KR20220036475A - System, method and application for verification and presentation of strabismus test results - Google Patents

Abstract

The present invention relates to a system and method that verifies and proposes a strabismus examination result through machine learning. Specifically, the present invention provides the system and method that, when the strabismus examination result is obtained, allow the obtained strabismus examination result to be pre-processed to perform CNN or DNN machine learning using the pre-processed data, compare and verify the two machine learning results, and analyze the new strabismus examination result, using the verified machine learning result, to output the diagnostic information. Therefore, the present invention is capable of having an effect of providing the system and method that can propose quickly and accurately.

Description

System, method and application for verification and presentation of strabismus test results

The present invention relates to a strabismus examination result strabismus examination result verification presentation system and method, and an application therefor. More specifically, a strabismus examination result verification presentation system and method for learning and verifying strabismus examination results using machine learning and performing a diagnosis on the strabismus examination results using artificial intelligence based on the learning results, and an application therefor is about

Six muscles, represented by the internal rectus, external rectus, superior rectus, inferior rectus, superior superior and inferior oblique, are connected to the human eye. Strabismus is a disease caused by the occurrence of abnormalities in at least one of these muscles. Strabismus surgery is performed to find the muscles that cause strabismus among these muscles and to normalize them by weakening or strengthening the eye muscles in the direction in which the eyes are straight. can

In general, a lot of retrospective surgery is performed to weaken the muscle strength by removing the muscle that moves the eye from its original position and attaching it a little further back than the original position. This is also implemented.

Before performing these surgeries, the degree of eye rotation, that is, the process of obtaining the angle of deviation is required. Currently, corneal reflection and prism reflection methods exist as methods to obtain the angle of deviation, but since they are all performed manually by a doctor without digital measuring equipment, it is difficult to obtain an accurate and objective angle of deviation.

In addition, although strabismus examination results require interpretation by a trained pediatric neuroophthalmologist, there is a problem in interpreting the examination results because ophthalmologists of other majors who do not have continuous training for accurate diagnosis are not familiar with the results.

In addition, there are cases where the results of the strabismus examination reflect the presence of brain lesions rather than ending with a simple visual acuity, and if this is missed, there is a problem that can lead to serious medical results.

In order to solve the above problems, the present invention intends to propose a strabismus examination result verification presentation system and method capable of performing fast and accurate artificial intelligence reading so as not to require interpretation by a trained pediatric neuroophthalmologist, and an application therefor.

In addition, the present invention intends to propose a strabismus examination result verification presentation system and method that can increase reliability by using a process of comparing and verifying machine learning with each other, and an application therefor.

In embodiments of the present invention, in a strabismus examination result verification presentation system and method and an application therefor, an examination result acquisition unit that obtains an examination result of a strabismus examination performed by a user and pre-processes the examination result to obtain a pre-processed examination result ; an inspection result learning unit that acquires the pre-processed inspection result and performs machine learning based on image information or coordinate information; and a diagnosis result providing unit that generates and compares diagnosis information using the results of the machine learning, selects any one of the diagnosis information as a diagnosis result, and provides it to an administrator, wherein the examination result includes the examination background and examination It is possible to provide a system for presenting verification of a strabismus examination result formed in an image.

Embodiments of the present invention include the steps of obtaining an examination result of a strabismus examination performed by a user using an examination result obtaining unit and pre-processing the examination result to obtain a pre-processed examination result, and using the examination result learning unit to obtain the pre-processed examination result A step of acquiring and performing machine learning based on image information or coordinate information, and generating and comparing diagnostic information using the results of machine learning using a diagnostic result providing unit, selecting any one of the diagnostic information as a diagnosis result, and selecting an administrator It is possible to provide a method for presenting strabismus examination results, including the step of providing to the

In the strabismus examination result verification presentation method according to embodiments of the present invention, the examination result may be formed of an examination background and an examination image.

