KR20230057489A - Method and apparatus for training a diagnostic module for quantitative analysis of emphysema - Google Patents

Abstract

The present disclosure relates to a learning method of a diagnostic module for quantitative analysis of emphysema and a device therefor. According to an embodiment of the present disclosure, the method includes the steps of: obtaining at least one training medical image generated corresponding to a continuous volume of a body part including a chest; and using a preprocessing module including at least one learned network function, extracting a lung area from the training medical image, correcting the volume and brightness of the lung area, and preprocessing the training medical image.

Description

Diagnostic module learning method for quantitative analysis of emphysema and device therefor

The technical idea of the present disclosure relates to a learning method of a diagnostic module for quantitative analysis of emphysema and an apparatus therefor.

Unlike X-ray imaging, computed tomography (CT) obtains a cross-sectional image of the human body cut horizontally. Compared to simple X-ray imaging, CT has the advantage of being able to see structures and lesions more clearly because structures are less overlapped. In addition, CT imaging is a basic examination method when there is a need for precise examination when lesions are suspected in most organs and diseases because the examination cost is cheaper and the examination time is shorter than that of MRI.

Recently, with the spread of multi-channel CT (MDCT, multi-detector CT), images can be reconstructed after imaging to freely obtain desired cross-sectional and three-dimensional (3D) images like MRI, and CT scan images can also be collected at medical institutions. there is.

On the other hand, chronic obstructive pulmonary disease (COPD) refers to various clinical symptoms in which the alveoli inside the lungs are permanently damaged, and representatively includes emphysema, in which tissues constituting the alveoli are destroyed and do not function normally. Recently, in order to quantitatively diagnose such emphysema, a technology has emerged that calculates the contrast of a CT scan image with software and automatically/semi-automatically analyzes an area below a specific contrast value as an emphysema area.

However, the contrast threshold for detecting the part corresponding to emphysema varies depending on the image reconstruction algorithm or the characteristics of the imaging equipment. If the threshold is too low, the surrounding tissue or the background of the image may be recognized together. Conversely, if it is too high, the part where emphysema has occurred is not normally recognized, making it difficult to trust the diagnostic value. In addition, even if the threshold value is within an appropriate range, it is difficult to achieve effective quantification because the diagnostic value varies depending on the threshold value.

In order to solve this problem, a method of automatically setting the contrast threshold using a statistical or learning method has emerged, but the contrast of the CT scan image or the size of the lungs has been changed depending on the characteristics of the imaging equipment or the algorithm for reconstructing the image. Since it is different, there is a problem that learning is not consistently performed and it is difficult to expect quantitative digitization to be performed accurately.

An object of the present disclosure is to provide a learning method of a diagnostic module for quantitative analysis of emphysema and an apparatus therefor to solve the above problems.

The technical tasks to be achieved by the learning method of the diagnostic module for quantitative analysis of emphysema according to the technical idea of the present disclosure and the apparatus therefor are not limited to the above-mentioned technical tasks, and other technical tasks not mentioned are as follows It will be clear to those skilled in the art from the descriptions.

According to one aspect of the technical concept of the present disclosure, a learning method of a diagnostic module for quantitative analysis of emphysema includes at least one training medical image generated corresponding to a continuous volume of a body part including the chest. obtaining; and preprocessing the learning medical image by extracting a lung region from the learning medical image and correcting the volume and brightness of the lung region using a preprocessing module including at least one learned network function. can

According to an exemplary embodiment, the preprocessing may include extracting a lung region from the training medical image; segmenting a left lung region and a right lung region from the training medical image from which the lung region is extracted; adjusting the volume of the left lung region and the right lung region within a predetermined range; and converting the brightness of the lung region whose volume is adjusted within a predetermined range through histogram normalization.

According to an exemplary embodiment, the method may further include labeling a severity class of emphysema in the training medical image for which the preprocessing is completed.

According to an exemplary embodiment, the severity class may be classified into normal, mild, moderate, and severe.

According to an exemplary embodiment, the method may further include learning at least one network function of the diagnostic module based on the labeled training medical images.

