may be determined as the target fundus image. Can be.

As another specific example, the control unit performs quality determination for each region on a plurality of fundus images, extracts a region satisfying the quality criteria from the fundus image satisfying the quality criteria for each region, and merges the extracted partial images. Can be determined as the target fundus image. For example, the determining of the quality of the fundus image may include determining, by the controller, the quality of the first region and the second region included in the first fundus image b1, the first region included in the second fundus image b2, and The fourth image generated by determining the quality of the second region and merging the first region of the first fundus image b1 and the second region of the second fundus image b2 that satisfies the quality criterion is the target fundus image. You can decide.

As another specific example, a target fundus image may be generated by overlapping a plurality of fundus images. For example, determining the quality of the fundus image may include acquiring, by the controller, the first fundus image b1 and detecting a first region in which the artifact is generated from the first fundus image. The controller further acquires the second fundus image b2, determines whether an artifact has occurred in the first region of the second fundus image b2, and the artifact does not occur in the first region of the second fundus image b2. In this case, a partial image of the first area of the second fundus image b2 may be obtained. The controller may determine the fourth fundus image generated by overlapping or merging the obtained partial image with the first fundus image b1 as the diagnosis target image.

Merging or synthesizing the at least one fundus image may be performed using a neural network model. For example, an image formed by synthesizing or merging two or more images selected from a plurality of fundus images using a convolutional neural network (CNN) or a generative adversarial neural network (GAN), etc. Can be determined by an image.

48 is a diagram for describing selecting a target image according to an embodiment of the present invention. Referring to FIG. 48, when the first fundus image (a) and the second fundus image (b) having dark artifacts are obtained, the second fundus image (b) including the weaker dark artifacts is selected as the target fundus image. Can be. More specifically, when the number or ratio of pixels corresponding to the dark artifacts included in the first fundus image (a) is greater than the number or ratio of pixels corresponding to the dark artifacts included in the second fundus image (b), 2 The fundus image (b) may be selected as the target fundus image.

As another example, unlike illustrated in relation to FIG. 48, even when a plurality of images including different types of artifacts are obtained, the artifacts including the weaker artifacts may be selected as the target fundus image. For example, the first fundus image includes the first artifact and the second fundus image includes the second artifact, and the second fundus image is included in the second fundus image than the ratio of pixels corresponding to the first artifact included in the first fundus image. When the ratio of the pixels corresponding to the two artifacts is low, the second fundus image may be determined as the target fundus image.

49 is a diagram for describing generation of a target image according to an embodiment of the present invention. Referring to FIG. 49, when the first fundus image (a) including the bright artifact and the second fundus image (b) including the dark artifact are obtained, an area where the bright artifact of the first fundus image (a) is located The third fundus image c generated by merging (or overlapping) a partial region of the second fundus image b corresponding to the first fundus image a may be determined as the target fundus image.

As another example, unlike the example illustrated in relation to FIG. 49, the third fundus image c may remove a portion of the first fundus image a corresponding to an area where the dark artifact of the second fundus image b is located. It may be generated by merging or superimposing the two fundus images (b).

2.3.5 Use of Results

2.3.5.1 Labeling

According to the above-described quality determination method, quality information of the fundus image may be obtained. The quality information of the fundus image may be used in various cases.

In the database construction step, the quality information of the fundus image may be used for constructing the granular data set. In the learning phase of the diagnostic model, the quality information of the fundus image may be used to classify and use the training data for each learning purpose. In the diagnosis step, the quality information of the fundus image may be used to select a diagnosis target fundus image.

According to an embodiment of the present invention, when the quality information of the fundus image is obtained, the controller may add the quality information to the fundus image. For example, the controller may tag or label quality information corresponding to each fundus image. For example, the controller may label the corresponding information on an image determined as having a specific artifact as a result of the quality determination, as a result of being judged to be less than the reference quality, or having a defect.

For example, the controller may label the first label on the fundus image determined to include the first artifact. As another example, the controller may label the first label on the fundus image determined to not satisfy the first quality criterion.

As another example, the controller may label the second label with respect to the fundus image determined to not include the first artifact or to satisfy the first quality criterion.

As a result of the quality determination of the plurality of images, a data set including the quality information may be obtained. A plurality of data sets may be obtained according to the quality classification. For example, image data judged to have an artifact or an image data set (hereinafter referred to as a first data set) judged to be less than the reference quality and an image data set judged to have no artifact or satisfying the reference quality (hereinafter , The second data set) may be managed separately.

As an example, the first data set may be discarded. For example, when training a diagnostic model to perform a diagnosis based on the fundus image, the data may be removed from the training data because the model may be less accurate when using the first data set containing artifacts or composed of bad images. Can be.

As another example, the diagnostic model may be trained using the first data set. In this case, the diagnostic model may exhibit more robust characteristics for images with artifacts or of poor quality.

Regarding the training of the diagnostic model using the data set obtained as a result of the quality judgment and the performance of the diagnosis, the application of the quality judgment result will be described in detail.

2.3.5.2 Print

According to an embodiment of the present invention, the obtained quality information may be output. A processing device, an image pickup device, a client device, or a mobile device that performs quality determination may output quality information. The quality information may include visual, auditory and / or tactile information. The quality information may be provided to the user through the above-described user interface.

As an example, when it is determined that the fundus image is a normal image (or when it is determined that it does not include an artifact) as a result of the quality determination of the photographed fundus image, diagnostic assistance information based on the corresponding fundus image may be output.

When the quality judgment of the captured fundus image determines that the fundus image is an abnormal image (or when it is determined to include an artifact), the quality information of the fundus image is recognized so that the user can recognize the quality information of the image. May be output.

In this case, the output quality information may include the type, location, and degree of artifacts included in the fundus image. Diagnostic assistance information based on the fundus image may be provided along with the quality information. In addition to the quality information, a guide recommending a retake may be output. The reshooting guide may be output along with the indication information.

According to an embodiment of the present invention, the indication information determined in consideration of the result of the quality determination of the image may be output.

The indication information may include instructions for improving the quality of the fundus image. The instruction information may include instructions for causing the eyeball component to be included on the screen. The indication information may include instructions that cause the eye component to align in the reference direction. The indication information may include instructions for causing the ROI to be included in the imaging area. The indication information may include instructions for allowing the region of interest to be included in an appropriate position of the captured image, instructions for not including an artifact in the region of interest, or instructions for improving the quality of the region of interest. For example, the instruction information may include instructions regarding focus, shutter speed, aperture control, sensitivity control, artificial light control, and the like.

As an example, the instruction information may be provided in the form of a processing instruction that can be interpreted by the imaging apparatus. In this case, when the instruction information provided from the processing apparatus to properly image the fundus image is transferred to the imaging apparatus, the imaging apparatus can change the imaging condition based on the instruction information. Or, when the imaging device performs quality determination (for example, in the case of a portable fundus imaging device or a smartphone-mounted imaging device), the control unit of the imaging device generates instruction information based on the quality determination, The imaging condition can be changed based on this. For example, the amount of light, focus, shooting direction, and the like can be changed.

The indication information may be output as visual, auditory and / or tactile information. The instruction information may be output from the diagnostic apparatus or the imaging apparatus. For example, when quality information indicating that bright artifacts are generated in the vicinity of the standing side of the fundus is acquired, instruction information indicating to move the photographing lens in the direction of the standing side may be output. The visual indication information may be superimposed on the fundus image displayed on the display unit.

The indication information may be provided differently when the subject and the examiner exist and when the subject photographs the fundus by themselves. For example, the indication information may indicate the opposite direction when the subject and the inspector are present and when the subject photographs the fundus. In other words, when there is a subject and an inspector, the instruction information based on the position of the inspector may be output, and when the subject photographs himself, the instruction information based on the position of the subject may be output.

In particular, when the subject photographs herself, the indication information may be provided as tactile or auditory information. When the subject photographs herself, the visual indication information may be output so as to be displayed in the field of view of the eye to be picked up. For example, the controller may reduce the amount of light incident on the fundus during imaging and output instruction information in the direction of the eye to be imaged.

The instruction information can instruct a setting condition change of the imaging device. Alternatively, the indication information may include a notification for requesting approval for changing a setting condition of the imaging apparatus. Notifications requesting approval may be provided visually or audibly. The imaging apparatus or the diagnostic apparatus may obtain a user approval generated in response to the notification requesting the approval, and change the setting condition.

According to an embodiment of the present invention, quality information may be output even when photographing a fundus image or a continuous image. The processing device, the image pickup device, the client device, or the mobile device may output the quality information obtained from the fundus image. In this case, the indication information generated in consideration of the quality information may be output.

As a specific example, the fundus image capturing apparatus may configure a fundus image by continuously photographing the fundus image. In this case, the processing apparatus for determining the quality of the fundus image may acquire quality information for each acquired fundus image frame. The processing apparatus may output the instruction information indicating the end of photographing when the fundus image of appropriate quality is obtained based on the obtained quality information. Alternatively, the processing apparatus may output the indication information indicating that an appropriate fundus image can be obtained based on the obtained quality information.

The indication information may be provided together during the capturing of the fundus image. In other words, the indication information may be provided in real time. The controller outputs first indication information obtained as a result of the first quality determination on the first fundus image corresponding to the first frame before a specific time point, and includes a first frame corresponding to the second frame obtained in response to the output of the first indication information. The second quality determination may be performed on the two fundus images, and the second indication information may be output.

As a specific example, when the dark artifact is detected near the macula, the controller increases the amount of light, adjusts the distance to the fundus, or the direction of incidence of the light, as a result of performing a first quality determination on the first fundus image corresponding to the first frame. You can output the command information.