Embodiments of the present invention include the steps of obtaining an examination result of a strabismus examination performed by a user, pre-processing the examination result, and obtaining an examination result formed of an examination background and examination image, and acquiring the pre-processed examination result to obtain image information or In order to execute the steps of performing machine learning based on the coordinate information, and generating and comparing diagnostic information using the results of machine learning, selecting any one of the diagnostic information as a diagnostic result and providing it to the administrator An application stored in the medium may be provided.

The present invention described above has the effect of providing a system and method capable of quickly and accurately presenting strabismus examination results.

In addition, the present invention has an effect of providing a system and method capable of obtaining an accurate diagnosis result even if an unskilled person is located for analyzing the examination result by analyzing the examination result using artificial intelligence and presenting the diagnosis result.

1 is a diagram simply showing a strabismus examination result verification presentation system according to an embodiment of the present invention.
2 is a simple block diagram illustrating the configuration of (a) an examination result acquisition unit, (b) an examination result learning unit, and (c) a diagnosis result providing unit in the strabismus examination result verification presentation system according to an embodiment of the present invention.
3 is a view showing an example of one result of (a) Lancaster test and (b) Hess test.
4 is a diagram illustrating (a) a flowchart illustrating a method for presenting verification of a strabismus examination result according to an embodiment of the present invention, and (b) a diagram visualizing some steps included in the flowchart.

Hereinafter, some embodiments of the present invention will be described in detail with reference to exemplary drawings. In adding reference numerals to the components of each drawing, it should be noted that the same components are given the same reference numerals as much as possible even though they are indicated on different drawings. In addition, in describing the present invention, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present invention, the detailed description thereof will be omitted.

1 is a diagram simply showing a strabismus examination result verification presentation system according to an embodiment of the present invention, and FIG. 2 is (a) an examination result obtaining unit ( It is a block diagram showing the configuration of b) a test result learning unit and (c) a diagnosis result providing unit, and FIG. 3 is a result of one of (a) Lancaster test and (b) Hess test. It is a drawing showing an example.

Hereinafter, the strabismus examination result verification presentation system 120 according to an embodiment of the present invention will be described in detail with reference to FIGS. 1 to 3 . At this time, the strabismus examination result verification presentation system 120 to be described below is only one core example of the present invention, and it is evident that various accessory configurations that can be changed and added are included in the scope of the present invention as needed. .

Referring to FIG. 1 , the strabismus examination result verification presentation system 120 according to an embodiment of the present invention is connected to the strabismus examination terminal 110 and the diagnosis terminal 130 using wireless or wired communication.

The strabismus examination terminal 110 performs a strabismus examination of the user, and transmits the performed strabismus examination result to the strabismus examination result verification presentation system 120 .

The diagnosis terminal 130 may obtain a diagnosis result from the strabismus examination result verification presentation system 120 and allow an administrator to check the obtained diagnosis result.

Here, the strabismus examination result verification and presentation system 120 of the present invention obtains the strabismus examination result, pre-processes it, performs machine learning, compares the diagnostic information generated using the result, selects the diagnosis result, and selects the selected diagnosis result is formed to provide managers with

To this end, the strabismus examination result verification presentation system 120 of the present invention may include an examination result obtaining unit 121 , an examination result learning unit 122 , and a diagnosis result providing unit 123 .

The examination result obtaining unit 121 is formed to obtain the examination result of the strabismus examination performed by the user and pre-process the examination result to obtain the pre-processed examination result, the examination result obtaining module 211 and the examination result pre-processing module 212 ) may be included.

The examination result obtaining module 211 receives the examination result, which is the isometric examination result, from the isometric examination terminal 110 connected through a wired or wireless communication network. At this time, the strabismus test terminal 110 performs a) a Lancaster test or b) a Hess test as in the example shown in FIG. 3 to obtain a strabismus test result, and performs two tests. You can do all of them.

For this reason, the test result transmitted from the test result acquisition module 211 may be formed, for example, by including the test image on the test background line, which can be described as a graph paper.

The inspection result pre-processing module 212 is configured to receive the inspection result and perform pre-processing to obtain the pre-processed inspection result. The inspection result pre-processing module 212 may generate image information or coordinate information as the pre-processed inspection result, and for this purpose, image pre-processing or coordinate pre-processing may be performed on the inspection result.