According to one aspect of the technical idea of the present disclosure, an apparatus for learning a diagnostic module for quantitative analysis of emphysema includes at least one processor; and a memory storing a program executable by the processor, wherein the processor acquires at least one training medical image generated corresponding to the continuous volume of the body part including the chest by executing the program, , By using a pre-processing module including at least one learned network function, the lung region is extracted from the training medical image, and the volume and brightness of the lung region are converted within a predetermined range to pre-process the training medical image. there is.

According to a method for learning a diagnostic module for quantitative analysis of emphysema and an apparatus therefor according to embodiments according to the technical idea of the present disclosure, a medical image for learning is preprocessed to be converted into a certain size and brightness range, and then diagnosed based thereon. Since the module is learned, the learning result of the diagnosis module can be prevented from being affected by the size and shape of the lungs of each patient, and the type of image acquisition equipment or image composition algorithm.

According to the diagnostic module learning method and apparatus for quantitative analysis of emphysema according to embodiments according to the technical idea of the present disclosure, when generating learning data of the diagnostic module, the specialist only needs to determine and label the severity class, The efficiency of labeling is increased so that a large amount of training data can be efficiently acquired.

Effects obtainable by the learning method of the diagnostic module for quantitative analysis of emphysema and the apparatus therefor according to embodiments according to the technical idea of the present disclosure are not limited to the above-mentioned effects, and other effects not mentioned are From the description below, it will be clear to those skilled in the art to which this disclosure belongs.

A brief description of each figure is provided in order to more fully understand the figures cited in this disclosure.
1 is a flowchart of a method for quantitative analysis of emphysema according to an embodiment of the present disclosure.
2 is a flowchart of a learning method of a diagnostic module for quantitative analysis of emphysema according to an embodiment of the present disclosure.
3 is a flowchart of a preprocessing step according to an embodiment of the present disclosure.
4 is a diagram for explaining a preprocessing step according to an embodiment of the present disclosure.
5 and 6 are views for explaining a step of learning at least one network function of a diagnosis module according to an embodiment of the present disclosure.
7 is a block diagram of a device according to an embodiment of the present disclosure.

Since the technical spirit of the present disclosure may be subject to various changes and may have various embodiments, specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the technical spirit of the present disclosure to specific embodiments, and should be understood to include all changes, equivalents, or substitutes included in the scope of the technical spirit of the present disclosure.

In describing the technical idea of the present disclosure, if it is determined that a detailed description of a related known technology may unnecessarily obscure the subject matter of the present disclosure, the detailed description will be omitted. In addition, numbers (eg, first, second, etc.) used in the description process of the present disclosure are only identifiers for distinguishing one component from another component.

In addition, in the present disclosure, when one component is referred to as "connected" or "connected" to another component, the one component may be directly connected or directly connected to the other component, but in particular Unless otherwise described, it should be understood that they may be connected or connected via another component in the middle.

In addition, terms such as "~ unit", "~ group", "~ character", and "~ module" described in the present disclosure mean a unit that processes at least one function or operation, which includes a processor, a micro Processor (Micro Processor), Micro Controller, CPU (Central Processing Unit), GPU (Graphics Processing Unit), APU (Accelerate Processor Unit), DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array) may be implemented by hardware or software or a combination of hardware and software.

In addition, it is intended to make it clear that the classification of components in the present disclosure is merely a classification for each main function in charge of each component. That is, two or more components to be described below may be combined into one component, or one component may be divided into two or more for each more subdivided function. In addition, each component to be described below may additionally perform some or all of the functions of other components in addition to its main function, and some of the main functions of each component may be performed by other components. Of course, it may be dedicated and performed by .

Hereinafter, embodiments of the present disclosure will be described in detail in turn.

Throughout this specification, network function may be used interchangeably with neural network and/or neural network. Here, a neural network (neural network) may be composed of a set of interconnected computational units, which may be generally referred to as nodes, and these nodes may be referred to as neurons. A neural network is generally composed of a plurality of nodes. Nodes constituting a neural network may be interconnected by one or more links.

Some of the nodes constituting the neural network may form one layer based on distances from the first input node. For example, a set of nodes having a distance of n from the first input node may constitute n layers.

The neural network described herein may include a deep neural network (DNN) including a plurality of hidden layers in addition to an input layer and an output layer.