2.4 Disease specific quality judgment method (compatibility judgment method)

2.4.1 Disease specific characteristics

When diagnosing a disease using an ocular fundus image, a region of interest may be determined differently according to the disease. In other words, in the case of diagnosing a disease using the fundus image, there may be a case in which the disease is diagnosed based on the entire area of the fundus or a case of diagnosing the disease based on information obtained from a part of the fundus image. .

In this regard, criteria for determining the quality of an image according to a target disease may be set differently. In other words, quality criteria for determining the suitability of an image may vary according to a target disease. By performing image quality determination differently for each target disease, diagnosis accuracy for each disease may be improved and image use efficiency may be increased.

For example, the area to be considered when determining the quality of the fundus image may be determined differently according to the target disease. Specifically, since the region of interest at the time of diagnosis may be different according to the target disease, regions other than the region of interest of diagnosis may be excluded from the target region.

Also, for example, a quality condition determined when determining the quality of the fundus image may be differently determined according to a target disease. In detail, whether to detect a pixel, a range of pixel values of the detected pixel, and the like may be determined according to the target disease.

In determining the quality or suitability of an ocular fundus image, the target disease may be divided into several types. Target diseases may include eye diseases, chronic diseases, systemic diseases, ophthalmic lesions, ophthalmic findings, and the like described throughout this specification.

For example, a target disease may be a first type disease that can be diagnosed based on an ocular fundus image containing any artifact, a second type disease that is difficult to diagnose based on an ocular fundus image that includes any artifact, and a fundus containing a specific type of artifact. Based on the image, it can be considered to be divided into a third type of disease that is difficult to diagnose.

The first quality criterion for determining suitability of the fundus image for a type 1 disease is the second quality criterion and / or criterion for determining suitability of the fundus image for a second type disease is fundus image for a third type disease. It may include a more relaxed criterion than the third quality criterion for determining the suitability of a.

When the first type of disease is a target disease, when at least one artifact is detected, the fundus image may be determined as a bad image. When the second type of disease is a target disease, the quality judgment of the fundus image related to the artifact detection may be omitted.

When the third type of disease is a target disease, the suitability of the fundus image may be determined by performing a quality criterion determination including determining whether there is a target artifact that affects the diagnosis of the target disease, that is, whether there is a target artifact corresponding to the target disease. have. In the case where the third type of disease is a target disease, the suitability of the fundus image is determined by performing a quality criterion including a quality judgment on a target area of the fundus image used for diagnosis of the target disease, that is, a target area corresponding to the target disease. Can be judged.

On the other hand, artifacts occurring in the fundus image can be thought of as being divided into several types. For example, an artifact may be a first type artifact that makes it difficult to diagnose all diseases based on the fundus image, a second type artifact that does not interfere with the diagnosis of a disease based on the fundus image, an agent that interferes with the diagnosis of a specific disease group based on the fundus image. There can be three type artifacts.

The fundus image from which the first type artifact is detected may be determined as a bad image. Determination for the second type artifact may be omitted. Even if the second type artifact is detected, a target operation based on the fundus image, such as a diagnostic process, may be performed. However, the occurrence of the second type artifact may be notified to the user.

In the case of the third type artifact, when the disease that is difficult to diagnose due to the target artifact is the target disease, that is, when the disease corresponding to the target artifact is the target disease, the target operation based on the fundus image may not be performed. Alternatively, when a disease that is not a disease corresponding to the target artifact is a target disease, a target operation based on the fundus image may be performed. In this case, the occurrence of the third type artifact may be notified to the user.

In the following, the suitability of the fundus image according to the disease is mainly considered in consideration of the correspondence relationship between the artifact and the disease, the correspondence relationship between the specific region and the disease, and the correspondence relationship between the specific artifact and the region, based on the third type disease and the third type artifact. It will be described how to judge. Diseases and artifacts, diseases and domains, or artifacts and domains may have a one-to-one correspondence.

The correspondence between disease and artifact, disease and territory, or artifact and territory, may be a many-to-one or one-to-many response. For example, the first disease may correspond to the first artifact and / or the second artifact. Alternatively, the first disease may correspond to the first region and / or the second region. Alternatively, the first artifact may correspond to the first region and / or the second region. Alternatively, the first artifact may correspond to the first disease and / or the second disease. Alternatively, the first region may correspond to the first artifact and / or the second artifact. Alternatively, the first region may correspond to the first disease and / or the second disease.

50 is a diagram illustrating a method for diagnosing lesions. FIG. 50 illustrates the areas considered for the diagnosis, for various findings, lesions or diseases that can be diagnosed from the fundus image.

Referring to FIG. 50A, hemorrhage may be diagnosed based on the fundus image. For example, as shown in (a), bleeding L1 can be detected from any position throughout the fundus region.

Referring to FIG. 50B, a drusen may be diagnosed based on the fundus image. For example, as shown in (b), drusen (L2) can be found from any position throughout the fundus region.

Referring to FIG. 50C, a glaucomatous disc change may be diagnosed based on the fundus image. For example, as shown in (c), glaucoma finding (L3) can be detected from the optic nerve papilla of the fundus image.

Referring to FIG. 50D, macular holes may be diagnosed based on the fundus image. For example, as shown in (d), macular puncture finding L4 can be found from the macular portion of the fundus image.

Referring to FIG. 50E, the macular epiretinal membrane may be diagnosed from the fundus image. For example, as shown in (e), macular epithelial findings (L5) can be found near the macula of the fundus image. Yellow foil epithelium findings can be found from the macula of the fundus image.

Referring to FIG. 50F, symptoms of myelinated nerve fiber may be diagnosed from the fundus image. For example, as shown in (f), myelinated nerve fiber (L6) findings can be found from any region of the fundus image.

When lesions are detected from any area of the fundus image as shown in FIGS. 50A, 50B and 50F, it may be desirable to evaluate the quality over the entire fundus area. As shown in (c) of FIG. 50, in the case of findings or lesions diagnosed based on the shape of the optic nerve papilla, it may be efficient to perform the quality evaluation around the area around the optic nerve papilla. In addition, when findings are found from around the macula, as shown in FIGS. 50D and 50E, it may be efficient to determine the quality around the macula.

Although some diseases have been exemplarily described in FIG. 50, they may be similarly applied to various other fundus-based diagnosis.

In consideration of the above-described background, the method for determining the quality of the fundus image described herein may be implemented differently according to the target disease. Hereinafter, in the case where the above-described diagnosis of the plurality of lesions is to be performed, a method of determining the quality of an image in consideration of diagnosis characteristics of each lesion will be described with reference to some embodiments.

2.4.2 How to determine disease suitability

The image quality determination method may be performed differently according to the diagnosis target lesion or disease. In other words, other criteria may be applied according to the target disease, the disease of interest, the target lesion, or the lesion of interest to determine suitability of the image. The method of determining the quality of the fundus image or the method of determining the suitability of the fundus image described below may be performed by the controller described above. The order of each step of the quality determination method described below may be changed.

Hereinafter, some embodiments will be described with reference to a method for determining suitability of an image in consideration of a target disease.

51 is a diagram for describing a method of determining suitability of an image according to an embodiment of the present invention.

Referring to FIG. 51, a method of determining suitability of an image according to an embodiment of the present invention may include acquiring a fundus image (S31) and acquiring suitability information of a fundus image (S33). The method of determining suitability of an image according to an embodiment of the present invention may further include converting a fundus image. With regard to the transformation of the fundus image, the above description may be analogized and applied herein.

Acquiring the suitability information of the fundus image (S33) may include performing a quality determination on the fundus image, and obtaining the suitability information for the target disease of the fundus image based on the quality determination result. For the quality judgment of the fundus image, the foregoing description may be inferred.

52 is a view for explaining a method of determining suitability of an image according to an embodiment of the present invention. Referring to FIG. 52, in the method of determining suitability of an image according to an embodiment of the present disclosure, the method may include obtaining a fundus image (S31) and determining whether the fundus image satisfies a quality condition corresponding to a target disease ( S33a), may include obtaining information on suitability for the target disease of the fundus image (S33b).

Obtaining suitability information of the fundus image (S33) may include determining whether the fundus image satisfies a quality condition (or object suitability condition) corresponding to the target disease (S33a).

The subject suitability condition may be a condition for determining whether the fundus image is an image that can be used to diagnose the subject disease. The quality condition corresponding to the target disease may refer to a quality condition determined in consideration of specific characteristics of the target disease in order to select an image that may be used for diagnosis of the target disease. The subject suitability condition may be a criterion for determining whether the accuracy of diagnosis for the subject disease based on the fundus image can be sufficiently ensured.

Target conformance conditions may include quality conditions. Determination of the quality conditions and quality conditions can be analogously applied to the above-described content herein. For example, the quality condition may include a reference range of pixel values, a ratio of pixels having pixel values belonging to the reference range, and the like.

Target suitability conditions may be prepared in advance to correspond to the target disease. In other words, the suitability determination device (or the controller) may determine whether the fundus image satisfies the target suitability condition corresponding to the target disease by using a database including a plurality of target suitability conditions prepared in advance to correspond to each of the plurality of target diseases. have.

For example, when the target disease is the first disease, the suitability determination method of the fundus image may include determining whether the fundus image satisfies the first target suitability condition.

Target suitability conditions may include quality conditions for the area. For example, the object suitability condition may include a first quality condition associated with a first area from which at least a portion of the fundus area is excluded. Target eligibility conditions may include quality-specific criteria. For example, the object suitability condition may include a first quality condition for the first area and a first quality condition for the second area.

Subject suitability conditions may include quality conditions for the area corresponding to the subject disease. In other words, the subject suitability condition may include a quality condition for the subject area, where the subject area may be determined based on the subject disease.

As a specific example, when the disease is a disease diagnosed based on lesions or findings mainly occurring in the internal area IA of the fundus area FA, the first target eligibility condition includes a reference quality condition for the internal area IA. can do. If the first disease is glaucoma diagnosed based on the form of the optic nerve papilla, the first target suitability condition may include a reference quality condition for the internal area IA.