Image preprocessing is a preprocessing performed on an inspection result in order to machine-learning the inspection result into an image state. In order to perform image pre-processing, the inspection result pre-processing module 212 removes the inspection background line constituting the inspection result and the inspection background line among the inspection images, standardizes the size of the inspection image, and uses center coordinates and border information to Acquire the relative coordinates of the inspection image. Thereafter, the inspection result preprocessing module 212 may generate image information including a standardized inspection image and relative coordinates to end the image preprocessing process.

In this case, the inspection result preprocessing module 212 detects an outer edge of a rectangle included in the inspection result in order to standardize the size of the inspection image in the image preprocessing process, and adjusts the scale of the inspection image to a fixed size based on the outer edge. can be converted to

In addition, the inspection result preprocessing module 212 obtains the representative coordinates to obtain the relative coordinates. To this end, the inspection result preprocessing module 212 detects the outer edge of the rectangle included in the inspection result, extracts central coordinates from the detected edge information, and acquires 24 coordinates at preset intervals based on the central coordinates, , it is possible to obtain the coordinate margin, which is the distance between the 25 coordinates including the center coordinate and the border information.

This is to check how much the inspection image is deformed by acquiring a coordinate margin, which is a distance between each coordinate and edge information, because 25 coordinates are initially located on the line of the reference image presented as the inspection standard.

The coordinate preprocessing is a preprocessing performed on the inspection result in order to machine learning the inspection result in the form of coordinates. In order to perform the coordinate preprocessing, the inspection result preprocessing module 212 detects an outer edge of a rectangle constituting the inspection result, extracts central coordinates from the detected edge information, and a straight line extending radially in at least three directions from the central coordinates. , obtain coordinates of a point where a straight line and a border meet, and detect 8 vertices to obtain at least 13 representative coordinates.

When at least 13 representative coordinates are obtained, the inspection result preprocessing module 212 may generate coordinate information including a distance and an angle of each representative coordinate from the central coordinates.

The inspection result learning unit 122 is formed to perform machine learning based on image information or coordinate information by acquiring pre-processed inspection results, and includes an image-based learning module 221 and a coordinate-based learning module 222 for this purpose can be formed to

The image-based learning module 221 performs machine learning on the inspection result using image information. The image-based learning module 221 preferably uses a convolutional neural network (CNN) algorithm to perform machine learning of image information, and can obtain an image learning result using a CNN algorithm. there is.

The coordinate-based learning module 222 performs machine learning on the inspection result by using the coordinate information. The coordinate-based learning module 222 may preferably use a deep neural network (DNN) algorithm to perform machine learning of coordinate information, and obtain a coordinate learning result using the DNN algorithm. .

The diagnosis result providing unit 123 generates and compares diagnosis information using the results of machine learning, selects any one of the diagnosis information as a diagnosis result, and transmits the selected diagnosis result to the diagnosis terminal 130 to output it to the administrator. is formed To this end, the diagnosis result providing unit 123 may include a diagnosis information comparison module 231 and a diagnosis result selection module 232 .

The diagnostic information comparison module 231 is configured to compare two pieces of diagnostic information to generate comparison information. The diagnostic information comparison module 231 may generate diagnostic information by applying the same test result to the image learning result and the coordinate learning result.

In this case, the diagnosis information generated using the image learning result may be image diagnosis information, and the diagnosis information generated using the coordinate learning result may be expressed as coordinate diagnosis information.

The diagnosis information comparison module 231 generates comparison information by comparing the image diagnosis information and the coordinate diagnosis information. In this case, the comparison information may be a result of comparing the coincidence rate of the image diagnosis information and the coincidence rate of the coordinate diagnosis information, and the coincidence rate corresponds to the disease name included in the examination result used to generate the image diagnosis information and the coordinate diagnosis information. It can mean the degree

That is, it may be expressed that the diagnosis information comparison module 231 obtains and compares the accuracy of the image learning result and the coordinate learning result, and generates comparison information.

