Method for providing diagnostic auxiliary information and device for executing the same

Abstract

To provide a unified criterion in analyzing a medical image to calculate an index regarding a disease, thereby eliminating an effect of a doctor's subjective judgment.SOLUTION: A method of providing diagnosis assistance information by analyzing a medical image comprises: obtaining an image; labeling a feature value reflecting a region of a brain; determining a reference boundary in the image of the brain; calculating a first disease index and a second disease index; and providing diagnosis assistance information on the basis of the first and second disease indexes.SELECTED DRAWING: Figure 5

Description

[Technical field]

[0001] The present invention relates to a method for analyzing medical images to provide auxiliary diagnostic information and a system for implementing the same, and in particular to a method for segmenting a medical image into a plurality of regions and providing auxiliary diagnostic information based on the association relationships between the plurality of regions and a system for implementing the same. [Background technology]

[0002] In the field of modern medical technology, there is an increasing demand for various techniques to precisely analyze medical images and provide more accurate disease-related information. In this trend, next-generation medical technology that analyzes medical images and provides a unified disease index is gaining attention.

[0003] There are many publications that have published methods for calculating indexes to provide disease-related information by analyzing medical images. However, most of these publications have limitations in that the clinical judgment of the doctor analyzing the medical images is involved, and the disease-related information provided to patients is overly subjective and subject to large deviations depending on the doctor.

[0004] In addition, even when image segmentation is performed to provide disease-related information from medical images, the area that can be segmented in the medical image varies depending on the conditions under which the medical image was acquired. As a result, it is difficult to obtain sufficient information about a disease from a specific medical image, and in order to provide accurate diagnostic auxiliary information, it is necessary to perform segmentation on each medical image acquired under various conditions, which is troublesome. Summary of the Invention [Problem to be solved by the invention]

[0005] The problem to be solved by the present invention is to provide a standardized standard for analyzing medical images and calculating disease-related indexes, thereby eliminating the influence of subjective judgment by doctors.

[0006] Another problem to be solved by the present invention is to calculate an objective and standardized disease index by analyzing medical images using a highly trained artificial neural network.

[0007] Another problem to be solved by the present invention is to provide disease information from which the influence of the diversity of medical images is removed by calculating a disease index corresponding to the characteristics of the medical images.

[0008] Another problem to be solved by the present invention is to increase the diversity of disease-related information provided by obtaining a segmentation result having at least two or more unique characteristics from one medical image. [Means for solving the problem]

[0009] An apparatus for providing diagnostic auxiliary information according to an embodiment of the present invention includes a communications module for acquiring MRI images related to the brain, a memory storing a program for analyzing the images, and a controller for analyzing the MRI images related to the brain using the program stored in the memory, wherein the controller distinguishes between ventricular regions and white matter hyperintensity signal regions from the images, sets a reference boundary a predetermined distance away from the ventricular region, calculates a first disease index based on white matter hyperintensity signal regions present within the reference boundary, and a second disease index based on white matter hyperintensity signal regions present outside the reference boundary, and provides diagnostic auxiliary information based on the first disease index and the second disease index.

[0010] According to another embodiment of the present invention, there is provided a method for providing diagnostic auxiliary information related to the brain, the method comprising the steps of: acquiring a brain image including a plurality of cells; labeling the plurality of cells with feature values ​​reflecting brain regions, the brain regions including ventricles and white matter hyperintensity; determining a reference boundary within the brain image, the reference boundary being defined as a set of cells located a predetermined distance away from the ventricles; calculating a first disease index associated with cells labeled with feature values ​​representing white matter hyperintensity signals located within the reference boundary and a second disease index associated with cells labeled with feature values ​​representing white matter hyperintensity signals located outside the reference boundary; and providing diagnostic auxiliary information based on the first disease index and the second disease index. Effect of the Invention

[0011] According to the present invention, a standardized standard is provided when calculating a disease index from a medical image, so that objective and clear disease information can be provided from the patient's condition.

[0012] According to the present invention, a method for calculating a criterion that reflects characteristics contained in a medical image is provided, thereby providing a method for providing disease information that can be applied to various medical images.

[0013] According to the present invention, by analyzing medical images using a highly trained artificial neural network, it is possible to provide objective and clear disease information that is free from the clinical judgment of doctors. According to the present invention, by utilizing an artificial neural network trained through medical images having at least two or more different characteristics, it is possible to accurately provide disease information for medical images that correspond to the characteristics of the disease possessed by the patient. [Brief description of the drawings]