As another specific example, when the first disease is a disease diagnosed based on lesions or findings mainly occurring in the macular region MA of the fundus region FA, the first target suitability condition is a reference to the macular region MA of the fundus image. May include quality conditions. As a specific example, when the first disease is macular perforation, the first target suitability condition may include that no artifact is detected from the macular area MA.

Target suitability conditions may include conditions related to the artifact. Target conformance conditions may include conditions relating to the presence or absence of specific artifacts. For example, determining the object suitability condition may include determining whether the fundus image includes the first artifact. Target eligibility conditions may include conditions related to the presence or absence of artifacts in a particular area. For example, determining the object suitability condition may include determining whether a first artifact has occurred in the first area of the fundus image.

Subject suitability conditions may include conditions related to artifacts corresponding to the subject disease. For example, the artifacts considered in subject suitability conditions may be artifacts that occur in an area at least partially overlapping an area on the fundus that is considered in diagnosing a disease. In other words, the subject suitability condition may include a condition regarding the presence or absence of a subject artifact associated with the subject disease. The subject artifact and the subject disease may overlap some areas detected from the fundus image.

As a specific example, when the first disease is a disease diagnosed based on lesions or findings mainly occurring in the outer area OA of the fundus region FA, the first target suitability condition may include a fundus image having a first artifact (eg, bright). Or artifacts). As a specific example, when the first disease is a cardiovascular chronic disease diagnosed in consideration of the shape of the blood vessel, the first target suitability condition may include no bright artifact.

As another specific example, determining the object suitability condition of the fundus image may include determining whether the first artifact and the second artifact are included.

When the subject disease is the first disease and the fundus image does not include the first artifact, it may be determined that the fundus image has a first suitability for the first disease. If the subject disease is the second disease and the fundus image does not include the second artifact, it may be determined that the fundus image has a second suitability for the second disease. When the subject disease is the third disease and the fundus image does not include the first artifact and the second artifact, the fundus image may be determined to have a third suitability for the third disease.

Obtaining suitability information of the fundus image (S33) may include obtaining suitability information of the target disease of the fundus image (S33b).

The suitability information for the target disease of the fundus image may indicate whether the fundus image is in an appropriate state for diagnosis of the target disease. The suitability information may indicate the probability that the diagnostic assistance information obtained from the fundus image is correct. Suitability information may include suitability or degree of suitability.

According to an embodiment, the suitability of the image according to an embodiment of the present invention may be performed for a plurality of target diseases. Determination of suitability for a plurality of subject diseases may be performed in one device. Determination of suitability for a plurality of target diseases may be performed in each of the plurality of devices.

For example, when the target diseases are the first disease and the second disease, the suitability determination method of the fundus image determines whether the fundus image satisfies the first target suitability condition corresponding to the first disease, and the fundus image corresponds to the second disease. And satisfying a second target suitability condition that is at least in part different from the first target suitability condition.

Determination of suitability for multiple subject diseases may be performed sequentially. When determining suitability for a plurality of target diseases, suitability between the target diseases may have a priority or inclusion relationship.

For example, when it is determined that the first fundus image is compatible with the first disease, the first fundus image may be determined to be compatible with the second disease. Alternatively, when it is determined that the first fundus image is not suitable for the first disease, the first fundus image may be determined to be not suitable for the second disease.

As a specific example, when the first disease is a bleeding finding that is determined in the entire fundus region, the first subject fitness condition is that the fundus image does not include the above-mentioned third artifact (for example, a cloudy artifact) for the entire fundus region. It may include. In addition, the second target suitability condition corresponding to the second disease may include that the outer area OA satisfies the quality criteria. In this case, when the first fundus image has a third artifact, it may be determined that the first fundus image is not suited to the second disease.

In other words, each of the target areas determined according to the target suitability conditions corresponding to each disease or the degree of each quality criterion determined according to the target suitability conditions corresponding to each disease has an inclusion relationship or priority. The quality judgment of the fundus image may be performed by first determining a target suitability condition having a higher priority or a wider (or stricter) criterion, and according to the result, the determination of a subordinate target suitability condition is omitted. Can be performed.

53 is a view for explaining a method of determining suitability of an image according to an embodiment of the present invention. Referring to FIG. 53, in the method of determining suitability of an image according to an embodiment of the present disclosure, the method may include obtaining a fundus image (S31), acquiring quality information on a target region of the fundus image (S33c), and a target region. It may include the step of obtaining the suitability information for the target disease of the eye fundus image based on the quality information for (S33d).

Acquiring suitability information of the fundus image (S33) may include acquiring quality information of a target area of the fundus image (S33c).

According to an embodiment, acquiring quality information of the target area of the fundus image may include acquiring quality information of the plurality of target areas. For example, acquiring quality information of the target region of the fundus image may include acquiring quality information of the first region and quality information of the second region.

Acquiring the suitability information of the fundus image (S33) may include acquiring the suitability information of the target disease of the fundus image based on the quality information of the target area (S33d).

According to an embodiment, acquiring suitability information of the target disease of the fundus image may be performed in consideration of a corresponding relationship between the target area and the target disease. For example, acquiring the suitability information of the target disease of the fundus image includes acquiring the suitability information of the fundus image for the first disease corresponding to the first target region in consideration of the quality information of the first target region. can do.

As a specific example, acquiring the suitability information of the fundus image, in the case where the first target region is the macula MA, in consideration of the quality information of the first target region, suitability of the fundus image for the first disease, for example, macular perforation, May include obtaining information.

54 is a diagram for describing a method of determining suitability of an image according to an embodiment of the present invention. Referring to FIG. 54, the method of determining suitability of an image according to an embodiment of the present disclosure may include obtaining a fundus image (S31), obtaining quality information on at least one artifact of the fundus image (S33e), and In consideration of the target artifact corresponding to the target disease and the quality information of the fundus image of the at least one artifact, the method may include obtaining information on suitability of the fundus image (S33f).

Acquiring suitability information of the fundus image (S33) may include acquiring quality information about at least one artifact of the fundus image (S33e).

According to an embodiment, acquiring quality information on artifacts of the fundus image may include acquiring quality information on a plurality of target artifacts. For example, acquiring quality information for an artifact of the fundus image may include acquiring first quality information for the first object artifact and second quality information for the second object artifact.

According to another embodiment, acquiring quality information about an artifact of the fundus image may include obtaining quality information about a target region of the fundus image and considering a corresponding relationship between the target region and the target artifact. And obtaining quality information. For example, acquiring quality information about an artifact of the fundus image may be achieved by obtaining first quality information about the first target region, and based on the first quality information, about the first artifact corresponding to the first target region of the fundus image. And obtaining quality information.

Acquiring the suitability information of the fundus image (S33), in consideration of the target artifact corresponding to the target disease and the quality information of the fundus image for the at least one artifact, obtaining the suitability information of the fundus image (S33f) It may include.

According to an embodiment of the present disclosure, acquiring the suitability information of the target disease of the fundus image may include obtaining quality information about the target artifact of the fundus image and considering the correspondence relationship between the target artifact and the target disease to obtain suitability information of the fundus image. May include acquiring. For example, acquiring the suitability information of the fundus image includes acquiring quality information about the first artifact of the fundus image and acquiring first suitability information about the first target disease corresponding to the first artifact of the fundus image. can do.

As a more specific example, when the target disease is glaucoma detected mainly from the optic nerve papilla located close to the center of the fundus region (except when the imaging state of the fundus image is set such that the optic nerve papilla is located far from the center of the fundus region) Acquiring the suitability information of the fundus image for the target disease is based on whether the fundus image includes a first artifact (eg, a bright artifact), and the subject of the fundus image for glaucoma corresponding to the first artifact. And determining suitability.

According to an embodiment of the present invention, the method of determining the suitability of the fundus image may include performing suitability determination on the target disease with respect to the fundus image, wherein the suitability determination on the target disease may be based on a quality standard corresponding to the target disease. This can be done by determining if it is satisfied.

55 is a view to explain a method of determining suitability of a fundus image according to an embodiment of the present invention. Referring to FIG. 55, according to an embodiment of the present disclosure, a method of determining suitability of a fundus image may include obtaining a fundus image (S310) and determining a target suitability of the fundus image by using a quality criterion corresponding to a target disease. It may include (S330).

Target diseases include glaucoma, cataracts, bleeding, drusen, exudate, cotton patch, cloudiness, retinal pigment change, macular perforation, ocular diseases including macular degeneration and prematurity retinopathy, systemic diseases including diabetic, Alzheimer's, cytomegalovirus, It may be any one selected from cardiovascular diseases including hypertension, hypotension, stroke, atherosclerosis, angina pectoris and heart failure.

According to an embodiment of the present invention, in connection with the diagnosis of the target disease based on the fundus image, the method of determining suitability of the fundus image may include obtaining the fundus image (S310) and the fundus image corresponding to the target disease. In operation S330, the object suitability of the fundus image may be determined by using the quality standard. In this case, the target disease may include at least one of the first disease and the second disease.

Determining the subject suitability of the fundus image may include determining whether the fundus image meets a predetermined quality criterion to correspond to the subject disease. For example, determining the object suitability may include determining whether the fundus image satisfies the first quality standard when the target disease is the first disease, and determining whether the fundus image satisfies the second quality standard when the target disease is the second disease. It may include.

The quality criterion may include a quality criterion for the target area. For example, the quality criterion may include a criterion about a pixel value (eg, a brightness value) of pixels distributed in the target area or an amount or number of artifacts distributed in the target area.

Determining the object suitability of the fundus image determines whether the first region of the fundus image satisfies the first quality criteria when the target disease is the first disease, and when the target disease is the second disease, the second region of the fundus image And determining whether the quality criteria are satisfied.