1 is a flowchart of a method for quantitative analysis of emphysema according to an embodiment of the present disclosure.

The method 100 according to an embodiment of the present disclosure may be performed in a personal computer having computing capability, a workstation, a computer device for a server, or a separate device for this purpose.

Additionally, method 100 may be performed on one or more computing devices. For example, at least one or more steps of the method 100 according to an embodiment of the present disclosure may be performed in a client device and other steps may be performed in a server device. In this case, the client device and the server device may be connected through a network to transmit and receive calculation results. Alternatively, method 100 may be performed by distributed computing technology.

Also, an apparatus for performing the method 100 may include at least one module for performing each step. For example, each module may be implemented by a processor, a communication unit and/or a memory, or implemented to operate in conjunction therewith, and may be implemented to operate by at least one learned network function, but is not limited thereto. . In an embodiment, the device may include a preprocessing module and a diagnosis module, but is not limited thereto and may additionally include various types of modules such as a learning module.

In step S110, the device may obtain a medical image generated corresponding to the continuous volume of the body part including the chest. In an embodiment, the medical image may be a computed tomography (CT) image generated by tomography of a subject's chest. That is, the medical image may consist of a plurality of slices obtained by continuously photographing the chest of a subject in one direction through a computed tomography method. Also, in an embodiment, the medical image may be a 3D image of the head and neck generated by stacking a plurality of slices constituting a CT image.

In step S120, the device may pre-process the medical image using a pre-processing module including at least one learned network function. In this case, for example, the preprocessing module may preprocess the medical image by extracting a lung region from the medical image and correcting volume and brightness of the lung region. Accordingly, quantitative diagnosis may not be affected by the size and shape of lungs different for each patient, and the type of image acquisition equipment or image composition algorithm.

In step S130, the device may generate diagnostic information about emphysema using a diagnostic module including at least one learned network function. In this case, for example, the diagnosis module may generate diagnosis information about emphysema by detecting emphysema regions in the medical image and calculating a severity class of emphysema based on the number and ratio of emphysema regions.

In an embodiment, the network function of the diagnosis module may be learned based on a plurality of learning data each labeled with a severity class. In this case, the severity class may be classified into normal, mild, moderate, and severe. Details regarding this will be described later in the description of FIG. 2 .

Depending on embodiments, the method 100 may further include matching diagnostic information to the medical image and displaying the diagnostic information. In this case, the diagnostic information matched to the medical image and displayed may include at least one of the number, ratio, and severity class of emphysema regions. Specifically, for example, emphysema areas are displayed in a 3D medical image in a predetermined manner (specific color, shade, border, etc.), and in addition, the number, ratio, and severity class of emphysema areas are displayed in the form of text, etc. It can be provided to the diagnostician. However, it is not limited thereto.

2 is a flowchart of a learning method of a diagnostic module for quantitative analysis of emphysema according to an embodiment of the present disclosure.

The method 200 according to an embodiment of the present disclosure may be performed in a personal computer having computing capability, a workstation, a computer device for a server, or a separate device for this purpose.

Additionally, method 200 may be performed on one or more computing devices. For example, at least one or more steps of the method 200 according to an embodiment of the present disclosure may be performed in a client device and other steps may be performed in a server device. In this case, the client device and the server device may be connected through a network to transmit and receive calculation results. Alternatively, method 200 may be performed by distributed computing technology.

Also, an apparatus for performing the method 200 may include at least one module for performing each step. For example, each module may be implemented by a processor, a communication unit and/or a memory, or implemented to operate in conjunction therewith, and may be implemented to operate by at least one learned network function, but is not limited thereto. . In an embodiment, the device may include a preprocessing module, but is not limited thereto and may additionally include various types of modules such as a learning data generation module.

In step S210, the device may acquire at least one training medical image generated corresponding to the continuous volume of the body part including the chest. In an embodiment, the medical image for learning may be a computed tomography (CT) image generated by tomography of a subject's chest. That is, the medical image for learning may be composed of a plurality of slices obtained by sequentially photographing the chest of an object in one direction through a computed tomography method. Also, in an embodiment, the medical image for learning may be a 3D image of the head and neck generated by stacking a plurality of slices constituting a CT image.