[0014] [Figure 1] FIG. 1 is a diagram illustrating an example of a system for providing auxiliary diagnostic information according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a block diagram showing a schematic configuration of a system for providing auxiliary diagnostic information according to an embodiment of the present invention. [Diagram 3] FIG. 3 is a block diagram showing the configuration of an image capturing device according to an embodiment of the present invention. [Figure 4] FIG. 4 is a block diagram showing a configuration of a video analysis device according to an embodiment of the present invention. [Diagram 5] FIG. 5 is a general flowchart illustrating a method for providing auxiliary diagnostic information executed in a system for providing auxiliary diagnostic information according to an embodiment of the present invention. [Figure 6] FIG. 6 is a diagram illustrating a medical image according to an embodiment of the present invention. [Figure 7] FIG. 7 is a diagram illustrating medical images according to various acquisition conditions according to an embodiment of the present invention. [Figure 8] FIG. 8 is a flowchart illustrating a detailed operation of a medical image analysis performed by an image analysis apparatus according to an embodiment of the present invention. [Figure 9] FIG. 9 is an example diagram of a segmented medical image according to an embodiment of the present invention. [Figure 10] FIG. 10 is a flow chart illustrating an example method for extracting disease indexes from medical images according to an embodiment of the present invention. [Figure 11] FIG. 11 is a diagram illustrating an example of a medical image showing a reference region and a reference boundary according to an embodiment of the present invention. [Figure 12] FIG. 12 is a diagram illustrating an example of a process for obtaining region information of a target region according to an embodiment of the present invention. [Figure 13] FIG. 13 is a diagram illustrating another example of a process for acquiring area information of a target area according to an embodiment of the present invention. [Figure 14] FIG. 14 is a diagram illustrating the provision of diagnostic auxiliary information obtained based on a disease index calculated according to an embodiment of the present invention. [Figure 15] FIG. 15 is a flow chart illustrating an example of a method for providing diagnostic assistance information using a reference boundary modified in accordance with an embodiment of the present invention. [Figure 16] FIG. 16 is a diagram illustrating a case where a reference boundary and a region of non-interest overlap according to an embodiment of the present invention. [Figure 17a] FIG. 17a is an exemplary diagram of a method for calculating a modified reference boundary according to an embodiment of the present invention. [Figure 17b] FIG. 17b is an exemplary diagram of a method for calculating a modified reference boundary according to an embodiment of the present invention. [Figure 18] FIG. 18 is an exemplary diagram of a reference boundary that is partially modified and a reference boundary that is entirely modified according to an embodiment of the present invention. [Figure 19] FIG. 19 is a diagram illustrating a specific modification process of a partially modified reference boundary according to an embodiment of the present invention. [Figure 20] FIG. 20 is a schematic flow chart illustrating a method for an image analyzer to provide auxiliary diagnostic information from a plurality of medical images according to an embodiment of the present invention. [Figure 21] FIG. 21 is a diagram illustrating an example of multiple medical images according to an embodiment of the present invention. [Figure 22] FIG. 22 is a diagram illustrating an example of slice images including different pieces of information and the provision of auxiliary diagnostic information therefrom according to an embodiment of the present invention. [Diagram 23] FIG. 23 is a flow chart that generally illustrates a method for obtaining auxiliary diagnostic information from a candidate image according to an embodiment of the present invention. [Figure 24] FIG. 24 is a diagram illustrating candidate images according to an embodiment of the present invention. [Diagram 25] FIG. 25 is a diagram illustrating an example of providing auxiliary diagnostic information from a candidate image according to an embodiment of the present invention. [Figure 26] FIG. 26 is a diagram illustrating another example of providing auxiliary diagnostic information according to an embodiment of the present invention. [Figure 27] FIG. 27 is a flow chart illustrating a method for obtaining a disease index based on multiple images according to an embodiment of the present invention. [Figure 28] FIG. 28 is a diagram illustrating an example of a process for calculating a disease index of a second image by considering information of a first image according to an embodiment of the present invention. [Figure 29] FIG. 29 is a diagram illustrating another example of a process of calculating a disease index of a second image by considering information of a first image according to an embodiment of the present invention. [Diagram 30] FIG. 30 is a flow chart illustrating an example of a method for providing diagnostic auxiliary information based on a 3D medical model according to an embodiment of the present invention. [Diagram 31] FIG. 31 is a diagram illustrating an example of a process for calculating a disease index from a 3D medical model according to an embodiment of the present invention. [Diagram 32] FIG. 32 is a diagram illustrating another example of a process for calculating a disease index from a 3D medical model according to an embodiment of the present invention. [Diagram 33] FIG. 33 is a schematic diagram illustrating a medical image segmentation operation performed by a video analysis apparatus according to an embodiment of the present invention. [Diagram 34] FIG. 34 illustrates an example of an artificial neural network model according to an embodiment of the present invention. [Diagram 35] FIG. 35 illustrates another implementation of an artificial neural network model according to an embodiment of the present invention. [Diagram 36] FIG. 36 is a diagram illustrating a segmentation result using an artificial neural network model according to an embodiment of the present invention. [Figure 37] FIG. 37 is an illustrative diagram of a medical image segmented to include multiple features in accordance with an embodiment of the present invention. [Figure 38]FIG. 38 is a flow chart illustrating the training process of an artificial neural network according to an embodiment of the present invention. [Figure 39] FIG. 39 is an example showing a process of matching a first image with a second image according to an embodiment of the present invention. [Diagram 40] FIG. 40 illustrates an example of morphological correction according to an embodiment of the present invention. [Diagram 41] FIG. 41 is a flowchart illustrating a deploying process using an artificial neural network of an image analysis device according to an embodiment of the present invention. [Diagram 42] FIG. 42 is a diagram showing an example of a final output result of the artificial neural network according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] An apparatus for providing diagnostic auxiliary information according to an embodiment of the present invention includes a communication module for acquiring an MRI image related to a brain, a memory storing a program for analyzing the image, and a controller for analyzing the MRI image related to the brain using the program stored in the memory, wherein the controller is capable of distinguishing a ventricular region and a white matter hyperintensity signal region from the image, setting a reference boundary a predetermined distance away from the ventricular region, calculating a first disease index based on white matter hyperintensity signal regions present within the reference boundary, and a second disease index based on white matter hyperintensity signal regions present outside the reference boundary, and providing diagnostic auxiliary information based on the first disease index and the second disease index.

[0016] In another embodiment of the present invention, a method for providing diagnostic auxiliary information can provide diagnostic auxiliary information related to the brain, the method comprising the steps of obtaining a brain image including a plurality of cells; labeling the plurality of cells with feature values ​​reflecting brain regions, the brain regions including ventricles and white matter hyperintensity; determining a reference boundary within the brain image, the reference boundary being defined as a set of cells located a predetermined distance away from the ventricles; calculating a first disease index associated with cells labeled with feature values ​​representing white matter hyperintensity signals located within the reference boundary and a second disease index associated with cells labeled with feature values ​​representing white matter hyperintensity signals located outside the reference boundary; and providing diagnostic auxiliary information based on the first disease index and the second disease index.

[0017] The above objects, features and advantages of the present invention will become more apparent through the detailed description of the accompanying drawings. Of course, the present invention can be realized in various modified embodiments, but the following will illustrate specific embodiments by way of example and describe them in detail.

[0018] In the drawings, the thickness of layers and regions are exaggerated for clarity, and a description that an element or layer is located "on" or "on" another element or layer includes not only the case where the element or layer is located immediately above the other element or layer, but also the case where another layer or other element is interposed therebetween. In this specification, the same reference numerals generally indicate the same elements. In addition, in the drawings relating to each embodiment, elements having the same function within the same concept are described using the same reference numerals.

[0019] If a detailed description of well-known functions or configurations related to the present invention is deemed unnecessary for the gist of the present invention, the detailed description will be omitted. In addition, numbers (e.g., 1, 2, etc.) used in the description of this specification are merely identification symbols for distinguishing one component from another.

[0020] In the following description, the terms "module" and "section" which represent units of components are used for the ease of writing the specification, and do not have any distinguishing meaning or role by themselves.

[0021] The present disclosure relates to a method for analyzing medical images to provide auxiliary diagnostic information related to the medical images, and a system for implementing the method.

[0022] Here, the term "medical image" in this specification can be interpreted as including various images that can be obtained in the medical industry. That is, the term "medical image" in this specification is a general term for images that can be obtained from various devices used in the medical industry. One example is an image obtained by a computer tomography (CT) device such as a magnetic resonance imaging (MRI). Another example is a medical image obtained by an X-ray device (X-ray). Of course, the medical image is not limited to such images, and can include all images that can be obtained in the medical industry, such as photographs that can be obtained by a general camera device.