The first disease may be selected from a first region corresponding disease group that is determined based at least in part on the first region. For example, the first region corresponding disease group may be a group of diseases that are determined based at least in part on an inner region of the fundus image. When the first region-corresponding disease group is a soldier of a disease that is determined based at least in part on the medial region of the fundus image, the macula or optic nerve papilla is generally located in the medial region of the fundus region in the photographed fundus image, and thus the macula or optic nerve Diseases, lesions or findings diagnosed based on the type of papilla may correspond to the first disease group.

The second disease may be selected from a second region corresponding disease group that is determined based at least in part on the second region and at least some other second region. For example, the second area corresponding disease group may be a group of diseases that are determined based at least in part on the outer area of the fundus image. In the case where the second region-corresponding disease group is a soldier whose disease is determined based at least in part on the outer region of the fundus image, the disease is diagnosed based on the state or form of the vessel because most of the blood vessels of the fundus image are located outside the fundus region. , Lesions or findings may correspond to the second disease group.

When a disease belonging to the first region-corresponding disease group is a target disease, object suitability of the fundus image may be determined based on a first quality criterion for the first region. When the disease belonging to the second region-corresponding disease group is a target disease, the object suitability of the fundus image may be determined based on a second quality criterion for the second region.

The subject disease may be selected from the first or second artifact corresponding disease group. The first or second artifact corresponding disease group may be a group of diseases in which the occurrence position of the first or second artifact and the detection position thereof at least partially overlap.

The subject disease may be selected from the first or second quality criteria corresponding disease group. The first or second artifact-corresponding disease group may be a group of diseases detected from the first or second area to be determined in the first or second quality criteria, respectively.

The target area may include a first area spaced at least a first distance from the center of the area corresponding to the fundus of the fundus image, a second area within a first distance from the center of the area corresponding to the fundus, and an optic nerve papilla among the areas corresponding to the fundus. It may include at least one selected from the third region corresponding to the region in which is distributed and the fourth region corresponding to the region in which the macular is distributed among the regions corresponding to the fundus. The target area may be defined as at least a part of the detailed area of the fundus image described herein. Alternatively, the target area may be determined by the user.

Quality criteria may include quality criteria for the presence or absence of artifacts. In other words, the quality criteria for determining object suitability for a specific disease of the fundus image may include information about the presence or absence of artifacts corresponding to the disease. For example, the quality criteria may include the presence or absence of artifacts that are hindered in using the fundus image as a basis for diagnosis of a specific disease.

For example, determining the object suitability of the fundus image may include determining whether the fundus image includes a first artifact corresponding to the first disease when the object disease is the first disease, and determining the first quality criterion; When the target disease is the second disease, determining the second quality criterion may further include determining whether the fundus image includes a second artifact corresponding to the second disease.

The first artifact corresponding to the first disease may be detected from a first region associated with the first disease. The second artifact corresponding to the second disease may be detected from a second region associated with the second disease.

The first disease may be diagnosed based at least in part on a first area within a predetermined distance from the center of the fundus area of the fundus image, and the first artifact may be an artifact detected in the first area.

For example, the first disease is glaucoma diagnosed based on the optic nerve papilla where the optic papilla in the fundus region is distributed, and the first region may be determined to include the optic papilla on the fundus image from the center of the fundus region.

The second disease may be diagnosed based at least in part on any location of the second region, including the entire fundus region, and the second artifact may be an artifact detected in the second region.

For example, the second disease is a bleeding finding detected at any location of the fundus region, and the second region may be determined to include the entire fundus region.

56A is a diagram for describing a method of determining suitability of a fundus image according to an embodiment of the present invention. Referring to (a) of FIG. 56, in the method for determining the suitability of the fundus image according to the exemplary embodiment of the present invention, the method may include obtaining a fundus image (S310), determining a target suitability of the fundus image (S330), and a determination result. It may include the step (S351) to obtain the suitability information of the fundus image based on.

According to an embodiment, the method of determining suitability of the fundus image may further include obtaining suitability information of the fundus image based on a result of the object suitability determination (S351).

Obtaining the suitability information may include obtaining suitability information indicating that the fundus image is appropriate for the first disease if the subject disease is the first disease and the fundus image meets the first quality criterion.

Obtaining the suitability information may include obtaining suitability information indicating that the fundus image is suitable for the second disease if the subject disease is the second disease and the fundus image meets the second quality criterion.

The target disease may include one or more diseases, and since different quality criteria may correspond to each target disease, the fundus image may have different suitability for each target disease.

For example, the subject disease may include a first disease and a second disease. In this case, determining the object suitability of the fundus image may include determining whether the first quality criterion and the second quality criterion are satisfied with respect to the fundus image. If it is determined that the fundus image satisfies the first quality criterion but does not satisfy the second quality criterion, acquiring the suitability information of the fundus image is that the fundus image is suitable for the first disease but not suitable for the second disease. Acquiring suitability information indicating that

56 (b) is a diagram for describing a method of determining suitability of a fundus image according to an embodiment of the present invention. Referring to (b) of FIG. 56, the method for determining the suitability of the fundus image according to an embodiment of the present invention may include obtaining a fundus image (S310), determining a target suitability of the fundus image (S330), and a fundus image. It may include the step (S353) to obtain the diagnostic information on the target disease based on. The suitability determination method of the fundus image may include acquiring the suitability information (S351) and acquiring the diagnostic information about the target disease based on the suitability information (S353).

According to an embodiment, the method of determining suitability of the fundus image may further include acquiring diagnostic information about the target disease based on the fundus image.

In the obtaining of the diagnostic information, when the target disease is the first disease, when the fundus image satisfies the first quality criterion, the first diagnosis assistance information for the first disease may be acquired. The acquiring of the diagnostic information may further include acquiring second diagnostic assistance information for the second disease when the fundus image satisfies the second quality criterion when the target disease is the second disease.

According to an embodiment of the present invention, the method of determining the suitability of the ocular fundus image, prior to performing a quality judgment based on at least one quality criterion for the ocular fundus image, and if the target disease is obtained, This may include determining suitability for the disease.

57 is a view to explain a method of determining suitability of a fundus image according to an embodiment of the present invention. Referring to FIG. 57, according to an embodiment of the present disclosure, a method of determining suitability of a fundus image may include acquiring a fundus image (S410), determining whether a fundus image satisfies a quality criterion (S430), and a quality of a fundus image. Based on determining whether the criteria are satisfied, the method may include determining a target suitability for the target disease of the fundus image (S450).

Determining whether the ocular fundus image satisfies a quality criterion comprises determining whether the ocular fundus image satisfies a first quality criterion associated with the first region, and wherein the second quality criterion is associated with a second region in which the ocular fundus image is at least partially distinct from the first region. It may include determining whether to satisfy.

Determining subject suitability includes determining that the fundus image satisfies the first suitability when the subject disease is the first disease and the fundus image satisfies the first quality criterion, wherein the subject disease is the second disease and the fundus image is the second If the quality criteria are satisfied, the fundus image may include determining that the second suitability is satisfied.

Determining whether the fundus image satisfies a quality criterion may include determining the presence or absence of the first artifact in consideration of the first region, and determining the presence or absence of the second artifact in consideration of the second region.

Determining whether the fundus image meets the first quality criterion includes determining whether the fundus image includes the first artifact, and determining whether the fundus image meets the second quality criterion includes: Determining whether or not to include.

The first area may be an area spaced a predetermined distance or more from the center of the fundus area in which the fundus of the fundus image is distributed, and the second area may be an area within a predetermined distance from the center of the fundus area.

The first disease is selected from the first region-corresponding disease group determined based at least in part on the first region, and the second disease is the second region-corresponding disease determined based at least in part on a second region different from the first region at least in part. It can be selected from the group.

The first area or the second area may be an outer area spaced at least a first distance from the center of the area corresponding to the fundus of the fundus image, an inner area within a first distance from the center of the area corresponding to the fundus, and an area corresponding to the fundus. It may include at least one selected from the optic nerve papilla region corresponding to the region in which the optic nerve papilla is distributed and the macular region corresponding to the region in which the macular is distributed among the regions corresponding to the fundus.

58A is a diagram for describing a method of determining suitability of a fundus image according to an embodiment of the present invention. Referring to (a) of FIG. 58, in accordance with an embodiment of the present invention, a method of determining suitability of a fundus image may include obtaining a fundus image (S410), determining whether a fundus image satisfies a quality standard (S430), The method may include determining target suitability for the target disease of the fundus image (S450) and obtaining suitability information of the fundus image based on the result of the fitness stage (S471).

According to an embodiment, the method of determining suitability of the fundus image may further include acquiring suitability information of the fundus image based on the determination result.

Acquiring the suitability information, when the target disease is the first disease and the fundus image meets the first quality criterion, obtains the suitability information indicating that the fundus image is suitable for the first disease, and the target disease is the second disease. If it is a disease and the fundus image satisfies the second quality criterion, it may include obtaining suitability information indicating that the fundus image is suitable for the second disease.

58 (b) is a diagram for describing a method of determining suitability of a fundus image according to an embodiment of the present invention. Referring to (b) of FIG. 58, in accordance with an embodiment of the present invention, a method for determining suitability of a fundus image may include obtaining a fundus image (S410), determining whether a fundus image satisfies a quality standard (S430), The method may include determining target suitability for the target disease of the fundus image (S450) and obtaining diagnostic information about the target disease based on the fundus image (S473).

According to an embodiment, the method of determining suitability of the fundus image may further include acquiring diagnostic information about the target disease based on the fundus image.

Acquiring the diagnostic information, when the target disease is the first disease, if the fundus image meets the first quality criteria, obtains first diagnostic assistance information for the first disease, and when the target disease is the second disease, If the fundus image satisfies the second quality criterion, the method may include acquiring second diagnostic assistance information about the second disease. The first quality criterion and the second quality criterion may be independently determined.