The diagnosis result selection module 232 is configured to select a diagnosis result and transmit it to the diagnosis terminal 130 so that an administrator can check the diagnosis result through the diagnosis terminal 130 .

The diagnosis result selection module 232 receives the comparison information generated by the diagnosis information comparison module 231 to determine which learning result has a higher matching rate, that is, higher accuracy, and diagnoses generated from the learning result. Information may be selected as a diagnosis result and transmitted to the diagnosis terminal 130 .

Meanwhile, in another embodiment of the present invention, the test result learning unit 122 may further include a virtual learning data generating module (not shown) not shown in the drawing.

In order to perform machine learning, sufficient basic data must be supplied. However, there is a possibility that a sufficient amount of basic data may not be supplied only with the actual strabismus examination result data.

Therefore, in order to smoothly perform machine learning, another embodiment of the present invention may further include a virtual learning data generating module (not shown) so that the virtual learning data generating module generates a large amount of virtual learning data. there is.

The virtual learning data generation module uses a well-known generative adversarial network (GAN) algorithm. The GAN algorithm has a generator and a discriminator, and repeats the process of determining whether the virtual data generated through the generator is data usable by the discriminator.

In this process, the creator of the virtual learning data generation module generates more sophisticated virtual data, and the discriminator can accumulate the virtual data by discriminating it as usable data.

The learnable virtual data generated in the virtual learning data generating module through this GAN algorithm is preferably image pre-processed or coordinate pre-processed inside the virtual learning data generating module, and the inspection result learning unit 122 described above as image information or coordinate information can be transmitted to

Meanwhile, FIG. 4 is a diagram illustrating (a) a flowchart illustrating a method for presenting verification of a strabismus examination result and (b) a diagram visualizing some steps included in the flowchart according to an embodiment of the present invention.

Hereinafter, a method for presenting verification of a strabismus examination result according to an embodiment of the present invention will be described in detail with reference to FIGS. 3 and 4 . Hereinafter, for convenience of explanation, the structure of the strabismus examination result verification presentation system described with reference to FIGS. 1 to 3 will be used, but the present invention is not necessarily limited thereto. It is self-evident that it can also be used in devices.

Referring to FIG. 4A , the strabismus examination result verification presentation method 400 according to an embodiment of the present invention comprises: obtaining a strabismus examination result, performing pre-processing ( S410 ), obtaining the pre-processed examination result, and machine learning It is formed to include a step of performing (S420) and a step of selecting a diagnostic result by generating and comparing diagnostic information (S430).

First, in the strabismus examination result verification presentation method 400, a strabismus examination result is obtained using the examination result acquisition unit and pre-processing is performed (step S410). Step S410 is to obtain the examination result of the strabismus examination performed by the user by using the examination result acquisition unit and pre-process the examination result to obtain the pre-processed examination result, and for this purpose, the step of obtaining the examination result (S411) and image pre-processing or It may include performing coordinate pre-processing (S412).

In step S411, a test result of the Hess test or the Lancaster test is obtained. In step S411, an examination result, which is a strabismus examination result, is transmitted through a wired or wireless communication network. In this case, the strabismus test result may be a) a Lancaster test or b) a Hess test, as in the example shown in FIG. 3 , and both test results may be obtained.

For this reason, the inspection result obtained in step S411 may be formed by including the inspection image on the inspection background line, which can be described with a graph paper, for example, and an example of this is shown as S411 in FIG. 4B .

Next, in step S412, the pre-processing is performed upon receiving the test result to obtain the pre-processed test result.

In step S412, image information or coordinate information may be generated as the pre-processed inspection result, and for this, image pre-processing or coordinate pre-processing is performed on the inspection result.

Image preprocessing is a preprocessing performed on an inspection result in order to machine-learning the inspection result into an image state. In order to perform image pre-processing, in step S412, the inspection background line constituting the inspection result and the inspection background line among the inspection images are removed, the size of the inspection image is standardized, and the relative coordinates of the inspection image are calculated using center coordinates and border information. acquire

Thereafter, in step S412, the image preprocessing process may be ended by generating image information including the standardized inspection image and relative coordinates.