[0023] Furthermore, the target object of a medical image according to one embodiment may include all photographed objects that are targets of the medical industry. As an example, the medical image may be an image of a patient. Specifically, the medical image may be an image of a specific part of the patient's body. As a more specific example, the medical image may be a photo related to the brain of a patient suspected of having dementia, or an image related to the lungs of a patient suspected of having lung cancer. In addition, the target object of a medical image according to another embodiment may be bodily tissue taken from a human being.

[0024] That is, the medical image in this specification may be an image of any object to be photographed in the medical industry from which disease information is to be obtained by the image analysis method according to the concept of the present invention, or may be an image having a form obtained by various devices used in the medical industry. However, in the following description of this specification, for convenience of explanation, magnetic resonance imaging (hereinafter, referred to as "MRI") related to the human brain will be mainly described, but the concept of the present invention is not limited thereto.

[0025] Furthermore, in this specification, "diagnostic auxiliary information" refers to comprehensive information that can be determined from a medical image and that can be objectively determined in relation to a disease. As an example, the diagnostic auxiliary information in this specification is the presence or absence of a specific disease. As another example, the diagnostic auxiliary information is the degree of progression of a specific disease. As yet another example, the diagnostic auxiliary information is the severity of a disease. As yet another example, the diagnostic auxiliary information is a numerical representation of the degree of disease of a specific patient in relation to the average population. In addition, in this specification, all information related to a disease that can be determined within a medical image as described above can be collectively referred to as "diagnostic auxiliary information".

[0026] First, a configuration of a diagnostic auxiliary information providing system according to one embodiment and an example of its execution will be described with reference to FIGS. 1 and 2. FIG.

[0027] FIG. 1 shows an example of an implementation of a diagnostic auxiliary information providing system according to an embodiment of the present invention, and FIG. 2 is a block diagram showing a schematic configuration of the diagnostic auxiliary information providing system according to an embodiment of the present invention.

[0028] 1 and 2, a system (10000) according to an embodiment can analyze medical images to provide auxiliary diagnostic information.

[0029] In this embodiment, the system (10000) includes an image capture device (1000) and an image analysis device (2000).

[0030] The image acquisition device 1000 can acquire a medical image from a subject to be imaged. Here, the image acquisition device 1000 refers to various devices or systems capable of acquiring a medical image. As an example, as shown in FIG. 1, the image acquisition device 1000 is an MRI device. However, as described above, the image acquisition device 1000 referred to in this specification is not limited to an MRI device.

[0031] The image analysis device 2000 can provide auxiliary diagnostic information by analyzing medical images acquired from the image acquisition device 1000. In particular, the image analysis device 2000 can extract a number of indexes related to diseases from the medical images and calculate auxiliary diagnostic information based on the extracted indexes. The medical image analysis operation performed by the image analysis device 2000 will be described in detail later.

[0032] Also, although the image capturing device 1000 and the image analysis device 2000 have been described above as separate devices, this is merely an example and includes various forms that can be embodied as the diagnostic assistance information providing system 10000. That is, the image analysis device 2000 according to the embodiment can be embodied as a separate server, or can be embodied as a program integrated into the image capturing device 1000. However, for ease of explanation, the present specification will describe the system 10000 embodied as a separate image analysis device 2000.

[0033] Hereinafter, configurations of an image capturing device and an image analyzing device according to an embodiment will be described with reference to the accompanying drawings.

[0034] FIG. 3 is a block diagram showing the configuration of an image capturing device according to an embodiment.

[0035] As shown in FIG. 3, the image capture device (1000) according to the embodiment includes a first controller (1002), an image capture module (1200), a first memory (1400) and a first communication module (1800).

[0036] In this embodiment, the first controller (1002) can transmit the medical image acquired by the image acquisition module (1200) to the image analysis device (2000) through the first communication module (1800).

[0037] Hereinafter, each component of the image capturing device 1000 according to the embodiment will be described.

[0038] According to the present embodiment, the image acquisition module 1200 can acquire imaging results for an object to be imaged. Here, the image acquisition module 1200 can include a configuration for acquiring various medical images. As an example, the image acquisition module 1200 can be configured for acquiring a magnetic resonance imaging (MRI) image. As another example, the image acquisition module 1200 can be configured for acquiring an X-ray or CT image.

[0039] Here, the first controller (1002) can adjust the setting parameters of the image acquisition module. For example, if the image acquisition module (1200) is configured to acquire an MRI image, the first controller (1002) can adjust the repetition time (TR) and echo time (TE) of the MRI imaging device. This allows the MRI imaging device to acquire a T1-weighted image or a T2-weighted image. The first controller (1002) can also adjust parameters related to an inversion pulse so that the MRI imaging device acquires a FLAIR image.

[0040] The first communication module (1800) according to the embodiment may communicate with an external device or an external server. The image acquisition device (1000) may perform data communication with the image analysis device (2000) or an external device (or server) through the first communication module (1800). For example, the image capturing device (1000) can transmit a medical image or data related to the medical image to the image analyzing device (2000) or an external device using the first communication module (1800).

[0041] The first communication module (1800) is mainly divided into a wired type and a wireless type. Since the wired type and the wireless type each have advantages and disadvantages, depending on the case, the image acquisition device (1000) may be prepared with both a wired type and a wireless type.

[0042] Here, in the case of a wired type, LAN (Local Area Network) or USB (Universal Serial Bus) communication is representative, but other methods may also be used.

[0043] In addition, in the case of a wireless type, a communication method of the WPAN (Wireless Personal Area Network) series such as Bluetooth or Zigbee can be mainly used. However, since the wireless communication protocol is not limited to this, a wireless type communication module may use a communication method of the WLAN (Wireless Local Area Network) series such as Wi-Fi or other known communication methods.

[0044] The first memory (1400) can store various information. Various data is stored temporarily or semi-permanently in the first memory (1400). Examples of the first memory (1400) include a hard disk drive (HDD), a solid state drive (SSD), a flash memory, a read-only memory (ROM), and a random access memory (RAM).

[0045] The first memory (1400) stores various data necessary for the operation of the image capture device (1000), including an operating program (OS: Operating System) for driving the image capture device (1000) and programs for operating each component of the image capture device (1000).

[0046] The first controller (1002) according to the embodiment may control the overall operation of the image capture device (1000). For example, the first controller (1002) may generate a control signal to receive medical images from the capture module (1200) and transmit them to the data analysis device (2000) through the first communication module (1800).

[0047] The first controller (1002) can be realized as a CPU (Central Processing Unit) or a similar device by hardware, software, or a combination of both. In terms of hardware, it can be provided in the form of an electronic circuit that processes electrical signals and executes control functions, and in terms of software, it can be provided in the form of a program or code that drives a hardware circuit.

[0048] The image capturing device (1000) may have a separate power supply unit, or may receive power from an external source via wire or wirelessly, and may have a switch for controlling the power supply unit.

[0049] FIG. 4 is a block diagram showing the configuration of a video analysis device according to an embodiment.