If the target disease includes the first disease and the second disease, acquiring the diagnostic information does not acquire diagnostic assistance information for the first disease and the second disease if the fundus image does not satisfy the first quality criterion. It may include doing.

If the target disease includes the first disease and the second disease, acquiring the diagnostic information may include determining whether the fundus image satisfies the first quality criterion and does not meet the second quality criterion. The method may further include acquiring the first diagnostic assistance information and not acquiring the second diagnostic assistance information.

2.5 Application of Quality Judgment Results

By using the quality information of the image obtained as a result of the above-described image quality determination method, a neural network model for diagnosing a disease based on the fundus image may be learned or driven. Hereinafter, a method of managing fundus image data using quality information of an image, a method of learning a neural network model, a method of performing a diagnosis using a neural network model, and the like will be described with reference to some embodiments.

In addition, when determining the quality or suitability of the image in consideration of the characteristics of the target lesion as described above, the acquired lesion response quality information or lesion response suitability information may be used in various forms depending on the stage in which the fundus image is used. .

Regarding data management, learning of neural network models, and diagnosis using neural network models described below, information about fundus image data, neural network models, diagnosis, and the like described in the diagnosis assistance item using the fundus image may be inferred.

2.5.1 Data management stage

According to an embodiment of the present invention, the method of determining the quality of the fundus image or the method of determining the suitability of the fundus image may be used in the data management step.

In the fundus image data management step, a plurality of fundus images may be classified and managed according to whether quality criteria are satisfied. Alternatively, when a quality criterion is determined in consideration of the target lesion, images having suitability for each target lesion may be selected and managed.

For example, in the fundus image data management step, the fundus image data may be classified, screened, or classified using a method for determining the quality of the fundus image or a suitability for determining the suitability of the fundus image described herein. As a result, at least one fundus image data set that can be used for the purpose can be obtained.

According to an embodiment of the present invention, the method of determining the quality of the fundus image may be used in the data management step. As an embodiment, a method of managing fundus image data using a quality determination of a fundus image may be provided. According to an embodiment, the data management method may include managing the fundus image based on quality information of the fundus image. The data management method described below may be performed by the image management apparatus, the controller, or the processor.

According to an embodiment of the present disclosure, a fundus image data set including a plurality of fundus images labeled with quality information may be obtained by a method of managing fundus image data using the quality of the fundus image. The fundus image data set may include a plurality of fundus images classified according to the quality information.

The data management method according to an embodiment of the present invention may include assigning quality information to the fundus image based on the quality information of the fundus image. For example, the quality information may be tagged or labeled in the fundus image (eg, the quality information manager or the quality determination unit may assign the quality information to the fundus image). The quality information may include normal or abnormal information of the fundus image, whether an artifact of the fundus image is included, quality information of each region of the fundus image, and the like.

For example, the data management method may be configured to assign first quality information to the first fundus image based on quality information of the first fundus image, and to assign the first quality information to the second fundus image based on the quality information of the second fundus image. Imparting other second quality information.

A data management method according to an embodiment of the present invention may include generating, storing, and / or managing a fundus data set including a plurality of fundus images based on quality information of the fundus image.

For example, according to an embodiment of the present invention, a data management method includes a first fundus image data set including a plurality of fundus images having first quality information and a plurality of fundus images having second quality information. Two fundus image data sets may be provided. As a specific example, the first fundus image data set may include a plurality of fundus images to which normal quality information indicating that the fundus image is normal. The second fundus image data set may include a plurality of fundus images to which abnormal quality information has been assigned indicating that the fundus image is abnormal (eg, if it contains a defect, artifact, or abnormal condition).

The quality information may include various information as described above in the image quality determination item. For example, the quality information may include various detailed information. In this regard, the fundus image data set generated according to the quality information may include a plurality of sub data sets.

According to another embodiment of the present invention, a method for determining suitability of the fundus image may be used in the data management step. As one embodiment of the present invention, a method for managing fundus image data using the suitability of the fundus image may be provided. The data management method described below may be performed by the image management apparatus, the controller, or the processor.

According to an embodiment of the present disclosure, a fundus image data set including a plurality of fundus images labeled according to the suitability information may be obtained by the suitability determination method of the fundus image data using the suitability determination of the fundus image. The fundus image data set may include a plurality of fundus images classified according to suitability information.

The data management method according to an embodiment of the present invention may include assigning suitability information to the fundus image based on the suitability information of the fundus image. For example, suitability information may be tagged or labeled on the fundus image (eg, the suitability information management unit or the suitability determination unit may assign suitability information to the fundus image). The suitability information may include suitability information for the target disease of the fundus image. The suitability information may include identification information of the target disease and suitability information of the target disease of the fundus image.

For example, the data management method may assign first suitability information to the first fundus image based on suitability information of the first fundus image, and based on the suitability information of the second fundus image, the first suitability information may be added to the second fundus image. And may give other second suitability information.

A data management method according to an embodiment of the present invention may include generating, storing and / or managing a fundus data set including a plurality of fundus images based on suitability information of the fundus image.

For example, according to an embodiment of the present invention, a data management method includes a first fundus image data set including a plurality of fundus images having first suitability information and a plurality of fundus images having second suitability information. Two fundus image data sets may be provided. As a specific example, the first fundus image data set may include a plurality of fundus images to which first suitability information is assigned indicating that the fundus image is determined to be suitable for the first disease. The second fundus image data set may include a plurality of fundus images given second suitability information indicating that the fundus image is determined to be suitable for the second disease.

The suitability information may include various information as described above in the image suitability determination item. For example, the suitability information may include various detailed information, and the fundus image data set generated based on the suitability information may include a plurality of sub data sets.

The database can be managed based on the identification information. For example, the database may be managed based on identification information of the fundus image, identification information of a disease, identification information of an area, and identification information of an artifact. In addition, for example, the database may be managed in association with identification information and quality information and / or suitability information.

As a specific example, when the target region and the target disease are managed in association with each other, identification information of the target region and identification information of the target disease may be matched and stored. When the target disease and the target artifact are managed in association with each other, the identification information of the target disease and the identification information of the target artifact may be matched and stored. As another specific example, when the fundus image has quality information and / or suitability information, the fundus image and the quality information and / or suitability information may be matched and stored.

2.5.2 Learning steps

In the training or learning of the neural network model, the neural network model may be trained using the fundus image data set using the quality information or the suitability information. For example, in training a neural network model, in relation to the function or purpose of the model, a fundus image data set generated based on quality information or suitability information may be selectively used. Regarding the training of the neural network model, in particular, the training of the diagnostic auxiliary neural network model, the information about the diagnostic auxiliary neural network model described above in the Diagnostics Aid item may be inferred.

According to an embodiment of the present invention, the neural network model may be trained using the fundus image data set obtained by using the method for determining the quality of the fundus image described herein.

For example, the neural network model may be trained using a fundus image data set including a plurality of fundus images determined as normal images according to a quality determination method. The diagnostic assistance neural network model may be trained using a fundus image data set that includes a plurality of fundus images to which quality information has been assigned. The diagnostic assisted neural network model may be trained using a fundus image data set that includes a plurality of fundus images labeled with normal information.

For example, the neural network model may be trained by the fundus image data set acquired based on the quality information determined according to various embodiments described in the above-described image quality determination method item. For example, the diagnostic assistance neural network model may be trained by a fundus image data set that includes only fundus images labeled with normal quality information.

A neural network model trained using a fundus image data set including a plurality of fundus images labeled as normal images and determined as normal images is a neural network model trained using a fundus image data set including both normal and abnormal images. It can have improved prediction accuracy.

As another example, the neural network model may be trained using a fundus image data set including a plurality of fundus images including quality information obtained according to a quality determination method considering a region. The diagnostic assisted neural network model may be trained using a fundus image data set including a plurality of fundus images, to which quality information for a specific region is assigned.

For example, the neural network model may be trained using a fundus image data set that includes a plurality of fundus images labeled with quality information for the first region. For example, the diagnostic assistance neural network model may be trained using a fundus image data set including only a plurality of fundus images having first quality information indicating satisfaction of the quality information with respect to the first region.

In the case of a neural network model trained using training data composed of fundus images having normal quality information for a specific region, the neural network model trained using training data having a random fundus image is based at least in part on a specific region. The prediction accuracy of diagnostic assistance information for the disease to be diagnosed may be improved.

In addition, for example, the neural network model may be trained using a fundus image data set including only a plurality of fundus images having second quality information indicating dissatisfaction with quality information for the first region.

In the case of a neural network model trained using training data composed of fundus images having abnormal quality information for a specific region, the quality of a specific region is lower than that of a neural network model trained using training data having a random fundus image. When the input fundus image is input, a more enhanced result can be output.

As another example, the neural network model may be trained using a fundus image data set including a plurality of fundus images including quality information obtained according to a quality determination method considering an artifact. The diagnostic assistance neural network model may be trained by a fundus image data set including a plurality of fundus images to which quality information indicating the presence or absence of artifacts is assigned.

For example, the neural network model may be trained using a fundus image data set that includes a plurality of fundus images that are determined to not include the first artifact. For example, the diagnostic assisted neural network model may be trained using a fundus image data set consisting of a plurality of fundus images labeled with quality information indicating no first artifact.

The neural network model trained using the fundus image data composed of the fundus images without artifacts may have improved prediction accuracy compared to the neural network model trained using the fundus image data composed of arbitrary fundus images. .

In addition, for example, the diagnostic assisted neural network model may be trained using a fundus image data set including a plurality of fundus images determined to include the first artifact. For example, the diagnostic assistance neural network model may be trained using a fundus image data set consisting of a plurality of fundus images labeled with quality information indicating that the first artifact is included.