In this case, in step S412, in order to standardize the size of the inspection image in the image preprocessing process, an outer edge of a rectangle included in the examination result may be detected, and the scale of the examination image may be converted to a fixed size based on the outer edge. .

In addition, in step S412, representative coordinates are acquired to further acquire relative coordinates. To this end, step S412 detects the outer edge of the rectangle included in the inspection result, extracts center coordinates from the detected edge information, obtains 24 coordinates at a preset interval based on the center coordinates, and includes the center coordinates. A coordinate margin, which is the distance between 25 coordinates and border information, can be obtained.

This is to check how much the inspection image is deformed by acquiring a coordinate margin, which is a distance between each coordinate and edge information, because 25 coordinates are initially located on the line of the reference image presented as the inspection standard.

The coordinate preprocessing is a preprocessing performed on the inspection result in order to machine learning the inspection result in the form of coordinates. In order to perform the coordinate pre-processing, step S412 detects the outer edge of the rectangle constituting the inspection result, extracts center coordinates from the detected edge information, assumes a straight line extending radially in at least three directions from the center coordinate, and a straight line Coordinates of a point where the border and the border meet are acquired, and 8 vertices are detected to obtain at least 13 representative coordinates.

When at least 13 representative coordinates are obtained in step S412, coordinate information including a distance and an angle of each representative coordinate from the central coordinate may be generated.

The image preprocessing and coordinate preprocessing of step S412 described above are shown as an example in S412 of FIG. 4B . The left table may be coordinate information for which coordinate pre-processing has been completed, and the right figure may be image information for which image pre-processing has been completed.

Next, in the strabismus examination result verification presentation method, a pre-processed examination result is obtained using a diagnosis result learning unit, and machine learning is performed based on image information or coordinate information (step S420).

Step S420 is to obtain a preprocessed inspection result using the inspection result learning unit, perform machine learning based on image information or coordinate information, and in more detail, perform image-based learning (S421) and performing coordinate-based learning (S422) is formed to include.

In the step of performing image-based learning ( S421 ), machine learning is performed on the inspection result using image information. In step S421, preferably, a convolutional neural network (CNN) algorithm is used to perform machine learning of image information, and an image learning result can be obtained using the CNN algorithm.

In the step of performing coordinate-based learning ( S422 ), machine learning is performed on the inspection result using coordinate information. In step S422, preferably, a deep neural network (DNN) algorithm is used to perform machine learning of coordinate information, and a coordinate learning result may be obtained using the DNN algorithm.

Finally, in the method for intensively presenting strabismus examination results according to an embodiment of the present invention, the diagnosis result providing unit generates and compares diagnosis information using the results of machine learning and selects the diagnosis results (step S430).

In operation S430, the diagnosis information may be generated using the diagnosis result providing unit, any one of the generated diagnosis information may be selected as a diagnosis result, and the selected diagnosis result may be output to the manager.

To this end, step S430 may include generating comparison information by comparing diagnostic information ( S431 ) and selecting a diagnosis result using the comparison information ( S432 ).

In the step (S431) of generating comparison information by comparing diagnostic information, the comparison information is generated by comparing two pieces of diagnostic information. can

In this case, the diagnosis information generated using the image learning result may be image diagnosis information, and the diagnosis information generated using the coordinate learning result may be expressed as coordinate diagnosis information.

In step S431, the comparison information is generated by comparing the image diagnosis information and the coordinate diagnosis information. In this case, the comparison information may be a result of comparing the coincidence rate of the image diagnosis information and the coincidence rate of the coordinate diagnosis information, and the coincidence rate is the image diagnosis The information and coordinates may mean a degree of matching with the disease name included in the test result used to generate the diagnostic information.

That is, step S431 may be expressed as acquiring and comparing the accuracy of the image learning result and the coordinate learning result, and generating comparison information.

Finally, in the strabismus examination result verification and presentation method of the present invention, the diagnosis result is selected by using the diagnosis result providing unit, and the diagnosis result is transmitted to the outside so that the administrator confirms the diagnosis result (step S432).