[0050] Referring to FIG. 4, the image analysis device (2000) includes a second controller (2002), a second memory (2400), a display module (2600) and a second communication module (2800).

[0051] According to this embodiment, the second controller (2002) can acquire medical images from the image acquisition device (1000) through the second communication module (2800), analyze the medical images using an analysis program stored in the second memory (2400), and calculate auxiliary diagnostic information from the medical images.

[0052] Hereinafter, each component of the video analysis device 2000 according to the embodiment will be described.

[0053] The second memory (2400) can store various information of the image analysis device (2000).

[0054] The second memory (2200) can store various data required for the operation of the image analysis device (2000), including an operation program for driving the image analysis device (2000) and a program for operating each component of the image analysis device (2000). For example, the second memory (2400) can store a program for processing medical images and / or a program for analyzing the processed medical images. The above programs can be realized as machine running algorithms, and a detailed description will be given later.

[0055] Examples of the second memory (2400) include a hard disk drive (HDD), a solid state drive (SSD), a flash memory, a read-only memory (ROM), and a random access memory (RAM).

[0056] The second memory (2400) can store various data necessary for the operation of the video analysis device (2000), including an operating program (OS: Operating System) for driving the video analysis device (2000) and programs for operating each component of the video analysis device (2000).

[0057] The second communication module (2800) can communicate with an external device or an external server. The video analysis device (2000) can perform data communication with the video acquisition device (1000) or an external server using the second communication module (2800). For example, the video analysis device (2000) can acquire medical images necessary for providing auxiliary diagnostic information from the video acquisition device (1000) using the second communication module (2800).

[0058] The second communication module (2800) can be roughly divided into a wired type and a wireless type. Since the wired type and the wireless type each have advantages and disadvantages, in some cases, the image analysis device (2000) can be equipped with both the wired type and the wireless type at the same time.

[0059] Here, in the case of a wired type, LAN (Local Area Network) and USB (Universal Serial Bus) communication are typical examples, but other methods are also possible.

[0060] In addition, in the case of a wireless type, a communication method of the WPAN (Wireless Personal Area Network) series such as Bluetooth (registered trademark) or Zigbee can be mainly used. Of course, the wireless communication protocol is not limited to this, and the wireless type communication module can also use a communication method of the WLAN (Wireless Local Area Network) series such as Wi-Fi (Wireless Firmware) or other known communication methods.

[0061] The second controller (2002) can control the overall operation of the image analysis device (2000). For example, the second controller (2002) can load a program for processing and analyzing image data from the second memory (2400), process and analyze medical images acquired from the image acquisition device (1000), and generate a control signal to provide the results to an external device or an external server through the second communication module (2800). A specific method for providing auxiliary diagnostic information executed by the image analysis device (2000) will be described in detail later.

[0062] The second controller (2002) can be realized by a CPU (Central Processing Unit) or a similar device by hardware, software, or a combination of these. In terms of hardware, it can be provided in the form of an electronic circuit that processes electrical signals and executes control functions, and in terms of software, it can be provided in the form of a program or code that drives a hardware circuit.

[0063] Also, the image analyzing device (2000) may further include a separate display module (2600) for outputting the medical image analysis result. The display module may be embodied in various ways to provide information to the user. Here, the second controller (2002) may process and analyze the medical image acquired from the image acquiring device (1000) and generate a control signal for providing the result through the display module (2600).

[0064] Hereinafter, a method for analyzing a medical image and providing auxiliary diagnostic information, which is performed by the auxiliary diagnostic information providing system according to the embodiment, will be described in detail.

[0065] FIG. 5 is a general flowchart of a method for providing auxiliary diagnostic information executed by the auxiliary diagnostic information providing system according to the embodiment.

[0066] Referring to FIG. 5, the operation of providing diagnostic auxiliary information according to an embodiment includes a step of acquiring a medical image (S1000), a step of analyzing the acquired medical image (S1200), and a step of providing diagnostic auxiliary information based on the analysis results (S1400).

[0067] First, the system for providing diagnostic auxiliary information (10000) according to the embodiment can acquire a medical image from an object to be photographed. Specifically, the image acquisition device (1000) can acquire a medical image for acquiring diagnostic auxiliary information from an object to be photographed, such as a patient or a specimen. More specifically, the image acquisition module (1200) in the image acquisition device (1000) can acquire a medical image by photographing the object to be photographed, and transmit the acquired medical image to the image analysis device (2000) through the first communication module (1800). Here, there may be a plurality of medical images. That is, the image acquisition device (1000) can acquire a plurality of cross-sectional images in a specific direction or a specific plane from the object to be photographed.

[0068] Here, the medical images may contain three-dimensional information, i.e., the medical images may be acquired as 3D images.

[0069] In addition, the specific direction and specific plane may include information regarding all directions and planes that can be obtained from the object to be imaged, such as a transverse plane, a saggital plane, and a coronal plane of the object to be imaged.

[0070] And, the diagnostic assistance information providing system (10000) can analyze the medical image (S1200). The image analyzing device (2000) can analyze the medical image transmitted from the image acquiring device (1000). Specifically, the second controller (2002) can analyze the medical image received through the second communication module (2800) using a program for image analysis stored in the second memory (2400). As an example, the second controller (2002) can segment the medical image using the program stored in the second memory (2400).

[0071] Thereafter, the auxiliary diagnostic information providing system (10000) can obtain auxiliary diagnostic information based on the analysis result of the medical image (S1400). That is, the image analyzing device (2000) can obtain auxiliary diagnostic information included in the analyzed medical image from the medical image using a program for obtaining auxiliary diagnostic information stored in the second memory (2400). The second controller (2002) can transmit the calculated auxiliary diagnostic information to an external device or an external server through the second communication module (2800), or output the auxiliary diagnostic information using a separately prepared display module (2600), etc.

[0072] As described above, the embodiment has generally described the method for providing auxiliary diagnostic information executed by the auxiliary diagnostic information providing system (10000). Below, detailed operations and contents of each step of the method for providing auxiliary diagnostic information executed by the auxiliary diagnostic information providing system (10000) will be described.

[0073] First, regarding the medical image capturing step (S1000), various medical images that can be captured by the image capturing device (1000) according to the embodiment will be illustrated and described.

[0074] FIG. 6 illustrates an example of a medical image according to an embodiment.

[0075] 6, medical images according to the embodiment can be acquired by an image acquisition device 1000 implemented in various forms. The medical images are images captured by various computed tomography devices. As described above, the medical images can be a collection of tomographic images composed of a plurality of slice images.