In step S220, the device may pre-process the training medical image using a pre-processing module including at least one learned network function. In this case, for example, the preprocessing module may preprocess the learning medical image by extracting a lung region from the learning medical image and correcting the volume and brightness of the lung region. Accordingly, the learning result of the diagnostic module may not be affected by the size and shape of lungs different for each patient, and the type of image acquisition equipment or image composition algorithm.

In step S230, the device may label the severity class of emphysema in the preprocessed training medical images. Through this, learning data for learning the diagnosis module may be generated, where the severity class may be described in the same way as described in the method 100 . At this time, for example, the determination of the severity class for labeling may be performed by a specialist. In this case, since the specialist only needs to determine the severity class when generating the training data of the diagnostic module, the efficiency of labeling increases and a large amount of training data can be efficiently acquired. However, it is not limited thereto.

In step S240, the device may learn at least one network function of the diagnosis module based on the labeled training medical images (ie, training data). A plurality of learning data can be generated and used to learn the network function, and the accuracy of diagnosis can be improved as the number of learning data used to learn the diagnosis module increases.

3 is a flowchart of a preprocessing step according to an embodiment of the present disclosure, and FIG. 4 is a diagram for explaining a preprocessing step according to an embodiment of the present disclosure.

The method 300 is for performing step S120 of the method 100 or step S220 of the method 200, and the medical images described below include the medical image of the method 100 and the medical image for learning of the method 200. It is a concept encompassing, and the device is a concept encompassing a device for performing the method 100 and a device for performing the method 200 . Additionally, method 300 may be performed by a preprocessing module of method 100 or a preprocessing module of method 200 .

In step S310, the device may extract a lung region from the medical image. Step S310 may be performed by, for example, the first network function included in the preprocessing module, but is not limited thereto.

In step S320, the device may segment the left lung region and the right lung region in the medical image from which the lung region is extracted. Through this, even if the contrast or volume of the left lung region and the right lung region are different in one medical image, it is possible to correct the contrast and volume within a predetermined range. Step S320 may be performed by, for example, a second network function included in the preprocessing module, but is not limited thereto.

In step S330, the device may adjust the volume of the left lung region and the right lung region within a predetermined range. This is to maintain a uniform learning result or diagnosis result by normalizing the size of the lung area within a certain size range regardless of the size and shape of the different lungs for each patient. Step S330 may be performed by, for example, a third network function included in the preprocessing module, but is not limited thereto.

In step S340, the device may convert the brightness of the lung region whose volume is adjusted within a predetermined range through histogram normalization. This is to maintain a uniform learning result or diagnosis result by normalizing the contrast of a medical image within a predetermined range regardless of the type of image acquisition equipment or image composition algorithm. Step S340 may be performed by, for example, the fourth network function included in the preprocessing module, but is not limited thereto.

Meanwhile, in FIG. 3 , steps S310 to S340 are shown to be sequentially performed, but according to embodiments, at least some of steps S310 to S340 may be simultaneously performed or the order of performing them may be changed. For example, after step S310 is performed, step S340 is performed, and then steps S320 and S330 are performed.

5 and 6 are views for explaining the operation of a diagnostic module according to an embodiment of the present disclosure. Specifically, FIG. 5 illustratively illustrates a process of labeling a medical image 510 for learning with a severity class to generate learning data and then learning the diagnosis module 520 based on the learning data. FIG. ) into the learned diagnosis module 620 to output diagnosis information including a severity class.

Referring to FIG. 5 , a true/false value for each severity class may be input as a label to the training data. For example, as shown in FIG. 5 , 0 may be input for Normal, Mild, and Severe, and 1 may be input for Moderate. That is, the specialist diagnoses severity classes classified into Normal, Mild, Moderate, and Severe for the training medical image 510 and inputs True/False values as labels. . However, this is merely exemplary, and the technical spirit of the present disclosure is not limited thereto.

Referring to FIG. 6 , when a medical image is input to the diagnosis module 620 learned through a plurality of learning data, a probability for each severity class may be output. For example, as shown in FIG. 6 , when a medical image is input to the diagnosis module 620, Normal 0.012, Mild 0.041, Severe 0.135, and Moderate 0.812 are respectively output. It can be. In this case, since the probability that the severity is moderate is 81.2%, the severity can be diagnosed as moderate. However, this is merely exemplary, and the technical spirit of the present disclosure is not limited thereto.