For neural network models trained using fundus image data consisting of fundus images containing specific artifacts, they are robust to specific artifacts compared to neural network models trained using fundus image data composed of arbitrary fundus images. It can have one characteristic. That is, the neural network model trained using the fundus image data composed of the fundus images including the specific artifacts may exhibit more improved prediction accuracy when the fundus image including the specific artifacts is input.

According to an embodiment of the present invention, the diagnostic assistance neural network model may be trained using the fundus image data set obtained by using the determination of suitability of the fundus image described herein.

As an example, the neural network model may be trained using a fundus image data set including a plurality of fundus images determined as appropriate images for a target disease according to a suitability determination method. The diagnostic assisted neural network model may be trained using a fundus image data set including a plurality of fundus images to which suitability information for a target disease has been assigned.

For example, the neural network model may be trained using a fundus image data set that includes fundus images suitable for the subject disease. The diagnostic assisted neural network model may be trained using a fundus image data set comprising a plurality of fundus images labeled with first suitability information indicating that the fundus image is suitable for a first subject disease.

The neural network model trained using the training data composed of the fundus images determined to be suitable for the target disease predicts diagnosis aid information related to the target disease, compared to the fundus image trained using the training data composed of arbitrary fundus images. It can have an improved accuracy in.

Also, for example, the neural network model can be trained using a fundus image data set that includes fundus images that are inappropriate for the subject disease. The diagnostic assisted neural network model may be trained using a fundus image data set comprising a plurality of fundus images labeled with second suitability information indicating that the fundus image is inadequate for the first target disease.

The neural network model trained using training data composed of fundus images determined to be inappropriate for the target disease predicts diagnosis aid information related to the target disease, compared to the fundus image trained using training data composed of arbitrary fundus images. In this case, even when the quality of the fundus image is poor, it can have more improved accuracy.

The method of determining suitability of the fundus image disclosed herein may also be used when using a plurality of neural network models in parallel. For example, when learning a plurality of neural network models, the fundus image determined suitably for the target disease of each neural network model may be used for learning. Regarding the parallel configuration of a plurality of neural network models, the above-described contents of the diagnostic system item for the plurality of labels may be inferred.

For example, the plurality of neural network models may include a first neural network model for assisting diagnosis of the first disease and a second neural network model for assisting diagnosis for the second disease. The first neural network model may be learned and / or driven in the first device. The second neural network model may be learned and / or driven in the first device. The first neural network model and the second neural network model may be learned and / or driven in a third device. The first neural network model and / or the second neural network model may be trained in the fourth device and driven in the fourth device.

Hereinafter, when the first neural network model and the second neural network model are used, a method of learning each neural network model using the determination of suitability of the fundus image will be described with reference to some embodiments.

According to an embodiment of the present invention, the training of the diagnosis assisted neural network model may be performed by learning a first neural network model for assisting diagnosis of the first disease by using the first fundus image data, and using the second fundus image data. Training a second neural network model to assist in the diagnosis of the second disease. In other words, training the diagnostic assistance neural network model includes training the first neural network model using first fundus image data labeled with the first suitability information (or first quality information), and the second suitability information (or second). Training the second neural network model using the second fundus image data labeled quality information).

For example, learning a diagnostic assisted neural network model may be performed by learning a first neural network model for assisting diagnosis of a first disease using first fundus image data including a plurality of fundus images determined to be suitable for the first disease. And training a second neural network model to assist in the diagnosis of the second disease using second fundus image data comprising a plurality of fundus images determined to be suitable for the first disease and another second disease. Can be.

The first fundus image data includes a plurality of fundus images determined to be suitable for the first region, and the first disease may be a disease diagnosed based at least in part on the first region. The second fundus image data includes a plurality of fundus images determined to be suitable for a second region that is at least partially different from the first region, and the second disease may be a disease diagnosed based at least in part on the second region.

The first fundus image data includes a plurality of fundus images that are determined to not include the first artifact that is at least partially detected from the first region, and the first disease may be a disease diagnosed based at least in part on the first region. have. The second fundus image data includes a plurality of fundus images that are determined to not include a second artifact that is at least partially detected from a second region that is at least partially different from the first region, and wherein the second disease is at least partially relative to the second region. It can be a disease that is diagnosed on the basis.

In training each diagnostic assisted neural network model, when using the appropriate fundus image data set for the target disease of each model, each of the neural network models uses a random (or identical) data set to train the model. The accuracy of predictive diagnostic assistance information for individual diseases of the neural network model can be improved.

In addition, for example, training the diagnosis assisted neural network model may include: a first neural network model for assisting diagnosis of the first disease by using first fundus image data including a plurality of fundus images determined to be inappropriate for the first disease. Learning a second neural network model for assisting diagnosis of a second disease by using second fundus image data including a plurality of fundus images determined to be inappropriate for a first disease and another second disease. It may include.

The first fundus image data includes a plurality of fundus images that are determined to be inappropriate for the first region, and the first disease may be a disease diagnosed based at least in part on the first region. The second fundus image data includes a plurality of fundus images that are determined to be inappropriate for a second region that is at least partially different from the first region, and the second disease may be a disease diagnosed based at least in part on the second region.

The first fundus image data may include a plurality of fundus images determined to include a first artifact that is at least partially detected from the first region, and the first disease may be a disease diagnosed based at least in part on the first region. . The second fundus image data includes a plurality of fundus images determined to include a second artifact that is at least partially detected from a second region that is at least partially different from the first region, wherein the second disease is based at least in part on the second region. It can be a disease diagnosed by.

In training each diagnostic assisted neural network model, when using the appropriate fundus image data set for the target disease of each model, each of the neural network models uses a random (or identical) data set to train the model. The accuracy of predictive diagnostic assistance information for individual diseases of the neural network model can be improved.

2.5.3 Diagnostic steps

In the step of performing diagnosis or diagnostic assistance using the learned neural network model, quality determination may be applied when acquiring a diagnosis target fundus image. In addition, in the case of diagnosing a specific disease of the neural network model used for diagnosis, a fundus image suitable for the disease may be used as the diagnosis target image. The neural network model may be a diagnostic auxiliary neural network model trained using the fundus image data set obtained through the above-described image quality determination or suitability determination.

According to an embodiment of the present invention, the neural network model may output diagnostic assistance information using the fundus image obtained by using the method for determining the quality of the fundus image described herein.

For example, the diagnostic assistance method may include outputting diagnostic assistance information based on a fundus image determined as a normal image according to a quality determination method using a neural network model. For example, the diagnostic assistance method according to an embodiment of the present invention includes determining a quality of a diagnosis target fundus image and obtaining diagnostic assistance information from a diagnosis assistance neural network model based on the fundus image determined to have a normal quality. can do.

Alternatively, the diagnostic assistance method may include determining a quality of a diagnosis target fundus image and obtaining diagnostic assistance information from a diagnosis assistance neural network model based on the fundus image determined to have abnormal quality. In this case, the quality information indicating that the target fundus image is abnormal may be output together with the diagnosis assistance information obtained based on the target fundus image. Alternatively, the method for assisting diagnosis according to an embodiment of the present disclosure may include determining a quality of an image of a fundus to be diagnosed, and providing a reshooting notification to a user or switching to a reshooting mode when the fundus image has abnormal quality. can do.

When the diagnostic assistance information is obtained using the selected target fundus image based on a quality judgment result of whether the fundus image is normal, the output of the diagnostic assistance information based on the defective fundus image may be prevented. Reliability can be enhanced.

As another example, the diagnostic assistance method may further include obtaining diagnostic assistance information based on an object fundus image satisfying a quality criterion for the target region and using a neural network model. For example, the diagnostic assistance method according to an embodiment of the present invention may determine the quality of the target region of the diagnosis target fundus image and diagnose from the diagnosis assist neural network model based on the fundus image determined to have a normal quality with respect to the target region. Obtaining assistance information.

Alternatively, the diagnostic assistance method may include determining a quality of the diagnosis fundus image and obtaining diagnostic assistance information from the diagnosis assistance neural network model based on the fundus image determined to have an abnormal quality with respect to the target region. In this case, quality information indicating that the target fundus image is abnormal with respect to the target area may be output together with the diagnostic assistance information obtained based on the target fundus image. Alternatively, the diagnostic assistance method according to an embodiment of the present disclosure may include providing a reshooting notification to the user or switching to a reshooting mode when the target region of the fundus image has abnormal quality.

When diagnosing assistance information is obtained based on a target fundus image selected as a fundus image that satisfies a quality criterion for the target region, the accuracy of the diagnosis assistance information for the target disease associated with the target region may be improved.

As another example, the diagnostic assistance method may further include obtaining diagnostic assistance information based on an object fundus image satisfying a quality criterion with respect to the target artifact and using a neural network model. For example, the diagnostic assistance method according to an embodiment of the present invention may determine whether or not the target artifact is included in the diagnosis target fundus image, and diagnose assistance information from the diagnosis assist neural network model based on the fundus image determined to not include the target artifact. It may include acquiring.

Alternatively, the diagnostic assistance method may include determining a quality of the diagnosis fundus image, and obtaining diagnostic assistance information from the diagnosis assistance neural network model based on the fundus image determined to have abnormal quality with respect to the object artifact. In this case, quality information indicating that the target fundus image includes the target artifact may be output together with the diagnostic assistance information obtained based on the target fundus image. Alternatively, the diagnostic assistance method according to an embodiment of the present disclosure may include providing a reshooting notification to the user or switching to a reshooting mode when the fundus image includes a target artifact.

When the diagnostic assistance information is acquired based on the target fundus image selected as the fundus image that does not include the target artifact, the accuracy of the diagnosis assistance information on the target disease associated with the region in which the target artifact occurs may be improved.

According to an embodiment of the present invention, the diagnostic assistance neural network model may be trained using the fundus image data set obtained by using the determination of suitability of the fundus image described herein.