[0076] As an example, the medical image is an MRI image taken by an MRI device. (a) and (b) are diagrams showing various body parts of a patient imaged by an MRI device. As another example, the medical image is an X-ray image taken by an X-ray device. (c) is a diagram showing an X-ray image of a patient imaged by an X-ray device. As yet another example, the medical image is a CT image taken by a CT device. (d) is a diagram showing a CT image of a patient imaged by a CT device.

[0077] In addition, the medical images according to the embodiment are images relating to each body part. Referring again to FIG. 6, the medical images are images relating to body parts including not only the patient's organs such as the brain and lungs, but also the skeletal structure such as the spine and the nervous structure.

[0078] According to the embodiment, the information contained in the medical image may vary depending on the setting parameters of the image capture device 1000 that captures the medical image.

[0079] FIG. 7 shows examples of medical images obtained under various acquisition conditions according to this embodiment.

[0080] 7, the image acquiring device 1000 according to the embodiment can acquire various images by setting parameters of the device. For example, when the image acquiring device 1000 is implemented as a 12-inch MRI device, the image acquiring device 1000 can acquire medical images including various information by adjusting parameters related to magnetic conditions.

[0081] As a more specific example, the image acquisition device 1000 may acquire a T1-weighted image by setting parameters to have a short TR / TE time. The T1-weighted image has a high signal intensity and is useful for clearly distinguishing anatomical structures. That is, the T1-weighted image is mainly used to determine the anatomical characteristics of the human body.

[0082] In addition, the image acquisition device (1000) can acquire T2-weighted images by setting parameters to have a long TR / TE time, and can also acquire FLAIR (Fluid attenuated inversion recovery) images by using an inversion pulse. T2-weighted images and FLAIR images are characterized by areas containing water appearing white, and lesion areas are mainly high in water content, so they are effectively used to find lesion areas.

[0083] In addition, medical images using special imaging techniques such as diffusion weighted imaging (DWI), PET, and fMRI are also included in the concept of the present invention.

[0084] As described above, various examples of medical images acquired in the medical image acquiring step (S1000) according to the embodiment have been described. Hereinafter, the details of the acquired medical image analyzing step (S1200) will be described.

[0085] FIG. 8 is a flowchart illustrating a detailed operation of a medical image analysis operation performed by the image analysis apparatus according to an embodiment.

[0086] As shown in FIG. 8, a medical image analysis operation according to an embodiment may include a step of segmenting a medical image (S2002), a step of extracting a disease index from the segmented medical image (S2004), and a step of obtaining diagnostic auxiliary information based on the extracted disease index (S2006).

[0087] First, the image analysis device (2000) can segment the medical image (S2002). Specifically, the second controller (2002) can segment the medical image using a program for image segmentation stored in the second memory (2400). Exemplarily, the second controller (2002) can segment the medical image related to the brain into brain regions. Here, the brain region may refer to a lesion found in the medical image, or may refer to a structural region of the brain found in the medical image. That is, the brain region found in the medical image may include all objects that can be distinguished in the image. That is, the second controller (2002) can segment the medical image related to the brain so that it is divided into at least two or more regions.

[0088] The program for image segmentation stored in the second memory (2002) can be implemented as a machine running algorithm, which will be described in detail later.

[0089] After the medical image segmentation step (S2002), the image analysis device (2000) can extract a disease index from the segmented medical image (S2004). Specifically, the second controller (2002) can extract a disease index based on each region included in the segmented medical image using an algorithm stored in the second memory (2400).

[0090] In this specification, the disease index may be defined as an index that expresses the relationship between each brain region in a medical image in relation to a specific disease. The disease index may, for example, include the Fazekas scale, ARWMC (Age-related white matter change), Posterior atrophy score of parietal atrophy, MTA (Medial temporal lobe atrophy score), Orbito-Frontal, Anterior Cingulate, Fronto-Insula, Anterior Temporal Scale, etc., and various indexes that can extract disease-related information from medical images may be used.

[0091] After the disease index extraction step (S2004), the image analysis device (2000) can provide auxiliary diagnostic information based on the extracted disease index. Specifically, the second controller (2002) can obtain disease-related information from the medical image based on the extracted disease index using an algorithm stored in the second memory (2400) and output the obtained result.

[0092] Below, examples of segmented medical images for disease index extraction will be described, and then various embodiments for extracting a disease index from the segmented medical images will be described with reference to FIG.

[0093] FIG. 9 illustrates an example of a segmented medical image according to an embodiment.

[0094] Referring to FIG. 9, an example of a medical image segmented by the video analyzer (2000) is shown.

[0095] In FIG 9, the image analysis device (2000) can segment the medical image. Here, segmentation means that the image analysis device (2000) assigns a specific value to a unit cell included in the medical image. Specifically, segmentation means that the second controller (2002) labels pixels or voxels included in the medical image with feature values. As an example, the second controller (2002) can label a plurality of pixels included in the medical image with feature values ​​indicating brain regions.

[0096] The image analysis device (2000) may segment a medical image to correspond to the form of the medical image. Here, the form of the medical image may correspond to the acquisition conditions when the medical image is acquired. For example, when the medical image is taken under a specific magnetic condition, the medical image may have a T1-weighted image form. It should be understood that, for convenience of explanation, the terms "form of the medical image" and "acquisition conditions of the medical image" are used interchangeably, but the concept of this specification does not change depending on the use of the terms. In addition, the term "characteristics of the medical image" may be used interchangeably because the characteristics contained in the medical image (e.g., anatomical characteristics or pathological characteristics) may change depending on the form of the medical image or the acquisition conditions of the medical image.

[0097] Specifically, the second controller (2002) can segment the medical image using a program for image segmentation stored in the second memory (2400) so that the characteristics of the medical image can be reflected.

[0098] The second controller (2002) can label pixels included in the medical image with values ​​indicating brain regions. That is, in a medical image embodied as a FLAIR image, the second controller (2002) can label pixels classified as white matter, gray matter, ventricles, etc. with values ​​indicating the respective regions.

[0099] Illustratively, (a) shows the image analysis device (2000) segmenting a T1-weighted MRI image, and (b) shows the image analysis device (2000) segmenting a T2-FLAIR MRI image.

[0100] As shown in the drawing, the T1-weighted MRI image (a) makes it easy to grasp the anatomical characteristics of the brain. That is, when the image analysis device (2000) segments the T1-weighted MRI image, the image analysis device (2000) can perform segmentation so that each part constituting the brain can be distinguished. For example, the image analysis device (2000) according to the embodiment can segment the brain image so that the organs included in the brain, such as the cerebrum, cerebellum, diencephalon, and hippocampus, can be distinguished, and the brain can be segmented so that each part and position of the brain, such as the temporal lobe, frontal lobe, and occipital lobe, can be distinguished, or a combination of these can be segmented. In this way, when analyzing the T1-weighted MRI image in which the anatomical characteristics of the brain can be well observed to obtain auxiliary diagnostic information, the image analysis device (2000) can easily analyze disease information related to the anatomical characteristics of the brain as well as atrophic Alzheimer's.