Step S432 receives the comparison information generated in step S431, checks which learning result has a higher matching rate, that is, higher accuracy, and selects the diagnostic information generated from the corresponding learning result as a diagnostic result and transmits it to the outside can

On the other hand, in another embodiment of the present invention, step S420, a step not specified in the flowchart, generating the pre-processed virtual learning data (not shown); may further include.

In order to perform machine learning, sufficient basic data must be supplied. However, there is a possibility that a sufficient amount of basic data may not be supplied only with the actual strabismus examination result data.

Therefore, in order to smoothly perform machine learning, another embodiment of the present invention includes the step of generating preprocessed virtual learning data; in the step of performing machine learning, a large amount of virtual learning data is generated in step S420. It can be created and used for machine learning.

In the step of generating the virtual learning data, a well-known generative adversarial network (GAN) algorithm is used to generate the virtual data. The GAN algorithm has a generator and a discriminator, and repeats the process of determining whether the virtual data generated through the generator is data usable by the discriminator.

In this process, the generator creates more sophisticated virtual data, and the discriminator can accumulate the virtual data by discriminating it as usable data.

The learnable virtual data generated through the GAN algorithm is generated in an image pre-processed or coordinate pre-processed state and transferred to the above-described steps S421 and S422 as image information or coordinate information, so that it can be used as basic data for machine learning.

The strabismus examination result verification presentation method according to the embodiments of the present invention described above may be implemented as an application (computer program) stored in a storage medium of a computer.

Here, the computer may include a strabismus examination result verification presentation system 120 .

The operating system of the computer is an operating system such as Windows or Macintosh installed on general PCs such as desktops and laptops, or iOS and Android installed on mobile terminals such as smartphones and tablet PCs. It may be a mobile-only operating system of

The strabismus examination result verification presentation method according to the above-described embodiments of the present invention is implemented as an application (ie, computer program) installed by default on a computer or installed by a user, and stored (recorded) in a computer-readable storage medium can be

The application (computer program) that implements the strabismus examination result verification presentation method according to the embodiments of the present invention and is stored and executed in a computer storage medium obtains the examination result of the strabismus examination performed by the user and pre-processes the examination result. Acquiring pre-processed test results formed from the test background and test image, performing machine learning based on image information or coordinate information by acquiring pre-processed test results, and generating diagnostic information using the results of machine learning A step of selecting any one of the diagnosis information as a diagnosis result by generating and comparing it and providing it to an administrator may be performed.

In this way, in order for the computer to read the program recorded in the storage medium and execute the method for presenting the strabismus examination result verification according to the present embodiments implemented as a program, the above-described application (application program) is performed by the processor (CPU) of the computer. It may include code coded in a computer language such as C, C++, JAVA, and machine language that can be read.

Such codes may include function codes related to functions defining the above-mentioned functions, etc., and may include control codes related to an execution procedure necessary for the processor of the computer to execute the above-mentioned functions according to a predetermined procedure.

In addition, this code may further include additional information necessary for the processor of the computer to execute the above functions or code related to memory reference for which location (address address) in the internal or external memory of the computer to be referenced. .

In addition, when the computer's processor needs to communicate with any other computer or server in a remote location in order to execute the above-mentioned functions, the code is transmitted to the computer's communication module (eg, wired and/or wireless communication module). ) may be used to further include communication-related codes for how to communicate with any other computer or server in the remote, and what information or media to transmit/receive during communication.

And, in consideration of the system environment of a computer that reads a storage medium and executes a program by reading a storage medium, a functional program and related codes and code segments for implementing the present embodiments are a programmer in the technical field to which the present invention pertains. may be easily inferred or changed by

In addition, the computer-readable storage medium in which the above-described program is recorded is distributed in a computer system connected by a network, so that the computer-readable code can be stored and executed in a distributed manner. In this case, any one or more of the plurality of distributed computers may execute some of the functions presented above, and transmit the result of the execution to one or more of the other distributed computers, and receive the result The computer may also execute some of the functions presented above and provide the results to other distributed computers as well.