For example, the diagnostic assistance method may include outputting diagnostic assistance information based on a fundus image determined as an image suitable for a target disease according to a suitability determination method using a neural network model. For example, the diagnostic assistance method according to an embodiment of the present invention may include determining suitability of the diagnosis target fundus image and obtaining diagnostic assistance information from the diagnosis assistance neural network model based on the fundus image determined to be suitable for the target disease. It may include.

Alternatively, the diagnostic assistance method may include determining suitability of the diagnosis fundus image and obtaining diagnostic assistance information from the diagnosis assistance neural network model based on the fundus image determined to be inappropriate for the target disease. In this case, the suitability information indicating that the target fundus image is not suitable may be output together with the diagnostic assistance information obtained based on the target fundus image. Alternatively, the method for assisting diagnosis according to an embodiment of the present disclosure may determine suitability of the diagnosis target fundus image, and when the fundus image is not suitable for the target disease, provide a reshooting notification to the user or switch to a retake mode. It may include.

When acquiring diagnostic assistance information using the selected target fundus image based on the quality judgment result of the suitability of the fundus image, an image having a particularly inappropriate defect for the target disease may be excluded from the target fundus image. The reliability of the diagnostic assistance system for the disease can be enhanced.

The method of determining suitability of the fundus image disclosed herein may also be used when acquiring diagnostic assistance information using a plurality of neural network models in parallel. For example, when acquiring diagnostic assistance information using a plurality of neural network models, a plurality of conformance criteria or conditions may be used.

For example, in the case of a first diagnosis assistive neural network model for acquiring diagnosis assistance information about a first disease, diagnosis assistance information may be obtained using a fundus image satisfying a first suitability condition as a target fundus image. In the case of the second diagnostic assistive neural network model for acquiring the diagnostic assistance information for the second disease, the diagnosis assistance information may be obtained by using the fundus image that satisfies the second suitability condition as the target fundus image.

The plurality of neural network models may include a first neural network model for assisting diagnosis for the first disease and a second neural network model for assisting diagnosis for the second disease. The first neural network model may be learned and / or driven in the first device. The second neural network model may be learned and / or driven in the first device. The first neural network model and the second neural network model may be learned and / or driven in a third device. The first neural network model and / or the second neural network model may be trained in the fourth device and driven in the fourth device.

Hereinafter, some embodiments will be described using a first neural network model and a second neural network model to obtain diagnostic assistance information by considering suitability of the fundus image.

According to an embodiment of the present invention, the diagnostic assistance method determines whether the first fundus image corresponding to the first neural network model for assisting diagnosis of the first disease is satisfied with respect to the target fundus image, And determining whether a second suitability corresponding to the second neural network model for assisting the diagnosis of the disease is satisfied.

When the target fundus image satisfies the first suitability and the second suitability, the diagnostic assistance method acquires first diagnostic assistance information based on the target fundus image by using the first neural network model, and generates a second neural network model. And obtaining second diagnostic assistance information based on the target fundus image.

When the target fundus image satisfies the first suitability condition but does not satisfy the second suitability condition, the diagnostic assistance method acquires first diagnostic assistance information based on the target fundus image using the first diagnostic assistance neural network model. Only the first diagnostic assistance information may be output. In this case, the target fundus image may be output together with second suitability information indicating that the second suitability condition is not satisfied. Alternatively, the diagnostic assistance method may include outputting second diagnostic assistance information obtained based on the target fundus image by using the first diagnostic assistance information and the second diagnostic assistance neural network model, and outputting second suitability information together. can do.

When the object fundus image does not satisfy the first suitability condition and the second suitability condition, diagnostic assistance information may not be obtained. In this case, the first suitability information and the second suitability information indicating that the object fundus image does not satisfy the first suitability condition and the second suitability condition may be provided to the user. Alternatively, the diagnostic assistance method may further include a second diagnosis acquired based on the target fundus image using the first diagnosis assistance information and the second diagnosis assistance neural network model obtained using the second diagnosis assistance neural network model. The auxiliary information may be output, but the first and second suitability information may be output together.

According to an example, determining whether the target fundus image does not satisfy the first suitability condition may include determining whether a first region of the target fundus image meets a quality criterion. Determining whether the subject fundus image does not satisfy the second suitability condition may include determining whether a second region that is at least partially different from the first region of the subject fundus image meets a quality criterion.

According to an example, determining whether the object fundus image does not satisfy the first suitability condition may include determining whether the object fundus image includes the first artifact. Determining whether the subject fundus image does not satisfy the second suitability condition may include determining whether the subject fundus image includes a second artifact that is different from the first artifact.

When diagnosing assistance information is obtained using a diagnosis aid neural network model for diagnosing a plurality of target diseases, defects affecting diagnosis may be different for each disease. Therefore, the reliability of the diagnostic assistance information may be ensured for each diagnostic assistance neural network model by acquiring the diagnosis assistance information based on the fundus image having a fitness corresponding to the disease to be diagnosed.

In addition, even when the target fundus image is inappropriate, the subjective judgment of the user may be assisted by providing the diagnosis assistance information and the suitability information together.

In the above description, the quality judgment or suitability determination method of the fundus image has been described based on a case of applying the database construction, the neural network model learning, and the driving of the neural network model, but the content of the invention disclosed herein is not limited thereto. . Even in the case of an image other than a fundus image, defects may occur in a predetermined region in the image, and when predetermined information is obtained based on the image, the contents of the present disclosure disclosed herein may be inferred.

Although the embodiments have been described by the limited embodiments and the drawings as described above, various modifications and variations are possible to those skilled in the art from the above description. For example, the described techniques may be performed in a different order than the described method, and / or components of the described systems, structures, devices, circuits, etc. may be combined or combined in a different form than the described method, or other components. Or even if replaced or substituted by equivalents, an appropriate result can be achieved.

Therefore, other implementations, other embodiments, and equivalents to the claims are within the scope of the claims that follow.

Claims (37)