7 is a block diagram of a device according to an embodiment of the present disclosure.

The apparatus 700 according to FIG. 7 is a device for quantitative analysis of emphysema for performing the method 100 of FIG. 1 and/or a learning device for a diagnostic module for quantitative analysis of emphysema for performing the method 200 of FIG. 2 As schematically showing the configuration of, the device 700 may be implemented to include a communication unit 710, an input unit 720, a memory 730 and at least one processor 740. At this time, for example, one device may be implemented to simultaneously drive the method 100 of FIG. 1 and the method 200 of FIG. 2, or two or more devices may be implemented to simultaneously drive the method 100 of FIG. It may also be implemented to drive at least one step of the method 200 of each.

The communication unit 710 may receive input data for diagnosing emphysema (chest CT image, etc.) and/or input data for learning an emphysema diagnosis module (chest CT shape, etc.). The communication unit 710 may include a wired/wireless communication unit. When the communication unit 710 includes a wired communication unit, the communication unit 710 may include a local area network (LAN), a wide area network (WAN), a value added network (VAN), and a mobile communication network ( mobile radio communication network), a satellite communication network, and one or more components that enable communication through a mutual combination thereof. In addition, when the communication unit 710 includes a wireless communication unit, the communication unit 710 transmits and receives data or signals wirelessly using cellular communication, a wireless LAN (eg, Wi-Fi), and the like. can In an embodiment, the communication unit may transmit/receive data or signals with an external device or an external server under the control of the processor 740 .

The input unit 720 may receive various user commands through external manipulation. To this end, the input unit 720 may include or connect one or more input devices. For example, the input unit 720 may be connected to an interface for various inputs such as a keypad and a mouse to receive user commands. To this end, the input unit 720 may include an interface such as a thunderbolt as well as a USB port. In addition, the input unit 720 may receive an external user command by including or combining various input devices such as a touch screen and buttons.

The memory 730 may store programs and/or program commands for operation of the processor 740 and may temporarily or permanently store input/output data. The memory 730 is a flash memory type, a hard disk type, a multimedia card micro type, a card type memory (eg SD or XD memory, etc.), RAM , SRAM, ROM (ROM), EEPROM, PROM, magnetic memory, a magnetic disk, it may include at least one type of storage medium.

In addition, the memory 730 may store various network functions and algorithms, and may store various data, programs (one or more instructions), applications, software, commands, codes, etc. for driving and controlling the device 700. there is.

The processor 740 may control the overall operation of the device 700 . Processor 740 may execute one or more programs stored in memory 730 . The processor 740 may mean a central processing unit (CPU), a graphics processing unit (GPU), or a dedicated processor on which methods according to the technical idea of the present disclosure are performed.

In an embodiment, the processor 740 acquires at least one training medical image generated corresponding to the continuous volume of a body part including the chest, and uses a pre-processing module including at least one learned network function, The medical image for training may be pre-processed by extracting a lung region from the training medical image and correcting the volume and brightness of the lung region.

In an embodiment, the processor 740 extracts a lung region from the training medical image using a preprocessing module, divides a left lung region and a right lung region from the training medical image from which the lung region is extracted, and the left lung region. and adjust the volume of the right lung region within a predetermined range, and convert the brightness of the lung region whose volume is adjusted within a predetermined range through histogram normalization.

In an embodiment, the processor 740 may label a severity class of emphysema in the medical image for training on which the preprocessing is completed.

In an embodiment, the processor 740 may learn at least one network function of the diagnosis module based on the labeled training medical images.

In an embodiment, the processor 740 obtains a medical image generated corresponding to the continuous volume of a body part including the chest, and uses a diagnosis module including at least one learned network function to obtain a medical image from the medical image. Emphysema areas are detected, and a severity class of emphysema is calculated based on the number and ratio of the emphysema areas, thereby generating diagnosis information on emphysema.

In an embodiment, the processor 740 may display the diagnosis information matched to the medical image.