As described above, a computer-readable storage medium recording an application for executing the isometric examination result verification presentation method according to embodiments of the present invention is recorded, for example, ROM, RAM, CD-ROM, magnetic tape. , a floppy disk, an optical media storage device, and the like.

In addition, the computer-readable storage medium in which an application, which is a program for executing the isometric examination result verification presentation method according to embodiments of the present invention, is recorded is an application store server, an application store server, an application or a web related to the corresponding service. It may be a storage medium (eg, hard disk, etc.) included in an application providing server including a server (Web Server), or it may be the application providing server itself, or it may be another computer or its storage medium in which the program is recorded. may be

A computer capable of reading a storage medium recording an application, which is a program for executing the strabismus examination result verification presentation method according to embodiments of the present invention, is not only a general PC such as a general desktop or notebook computer, but also a smart phone, a tablet PC, It may include mobile terminals such as personal digital assistants (PDA) and mobile communication terminals, and should be interpreted as all computing devices.

The above description is merely illustrative of the technical spirit of the present invention, and various modifications and variations will be possible without departing from the essential characteristics of the present invention by those skilled in the art to which the present invention pertains. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical spirit of the present invention, but to explain, and the scope of the technical spirit of the present invention is not limited by these embodiments. The protection scope of the present invention should be construed by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

110: strabismus examination terminal
120: strabismus examination result verification presentation system
121: inspection result acquisition unit 122: inspection result learning unit
123: diagnosis result providing unit 130: diagnosis terminal
211: inspection result acquisition module 212: inspection result preprocessing module
221: image-based learning module 222: coordinate-based learning module
231: diagnosis information comparison module 232: diagnosis result selection module

Claims (17)