[0101] Also, the T2 FLAIR MRI image (b) makes it easy to grasp the pathological characteristics of the brain. When the image analysis device (2000) according to the embodiment segments the T2-FLAIR MRI image, the image analysis device (2000) can segment the medical image so that the pathological characteristics of the brain can be distinguished. Here, the pathological characteristics of the brain can be detected by white matter hyperintensity (hereinafter referred to as "WMH") that can be observed from the medical image. That is, the image analysis device (2000) can segment the medical image so that the WMH can be distinguished from the material and other regions that constitute the brain in the T2-FLAIR MRI image. Here, the material that constitutes the brain (or the human head) includes white matter, gray matter, skull, etc., and other regions include ventricles, etc. The image analyzer (2000) can extract disease indices based on the location of the WMHs and their relationship to other brain structures, as will be described in more detail below.

[0102] In the drawings, T1-weighted MRI images and T2-FLAIR MRI images are mainly described as examples, but as mentioned above, the idea of ​​the present specification is not limited thereto, and MRI images taken under other magnetic conditions such as DWI and SWI, and medical images taken with other imaging devices such as X-ray and CT can all be used.

[0103] In addition, the image analysis device (2000) may perform segmentation such that a medical image having a specific characteristic (i.e., acquired under a specific acquisition condition) includes information on another characteristic. Specifically, the second controller (2002) may perform segmentation using a machine running algorithm stored in the second memory (2400) such that a medical image acquired under a first condition and having a first characteristic includes not only the first characteristic but also information on a second characteristic that can be acquired when acquired under a second condition. For example, when segmenting a T2-FLAIR MRI image, the image analysis device (2000) may segment the T2-FLAIR image such that anatomical information that can be observed in a T1-weighted MRI image is also included. This will be described in detail later with reference to the drawings.

[0104] Hereinafter, a specific embodiment of a method for extracting a disease index from a segmented medical image will be described with reference to the accompanying drawings.

[0105] Fig. 10 is a flow chart showing an example of a method for extracting a disease index from a medical image according to the present embodiment, Fig. 11 shows an example of a medical image in which a reference region and a reference boundary are illustrated, Fig. 12 shows an example of a case in which region information of a target region is obtained, Fig. 13 shows another example of a case in which region information of a target region is obtained, and Fig. 14 shows an embodiment in which auxiliary diagnostic information obtained based on a calculated disease index is provided.

[0106] First, a general flow will be described with reference to Figure 10. In this embodiment, the method for extracting a disease index from a medical image segmented by the video analysis device (2000) includes a step of detecting a reference region from the segmented medical image (S2102), a step of setting a reference boundary based on the detected reference region (S2104), and a step of extracting a disease index based on the relationship between the reference boundary and a target region (S2106).

[0107] The image analysis device (2000) can detect a reference region from the segmented medical image (S2102). Specifically, the second controller (2002) can detect pixels labeled to indicate a reference region in the segmented medical image and determine a set of these as the reference region. Here, the reference region means a region that is included in the medical image and serves as a reference for calculating a specific disease index, as described below. As an example, the reference region is a ventricle region. As another example, the reference region is a white matter region.

[0108] Here, the reference region is a region in which a buffer region of a certain area or more is added to a set of pixels labeled to indicate the reference region. The reference region may be defined as a region in which a certain pixel on the outer edge is removed from a set of pixels labeled to indicate the reference region.

[0109] Also, the image analysis device (2000) can detect a region corresponding to the analysis target from the segmented medical image. Specifically, the second controller (2002) can detect pixels labeled to indicate the analysis target and determine a set of these pixels as a region to be analyzed (hereinafter referred to as a "target region"). Here, the target region may be included in a medical image as described below, and may refer to a region that is the target of analysis for calculating a specific disease index. As an example, the target region refers to a WMH region. A method for detecting the target region is similar to the method for detecting the ventricular region described above, and therefore a detailed description will be omitted.

[0110] When the reference region is detected, the image analysis device (2000) can set a reference boundary based on the detected reference region (S2104). Specifically, the second controller (2002) can set a boundary that is a predetermined distance away from the ventricular region detected in the medical image as the reference boundary.

[0111] Here, the reference boundary means a boundary that serves as a reference for extracting a disease index by deriving a relationship between a reference region and a target region, as described below. That is, when a disease index such as the Fazekas scale is used to determine disease information related to Alzheimer's, the positional relationship between the ventricle and the WMH can be used as an important index, and the reference boundary is used as a reference for grasping the positional relationship between the ventricle and the WMH. In this case, the ventricle can be determined as the reference region of the Fazekas scale, and the WMH can be determined as the target region of the Fazekas scale.

[0112] In addition, to obtain the reference boundary, a predetermined distance can be calculated from any point within the reference area. That is, for example, the reference boundary means a set of pixels that are a predetermined distance away from pixels on the outer edge of the reference area. In this case, the direction that is the reference for the predetermined distance is preferably the normal direction of the boundary surface of the reference area. As another example, the reference boundary means a set of pixels that are a predetermined distance away from the center of the reference area.

[0113] The reference boundary will be described with reference to FIG.

[0114] FIG. 11 shows the segmentation results of the ventricles, WMH, white matter, gray matter, skull, etc. in a T2-FLAIR MRI image.

[0115] The image analysis device (2000) can use the Fazekas scale as a disease index. Specifically, the second controller (2002) can determine the ventricular region as a reference region and the WMH region as a target region to calculate the Fazekas scale. Here, the second controller (2002) can set a set of pixels that are a predetermined distance away from pixels labeled as the ventricle as a reference boundary for calculating the Fazekas scale. Also, the reference boundary can be calculated from pixels on the outer edge of the ventricular region, and the predetermined distance can be calculated as 10 mm. However, this is merely an example value, and any distance within a range of 5 mm to 15 mm can be determined.

[0116] Referring to FIG. 11, it is illustrated that WMHs exist both inside and outside the reference boundary. WMHs existing inside and outside the reference boundary act as important factors in determining auxiliary diagnostic information. In particular, information on the area occupied by WMHs (hereinafter referred to as "area information") and the relationship between the reference boundary are closely related to auxiliary diagnostic information. A method for calculating area information of WMHs from a medical image in which WMHs are segmented will be described with reference to FIG. 12 and FIG. 13.

[0117] According to medical research, white matter regions in the brain that are relatively close to the ventricles are generally more susceptible to degeneration than white matter regions that are relatively far from the ventricles. Therefore, WMH regions formed around the ventricles and WMH regions that are relatively far from the ventricles contained in medical images can each contain different information about diseases. Accordingly, when analyzing WMH regions contained in medical images, it is necessary to take into account their distance from the ventricles.