In an embodiment, the processor 740 acquires at least one training medical image generated corresponding to the continuous volume of a body part including the chest, and uses a pre-processing module including at least one learned network function, The medical image for training may be pre-processed by extracting a lung region from the training medical image and correcting the volume and brightness of the lung region.

In an embodiment, the processor 740 extracts a lung region from the training medical image using a preprocessing module, divides a left lung region and a right lung region from the training medical image from which the lung region is extracted, and the left lung region. and adjust the volume of the right lung region within a predetermined range, and convert the brightness of the lung region whose volume is adjusted within a predetermined range through histogram normalization.

In an embodiment, the processor 740 may label a severity class of emphysema in the medical image for training on which the preprocessing is completed.

In an embodiment, the processor 740 may learn at least one network function of the diagnosis module based on the labeled training medical images.

The method according to an embodiment of the present disclosure may be implemented in the form of program instructions that can be executed through 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. Program commands recorded on the medium may be specially designed and configured for the present disclosure, or may be known and usable to those skilled in computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. - includes hardware devices specially configured to store and execute program instructions, such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language codes that can be executed by a computer using an interpreter, as well as machine language codes such as those produced by a compiler.

In addition, the method according to the disclosed embodiments may be provided by being included in a computer program product. Computer program products may be traded between sellers and buyers as commodities.

A computer program product may include a S/W program and a computer-readable storage medium in which the S/W program is stored. For example, a computer program product may include a product in the form of a S/W program (eg, a downloadable app) that is distributed electronically through a manufacturer of an electronic device or an electronic marketplace (eg, Google Play Store, App Store). there is. For electronic distribution, at least a part of the S/W program may be stored in a storage medium or temporarily generated. In this case, the storage medium may be a storage medium of a manufacturer's server, an electronic market server, or a relay server temporarily storing SW programs.

A computer program product may include a storage medium of a server or a storage medium of a client device in a system composed of a server and a client device. Alternatively, if there is a third device (eg, a smart phone) that is communicatively connected to the server or the client device, the computer program product may include a storage medium of the third device. Alternatively, the computer program product may include a S/W program itself transmitted from the server to the client device or the third device or from the third device to the client device.

In this case, one of the server, the client device and the third device may execute the computer program product to perform the method according to the disclosed embodiments. Alternatively, two or more of the server, the client device, and the third device may execute the computer program product to implement the method according to the disclosed embodiments in a distributed manner.

For example, a server (eg, a cloud server or an artificial intelligence server) may execute a computer program product stored in the server to control a client device communicatively connected to the server to perform a method according to the disclosed embodiments.

Although the embodiments have been described in detail above, the scope of the present disclosure is not limited thereto, and various modifications and improvements of those skilled in the art using the basic concepts of the present disclosure defined in the following claims are also within the scope of the present disclosure. belongs to

Claims (7)

As a learning method of a diagnostic module for quantitative analysis of emphysema,
acquiring at least one training medical image generated corresponding to a continuous volume of a body part including the chest; and
Extracting a lung region from the learning medical image using a preprocessing module including at least one learned network function, and preprocessing the learning medical image by correcting the volume and brightness of the lung region, method. According to claim 1,
In the preprocessing step,
extracting a lung region from the training medical image;
segmenting a left lung region and a right lung region from the training medical image from which the lung region is extracted;
adjusting the volume of the left lung region and the right lung region within a predetermined range; and
And converting the brightness of the lung region, the volume of which is adjusted, within a predetermined range through histogram normalization. According to claim 1,
The method further comprising the step of labeling the severity class of emphysema in the training medical image for which the preprocessing is completed. According to claim 3,
The severity class is divided into normal, mild, moderate and severe. According to claim 3,
The method further comprising learning at least one network function of the diagnosis module based on the labeled training medical images. As a device for learning a diagnostic module for quantitative analysis of emphysema,
at least one processor; and
A memory for storing a program executable by the processor;
The processor, by executing the program, obtains at least one training medical image generated corresponding to the continuous volume of the body part including the chest, and uses a preprocessing module including at least one learned network function. , An apparatus for pre-processing the medical image for learning by extracting a lung region from the medical image for learning and correcting the volume and brightness of the lung region. A computer program stored in a recording medium to execute the method of any one of claims 1 to 5.

Images