an examination result acquisition unit for obtaining an examination result formed of an examination background and examination image of the strabismus examination performed by the user, and pre-processing the examination result to obtain a pre-processed examination result;
an inspection result learning unit that acquires the preprocessed inspection result and performs machine learning based on image information or coordinate information; and
The strabismus examination result verification and presentation system comprising a; a diagnosis result providing unit that generates and compares diagnosis information using the results of the machine learning, selects any one of the diagnosis information as a diagnosis result, and provides it to an administrator. The method of claim 1,
The inspection result obtaining unit,
Hess test or Lancaster test is used as the strabismus test,
an inspection result acquisition module configured to acquire the inspection result; and
A strabismus examination result verification presentation system comprising a; an examination result pre-processing module to obtain the pre-processed examination result by performing image pre-processing for generating the image information or coordinate pre-processing for generating the coordinate information. 3. The method of claim 2,
The image preprocessing is
A system for verifying and presenting isometric examination results by removing the examination background from the examination results, standardizing the size of the examination image, and displaying the image information including the relative coordinates of the examination image obtained using center coordinates and edge information. 3. The method of claim 2,
The coordinate preprocessing is
A strabismus examination result verification presentation system for extracting central coordinates from the examination image, and extracting a plurality of representative coordinates using the central coordinates to generate the coordinate information. 3. The method of claim 2,
The test result learning unit,
an image-based learning module that performs machine learning on the inspection result using the image information; and
Including; a coordinate-based learning module that performs machine learning on the inspection result by using the coordinate information;
The image-based learning module performs the machine learning using a CNN algorithm,
The coordinate-based learning module is a strabismus examination result verification presentation system for performing the machine learning using a DNN algorithm. 3. The method of claim 2,
The diagnosis result providing unit,
The diagnosis information is respectively generated by applying the inspection result to the image learning result, which is the machine learning result performed based on the image information, and the coordinate learning result, which is the machine learning result, which is the machine learning result performed based on the coordinate information, respectively. a diagnostic information comparison module that compares the diagnostic information of dogs and generates comparison information; and
Isometric examination result verification including; a diagnosis result selection module that selects the diagnosis information generated using either the image learning result or the coordinate learning result as the diagnosis result by using the comparison information and provides it to the administrator presentation system. 7. The method of claim 6,
The diagnostic information comparison module,
generating the comparison information by comparing the image diagnosis information that is the diagnosis information generated through the image learning result with the coordinate diagnosis information that is the diagnosis information generated through the coordinate learning result,
The comparison information is a result of comparing the coincidence rate of the image diagnosis information and the coincidence rate of the coordinate diagnosis information;
The concordance rate refers to a degree to which the image diagnosis information and the coordinate diagnosis information match the predicted disease name of the examination result obtained through the machine learning, respectively. 8. The method according to any one of claims 2 to 7,
The test result learning unit,
It further includes; a virtual learning data generation module for generating virtual learning data similar to the test result by using a generative adversarial network (GAN),
The virtual learning data generation module,
A strabismus examination result verification presentation system for generating the virtual learning data as data on which the image pre-processing and the coordinate pre-processing have been performed. In a strabismus examination result verification presentation method using a strabismus examination result verification presentation system,
obtaining an examination result formed of an examination background and examination image of a strabismus examination performed by a user, and pre-processing the examination result to obtain a pre-processed examination result;
performing machine learning based on image information or coordinate information by obtaining the pre-processed inspection result; and
The step of generating and comparing diagnostic information using the results of the machine learning, selecting any one of the diagnostic information as a diagnostic result, and providing it to an administrator; 10. The method of claim 9,
The step of obtaining the pre-processed test result is,
Hess test or Lancaster test is used as the strabismus test,
obtaining the test result; and
Obtaining the pre-processed test result by performing image pre-processing for generating the image information or coordinate pre-processing for generating the coordinate information; 11. The method of claim 10,
The image preprocessing is
A method for verifying and presenting isometric examination results by removing the examination background from the examination results, standardizing the size of the examination image, and presenting the image information including the relative coordinates of the examination image obtained using center coordinates and edge information. 11. The method of claim 10,
The coordinate preprocessing is
A method for presenting isometric examination results by extracting central coordinates from the examination image and extracting a plurality of representative coordinates using the central coordinates to generate the coordinate information. 11. The method of claim 10,
The step of performing the machine learning is,
performing image-based learning of performing machine learning on the inspection result using the image information; and
Including; performing coordinate-based learning of performing machine learning on the inspection result by using the coordinate information;
The step of performing the image-based learning, performing the machine learning using a CNN algorithm,
The performing of the coordinate-based learning is a method for presenting a strabismus examination result verification and performing the machine learning using a DNN algorithm. 11. The method of claim 10,
The step of selecting the diagnosis result is
The diagnosis information is respectively generated by applying the inspection result to the image learning result, which is the machine learning result performed based on the image information, and the coordinate learning result, which is the machine learning result, which is the machine learning result performed based on the coordinate information, respectively. generating comparison information by comparing the diagnosis information of dogs; and
Selecting the diagnosis information generated using either the image learning result or the coordinate learning result using the comparison information as the diagnosis result and providing the diagnosis information to the manager; 15. The method of claim 14,
The step of generating the comparison information comprises:
generating the comparison information by comparing the image diagnosis information that is the diagnosis information generated through the image learning result with the coordinate diagnosis information that is the diagnosis information generated through the coordinate learning result,
The comparison information is
It is a result of comparing the coincidence rate of the image diagnosis information and the coincidence rate of the coordinate diagnosis information,
The match rate is
A strabismus examination result verification and presentation method, which means the degree to which the image diagnosis information and the coordinate diagnosis information correspond to the expected disease name of the examination result obtained through the machine learning, respectively. 16. The method according to any one of claims 10 to 15,
performing the machine learning;
It further comprises; generating virtual learning data similar to the test result using a generative adversarial network (GAN);
The step of generating the virtual learning data includes:
A strabismus examination result verification presentation method for generating the virtual learning data as data on which the image pre-processing and the coordinate pre-processing have been performed. obtaining an examination result of a strabismus examination performed by a user, pre-processing the examination result, and obtaining an examination result formed of an examination background and examination image;
performing machine learning based on image information or coordinate information by obtaining the pre-processed inspection result; and
An application stored in a storage medium of a computer to execute; generating and comparing diagnostic information using a result of the machine learning, selecting any one of the diagnostic information as a diagnosis result, and providing it to an administrator.

Images