[0118] 12 and 13 show examples of methods for obtaining region information related to a target region, taking into account the distance between a reference region and the target region.

[0119] As shown in Fig. 12, the image analysis apparatus (2000) according to the embodiment can obtain region information of a target region in a medical image. Specifically, the second controller (2002) can obtain region information for the target region in the medical image based on the segmentation result. Here, the region information means the thickness or width of a particular region. In addition, as described below, when the medical image is realized in 3D, the region information means the volume.

[0120] First, when calculating a disease index based on the Fazekas scale, a method of calculating region information of a target region that exists near a reference region included in a medical image, i.e., exists within the above-mentioned reference boundary, will be described.

[0121] The image analysis device 2000 can obtain region information of the target region based on the distance from the reference region to the target region. Specifically, the second controller 2002 can calculate region information of the target region based on the distance from a region labeled as a reference region to a region labeled as a target region in the medical image.

[0122] For example, the second controller (2002) can calculate the thickness of the WMH region based on the distance from the region labeled as the ventricle in the medical image to the region labeled as the WMH. Here, the distance standard can be defined in the normal direction from the boundary of the region labeled as the ventricle. Also, the maximum length of the calculated WMH thicknesses can be measured as the thickness of the WMH. More specifically, the second controller (2002) can determine the WMH region located closest to the boundary of the region labeled as the ventricle in the normal direction as a first point and the WMH region located farthest from the boundary as a second point, and determine the distance between the first point and the second point as the thickness (width) of the WMH region. Here, if the WMH region is connected (attached) to the ventricular region, the first point can be determined as one point on the ventricle.

[0123] That is, in other words, the second controller (2002) can generate a ray from the boundary of the ventricular region in the medical image in the normal direction, and calculate the thickness of the WMH by considering the junction between the generated ray and the pixel labeled as the WMH. Here, the junction between the ray and the pixel labeled as the WMH can include a first junction that measures the closest distance between the ventricular region and the WMH region, and a second junction that measures the furthest distance. That is, the second controller (2002) can calculate the thickness (width) of the WMH region by considering the distance between the first junction and the second junction. Here, it has been described above that the first junction can be present on the ventricular region.

[0124] Although the method using rays is preferably applied to WMHs located within the reference boundary described above, the present invention is not limited thereto, and the same method can be applied to WMHs outside the reference boundary.

[0125] FIG. 13 also illustrates a method for calculating area information for WMHs other than the reference boundary.

[0126] 13, the image analysis device 2000 can obtain region information of the target region using principal component analysis. Specifically, the second controller 2002 can extract at least two principal components of the target region using principal component analysis and calculate region information of the target region based on the extracted principal components.

[0127] For example, the second controller (2002) can extract principal components for at least two or more axes in the area labeled with the WMH in the medical image. Here, the principal components can be extracted in various ways, but it is preferable to extract them in the short axis and the long axis in the area labeled with the WMH. Once the principal components are extracted, the second controller (2002) can calculate the thickness of the area labeled with the WMH based on the extracted principal components.

[0128] That is, in other words, the second controller (2002) can perform a principal component analysis on the area labeled as a WMH to obtain the direction and size of the major and minor axes, and calculate the thickness of the area labeled as a WMH based on the length of the major axis. Alternatively, it can be expressed as the second controller (2002) calculating the width of the WMH area.

[0129] The above describes a method for calculating region information of a target region used to obtain a disease index from a medical image. It has also been described that the positional relationship between the target region and a reference region is also taken into consideration when obtaining the region information of the target region.

[0130] In the following, referring again to FIG. 10, a method for extracting a disease index based on the calculated WMH area information and reference boundaries will be described.

[0131] Referring again to FIG. 10, the image analysis device (2000) can calculate a disease index based on the reference boundary set in the medical image and area information of the target area (S2106).

[0132] Specifically, the second controller (2002) can calculate the disease index by comprehensively considering the association relationship between the reference boundary and the target area, and the area information of the target area.

[0133] Here, an example of the association relationship is whether or not the WMH is located inside or outside the reference boundary.

[0134] As a specific example, it is generally determined that the greater the thickness of the WMH, the more serious the disease; however, the method for extracting a disease index according to an embodiment can comprehensively take into consideration not only the thickness of the WMH but also whether the WMH is within or outside the reference boundary.

[0135] JPEG0007650530000001.jpg48170

[0136] JPEG0007650530000002.jpg77170

[0137] Table 1 shows an example of the criteria for calculating the disease index based on the subjective opinion of existing doctors.

[0138] Table 1 shows an example of a standard for calculating the Fazekas scale used in the existing medical industry. The Fazekas scale is a method for calculating disease indexes that has been published in various papers and is used in the medical industry as a credible method for calculating disease indexes. However, as shown in Table 1, in the existing medical industry, disease indexes are calculated based on vague standards and subjective judgments of doctors, which creates a problem in that objective diagnostic information or diagnostic support information cannot be provided to patients.

[0139] Table 2 shows the criteria for disease index calculation performed by the image analysis device of this embodiment.

[0140] In contrast, referring to Table 2, the image analysis device (2000) according to the embodiment can calculate the disease index according to a clear criterion. Specifically, the second controller (2002) can calculate the disease index from the medical image according to the criterion for calculating the disease index stored in the second memory (2400).

[0141] Here, the criteria for the image analysis device (2000) to calculate the disease index can be determined based on various diagnostic results or diagnostic assistance results previously obtained in the medical industry. That is, according to this embodiment, the image analysis device (2000) can calculate the disease index from the medical image using a quantified criterion as a result of analyzing a plurality of medical images previously obtained in the medical industry and having disease indexes calculated.

[0142] That is, the image analysis device (2000) can analyze a medical image, determine a disease index to be calculated according to a predetermined criterion, and calculate a grade of the disease index corresponding to the disease index whose calculation has been determined. Specifically, the second controller (2002) can determine a reference boundary from a reference area, determine a disease index to be calculated in consideration of a relationship between the determined reference boundary and a target area, and calculate a grade value corresponding to the determined disease index.

[0143] Here, the predetermined criterion for determining the disease index to be calculated may be the above-mentioned reference boundary. For example, the type of disease index calculated by the image analysis device 2000 may be changed depending on whether the target area is within the reference boundary or whether the target area is outside the reference boundary.

[0144] Here, in calculating the same disease index, there may be a criterion for distinguishing the grade of the disease index. In one example, the predetermined criterion is a criterion related to the area information of the target area.