In the method for determining the suitability of the fundus image in relation to the diagnosis of the target disease based on the fundus image,
Acquiring the fundus image; And
Determining, on the fundus image, object suitability of the fundus image using a quality criterion corresponding to the subject disease, wherein the subject disease comprises at least one of a first disease and a second disease; Including,
The determining of suitability of the subject may include determining whether the fundus image satisfies a first quality criterion when the subject disease is a first disease, and when the subject disease is a second disease, the fundus image satisfies a second quality standard. Including judging
How to determine the suitability of a fundus image.
The method of claim 1,
The quality criteria include quality criteria for the target area,
Determining a subject suitability of the fundus image determines whether the first region of the fundus image satisfies the first quality criterion when the subject disease is a first disease, and when the subject disease is a second disease, the fundus image Determining whether the second region of satisfies the second quality criterion,
How to determine the suitability of a fundus image.
The method of claim 1,
The first disease is selected from a first region corresponding disease group that is determined based at least in part on the first region,
The second disease is selected from a second area corresponding disease group that is determined based at least in part on a second area that is at least partially different from the first area
How to determine the suitability of a fundus image.
The method of claim 2,
The target area corresponds to a first area spaced at least a first distance from a center of an area corresponding to the fundus of the fundus image, a second area within a first distance from a center of an area corresponding to the fundus, and the fundus. And at least one selected from a third region corresponding to a region in which the optic nerve papilla is distributed and a fourth region corresponding to a region in which the macular is distributed among the regions corresponding to the fundus.
How to determine the suitability of a fundus image.
The method of claim 1,
The quality criteria include quality criteria for the presence or absence of artifacts,
Determining the object suitability of the fundus image, if the target disease is a first disease, determining the first quality criterion determines whether the fundus image includes a first artifact corresponding to the first disease And when the target disease is a second disease, determining the second quality criterion further comprises determining whether the fundus image includes a second artifact corresponding to the second disease.
How to determine the suitability of a fundus image.
The method of claim 5,
A first artifact corresponding to the first disease is detected from a first region associated with the first disease, and a second artifact corresponding to the second disease is detected from a second region associated with the second disease,
How to determine the suitability of a fundus image.
The method of claim 5,
The first disease is diagnosed based at least in part on a first area within a predetermined distance from a center of the fundus area of the fundus image, wherein the first artifact is an artifact detected in the first area,
The second disease is diagnosed based at least in part on an arbitrary location of a second region including the entire fundus region, and the second artifact is an artifact detected in the second region.
How to determine the suitability of a fundus image.
The method of claim 7, wherein
The first disease is glaucoma diagnosed based on the optic nerve papilla where the optic nerve papilla is distributed in the fundus region, wherein the first region is determined to include the optic nerve papilla on the fundus image from the center of the fundus region,
The second disease is a bleeding finding detected at any position of the fundus region, wherein the second region is determined to include the entire fundus region,
How to determine the suitability of a fundus image.
The method of claim 1,
Obtaining diagnostic information about the target disease based on the fundus image; Include more,
Acquiring the diagnostic information,
When the target disease is the first disease, when the fundus image satisfies the first quality criterion, first diagnosis assistance information about the first disease is acquired.
If the target disease is the second disease, further comprising acquiring second diagnostic assistance information for the second disease if the fundus image satisfies the second quality criterion.
How to determine the suitability of a fundus image.
The method of claim 1,
Obtaining suitability information of the fundus image based on the determination result; Include more,
Acquiring the suitability information,
If the target disease is the first disease and the fundus image meets the first quality criterion, obtaining the suitability information indicating that the fundus image is suitable for the first disease,
If the subject disease is the second disease and the fundus image meets the second quality criterion, obtaining the suitability information indicating that the fundus image is suitable for the second disease
How to determine the suitability of a fundus image.
The method of claim 10,
The subject disease includes the first disease and the second disease,
Determining object suitability of the fundus image includes determining whether the first quality criterion and the second quality criterion are satisfied with respect to the fundus image,
If it is determined that the fundus image satisfies the first quality criterion but does not satisfy the second quality criterion, acquiring suitability information of the fundus image may be suitable for the first disease. Obtaining the suitability information indicating inadequacy for the second disease,
Method of suitability of the fundus image.
The method of claim 1,
Target diseases include glaucoma, cataracts, bleeding, drusen, exudate, cotton patch, cloudiness, retinal pigment change, macular perforation, ocular diseases including macular degeneration and prematurity retinopathy, systemic diseases including diabetic, Alzheimer's, cytomegalovirus, Any one selected from cardiovascular diseases including hypertension, hypotension, stroke, atherosclerosis, angina pectoris and heart failure,
How to determine the suitability of the fundus image
In the method for determining the suitability of the fundus image in relation to the diagnosis of the target disease based on the fundus image,
Acquiring the fundus image;
Determining whether the fundus image meets a quality criterion; And
Determining subject suitability of the fundus image for the subject disease based on determining whether the fundus image meets the quality criteria; Including,
Determining whether the fundus image satisfies the quality criterion comprises: determining whether the fundus image satisfies a first quality criterion associated with a first region, wherein the fundus image is associated with a second region that is at least partially distinct from the first region; Determining whether the second quality criterion is satisfied,
The determining the suitability of the subject may include determining that the fundus image satisfies the first suitability when the subject disease is the first disease and the fundus image satisfies the first quality criterion, and wherein the subject disease is the second disease. And when the fundus image satisfies the second quality criterion, determining that the fundus image satisfies a second suitability.
How to determine the suitability of a fundus image.
The method of claim 13,
Determining whether the fundus image satisfies the quality criteria includes determining the presence or absence of a first artifact in consideration of the first region, and determining the presence or absence of a second artifact in consideration of the second region.
How to determine the suitability of a fundus image.
The method of claim 13,
Determining whether the fundus image satisfies the first quality criterion includes determining whether the fundus image includes a first artifact,
Determining whether the fundus image meets the second quality criterion includes determining whether the fundus image includes a second artifact,
How to determine the suitability of a fundus image.
The method of claim 13,
The first area is an area spaced a predetermined distance or more from the center of the fundus area in which the fundus of the fundus image is distributed,
The second area is an area within a predetermined distance from the center of the fundus area.
How to determine the suitability of a fundus image.
The method of claim 13,
The first disease is selected from a first region corresponding disease group that is determined based at least in part on the first region,
The second disease is selected from a second area corresponding disease group that is determined based at least in part on a second area that is at least partially different from the first area,
How to determine the suitability of a fundus image.
The method of claim 13,
The first region or the second region may be an outer region spaced at least a first distance from a center of the region corresponding to the fundus of the fundus image, an inner region within a first distance from the center of the region corresponding to the fundus, And at least one selected from among optic nerve papilla regions corresponding to the areas where the optic nerve papilla is distributed among the areas corresponding to the fundus, and macular regions corresponding to the areas where the macula is distributed among the areas corresponding to the fundus.
How to determine the suitability of a fundus image.
The method of claim 13,
Obtaining diagnostic information about the target disease based on the fundus image; Include more,
Acquiring the diagnostic information,
When the target disease is the first disease, when the fundus image satisfies the first quality criterion, first diagnosis assistance information about the first disease is acquired.
If the target disease is the second disease, further comprising acquiring second diagnostic assistance information for the second disease if the fundus image satisfies the second quality criterion.
How to determine the suitability of a fundus image.
The method of claim 19,
The subject disease comprises the first disease and the second disease,
Acquiring the diagnostic information,
When the fundus image does not satisfy the first quality criteria, comprising not obtaining diagnostic assistance information for the first disease and the second disease,
How to determine the suitability of a fundus image.
The method of claim 19,
The subject disease comprises the first disease and the second disease,
Acquiring the diagnostic information,
When the fundus image satisfies the first quality criterion and does not satisfy the second quality criterion, the first diagnosis assistance information for the first disease is acquired, and the second diagnosis assistance information is not obtained. Including more,
How to determine the suitability of a fundus image.
The method of claim 13,
Acquiring suitability information of the fundus image based on the determination result; Include more,
Acquiring the suitability information,
If the target disease is the first disease and the fundus image meets the first quality criterion, obtaining the suitability information indicating that the fundus image is suitable for the first disease,
If the subject disease is the second disease and the fundus image meets the second quality criterion, obtaining the suitability information indicating that the fundus image is suitable for the second disease
How to determine the suitability of a fundus image.
The method of claim 13,
Target diseases include glaucoma, cataracts, bleeding, drusen, exudate, cotton patch, cloudiness, retinal pigment change, macular perforation, ocular diseases including macular degeneration and prematurity retinopathy, systemic diseases including diabetic, Alzheimer's, cytomegalovirus, Any one selected from cardiovascular diseases including hypertension, hypotension, stroke, atherosclerosis, angina pectoris and heart failure,
How to determine the suitability of a fundus image.
In the fundus image management apparatus for determining the suitability of the fundus image in connection with the diagnosis of the target disease based on the fundus image,
A fundus image obtaining unit obtaining the fundus image; And
A suitability determination unit configured to determine a target suitability of the fundus image with respect to the fundus image using a quality standard corresponding to the target disease, wherein the target disease includes at least one of a first disease and a second disease; Including,
The suitability determination unit determines whether the fundus image meets a first quality criterion when the target disease is a first disease, and determines whether the fundus image meets a second quality standard when the target disease is a second disease.
Fundus image management device.
The method of claim 24,
The quality criteria include quality criteria for the target area,
The fitness determining unit determines the object suitability of the fundus image to determine whether the first region of the fundus image satisfies the first quality criterion when the target disease is the first disease, and wherein the target disease is the second disease. And determining whether the second region of the fundus image satisfies the second quality criterion.
Fundus image management device.
The method of claim 24,
The quality criteria include quality criteria for the presence or absence of artifacts,
The suitability determining unit, when the target disease is the first disease, determining the first quality criterion includes determining whether the fundus image includes a first artifact corresponding to the first disease, In the case of a second disease, determining the second quality criterion further includes determining whether the fundus image includes a second artifact corresponding to the second disease,
Fundus image management device.
The method of claim 26,
A first artifact corresponding to the first disease is detected from a first region associated with the first disease, and a second artifact corresponding to the second disease is detected from a second region associated with the second disease,
Fundus image management device.
The method of claim 24,
Further comprising: a diagnostic information acquisition unit for obtaining diagnostic information on the target disease based on the fundus image,
When the target disease is the first disease, the diagnostic information obtaining unit obtains first diagnostic assistance information for the first disease when the fundus image satisfies the first quality criterion, and the target disease is the second disease. In the case of a disease, when the fundus image satisfies the second quality criterion, the second diagnosis assistance information regarding the second disease is acquired.
Fundus image management device.
The method of claim 24,
A suitability information acquisition unit obtaining suitability information of the fundus image based on the determination result; Include more,
The fitness information acquiring unit acquires the fitness information indicating that the fundus image is suitable for the first disease when the target disease is the first disease and the fundus image satisfies the first quality criterion. Obtaining the suitability information indicating that the fundus image is suitable for the second disease when the target disease is the second disease and the fundus image satisfies the second quality criterion.
Fundus image management device.
The method of claim 29,
The subject disease includes the first disease and the second disease,
The suitability determination unit determines whether the fundus image satisfies the first quality criterion and the second quality criterion,
The fitness information acquiring unit, when it is determined that the fundus image satisfies the first quality criterion but does not satisfy the second quality criterion, the fundus image is suitable for the first disease and for the second disease. Acquiring the suitability information indicating incompatibility;
Fundus image management device.
In the fundus image management apparatus for determining the suitability of the fundus image in connection with the diagnosis of the target disease based on the fundus image,
A fundus image obtaining unit obtaining the fundus image;
A quality determination unit determining the quality of the fundus image; And
And a suitability determination unit configured to determine a target suitability for the target disease of the fundus image based on determining whether the fundus image satisfies the quality criteria.
The quality determining unit determines whether the first quality criterion related to the first area is satisfied, and whether the fundus image satisfies a second quality criterion related to the second area that is at least partially distinguished from the first area. Judge,
The suitability determination unit determines that the fundus image satisfies the first suitability when the target disease is the first disease and the fundus image satisfies the first quality criterion, and the target disease is the second disease and the fundus Determining that the fundus image satisfies a second suitability if the image satisfies the second quality criterion,
Fundus image management device.
The method of claim 31, wherein
The quality determining unit determines whether there is a first artifact in consideration of the first area, and determines whether there is a second artifact in consideration of the second area.
Fundus image management device.
The method of claim 31, wherein
The quality determining unit determines whether the fundus image includes the first artifact by determining whether the fundus image includes the first artifact, and determines whether the fundus image satisfies the second quality standard by determining whether the fundus image includes the second artifact. ,
Fundus image management device.
The method of claim 31, wherein
The first area is an area spaced a predetermined distance or more from the center of the fundus area in which the fundus of the fundus image is distributed,
The second area is an area within a predetermined distance from the center of the fundus area.
Fundus image management device.
The method of claim 31, wherein
A diagnostic information obtaining unit obtaining diagnostic information on the target disease based on the fundus image; Include more,
The diagnosis information obtaining unit, when the target disease is the first disease, acquires first diagnostic assistance information for the first disease when the fundus image satisfies the first quality criterion, and wherein the target disease is the first disease. In the case of a second disease, when the fundus image satisfies the second quality criterion, the second diagnosis assistance information regarding the second disease is acquired.
Fundus image management device.
The method of claim 35, wherein
The subject disease comprises the first disease and the second disease,
The diagnosis information acquiring unit, when the fundus image satisfies the first quality criterion and does not satisfy the second quality criterion, obtains first diagnostic assistance information for the first disease, and second diagnostic assistance information. Not acquiring,
Fundus image management device.
A computer-readable recording medium having recorded thereon a program for executing the method of any one of claims 1 to 23.

Images