[0145] Referring again to Table 2, the disease index according to the embodiment may include first to third disease indexes. The disease indexes shown in Table 2 refer to the Fazekas scale. The first disease index is a disease index related to peri-ventricular white matter, the second disease index represents a disease index related to deep white matter, and the third disease index may represent a total index calculated taking into account the first and second disease indexes.

[0146] Here, the reference boundary can be used to distinguish between the first disease index and the second disease index. That is, as described above, the second controller (2002) can use the reference boundary to calculate the first disease index from the target area within the reference boundary. The second controller (2002) can use the reference boundary to calculate the second disease index from the target area outside the reference boundary.

[0147] Also, it is possible to distinguish between grades within the same disease index, i.e., the second controller (2002) can calculate a grade value corresponding to a specific disease index from the medical image according to a predefined criterion.

[0148] Here, the predetermined criterion may relate to area information of the target area. In one example, if there is no cell indicating the target area or if it is determined that the number of cells is less than the predetermined number, the second controller (2002) may determine that the grade value of the specific disease index is 0. In another example, if the thickness of the target area is acquired to be a first value or more, the second controller (2002) may determine that the grade value of the specific disease index is 1. In yet another example, if the thickness of the target area is acquired to be a second value or more, the second controller (2002) may determine that the grade value of the specific disease index is 2. In yet another example, if the thickness of the target area is measured to be a third value or more or is connected to another target area, the second controller (2002) may determine that the grade value of the specific disease index is 3.

[0149] That is, referring to the example of Table 2, the second controller (2002) can calculate a first disease index based on the object area included within the reference boundary as a result of analyzing the medical image, and can calculate a grade value associated with the first disease index taking into account the area information of the object area included within the reference boundary.

[0150] In addition, the image analyzer (2000) can calculate all values ​​related to the first disease index and the second disease index within the same medical image. If the segmentation result of the medical image shows that all target regions are present both inside and outside the reference boundary, the second controller (2002) can calculate all grades related to the first disease index and the second disease index.

[0151] The image analysis device (2000) can also calculate a third disease index by taking into account the first disease index and / or the second disease index.

[0152] FIG. 14 shows an example of a display that provides a disease index and diagnostic support information based on the disease index in an embodiment.

[0153] 14, the image analysis device 2000 can provide auxiliary diagnostic information based on a disease index. Specifically, the second controller 2002 can obtain a disease index based on the analysis result of the medical image, and can provide auxiliary diagnostic information based on the obtained disease index.

[0154] To explain a specific example with reference to the drawings, the second controller (2002) can provide a disease index for a target area within a reference boundary based on the analysis result of the medical image, and can also provide information on what the disease index means based on the disease index.

[0155] As described above, the disease index refers to a scale (or standard) for the progression or severity of a disease that can be determined within a medical image. For example, if the standard for the disease index is the Fazekas scale, the disease index can be expressed as a grade on the Fazekas scale.

[0156] In addition, the auxiliary diagnostic information refers to disease-related information that can be determined based on the disease index. For example, when the disease index is calculated to have a specific grade using the Fazekas scale, the auxiliary diagnostic information refers to disease-related information that can be inferred from the disease index, such as the current degree of disease progression, the difference from the normal range, the difference from the age average, etc.

[0157] The diagnostic auxiliary information that can be obtained from the medical image and the diagnostic auxiliary information to be compared, for example, average values ​​by age, disease index values ​​of normal people, etc., can be obtained in advance and stored in the second memory (2400) and can be continuously updated.

[0158] As explained above, the image analysis device (2000) sets clear criteria, extracts disease indexes from medical images, and provides diagnostic auxiliary information based on the extracted disease indexes, thereby making it possible to obtain more objective and accurate diagnostic auxiliary information than disease-related information that has traditionally relied on the subjective judgment of doctors in the medical industry.

[0159] As described above, a method of using a predetermined reference boundary has been described as an embodiment of a method for providing auxiliary diagnostic information. However, in order to provide more accurate auxiliary diagnostic information, it may be necessary to modify the reference boundary in consideration of the characteristics of the medical image, rather than using a fixed reference boundary according to the characteristics of the medical image.

[0160] Hereinafter, a method of using a modified reference boundary will be described as another embodiment of a method of providing diagnostic auxiliary information with reference to the drawings.

[0161] First, with reference to FIG. 15, another embodiment for extracting disease indexes from medical images to obtain disease information will be generally described.

[0162] FIG. 15 is a flow chart illustrating an example of a method for utilizing modified reference boundaries to provide diagnostic assistance information, according to an embodiment.

[0163] Referring to FIG. 15, a method for providing diagnostic auxiliary information according to an embodiment includes a step of detecting a reference area from a segmented medical image (S2202), a step of setting a reference boundary based on the detected reference area (S2204), a step of determining whether the reference boundary overlaps with a region of non-interest (S2206), a step of modifying the reference boundary (S2208), and a step of obtaining a disease index based on the modified reference boundary (S2210).

[0164] The video analysis device (2000) performs operations similar to those performed in the embodiments described with reference to Figures 10 to 14, namely, a reference area detection step (S2202) and a reference boundary setting step (S2204) based on the detected reference area, and therefore detailed explanations are omitted here.

[0165] Once the reference boundary is set, the image analysis device (2000) can determine whether the set reference boundary overlaps with a region of non-interest in the segmented medical image (S2206). Specifically, the second controller (2002) can determine whether pixels corresponding to the set reference boundary in the medical image are included in pixels labeled with a value indicating the region of non-interest. Here, the region of non-interest is a gray matter region.

[0166] Here, a description will be given with reference to FIG.

[0167] FIG. 16 illustrates an overlap of the reference boundary and the region of non-interest according to an embodiment.

[0168] Referring to FIG. 16, a fiducial boundary is shown overlapping a grey matter labeled region in a medical image.

[0169] Generally, WMHs occur mainly in white matter, but are not found in gray matter regions. For this reason, if a disease index is extracted as it is in the case where the reference boundary invades the gray matter region, the disease index may be calculated inaccurately, and as a result, the diagnostic auxiliary information based on the disease index is also likely to be inaccurate. Therefore, when the reference boundary overlaps with the gray matter region, it is necessary to modify the reference boundary and calculate the disease index based on the modified reference boundary.

[0170] In the remainder of this specification, we will refer to regions where the target region (e.g., WMH) can be observed, such as white matter regions, as “regions of interest,” and to regions where the target region (e.g., WMH) cannot be observed, such as gray matter regions, as “regions of non-interest.”

[0171] Thus, the second controller (2002) can segment the medical image to set a reference boundary and then modify the reference boundary if the reference boundary overlaps with at least a portion of the area labeled with gray matter (S2208).

[0172] 15 again, if the reference boundary is modified, the image analysis device 2000 can obtain a disease index based on the modified reference boundary (S2210). The method of obtaining the disease index is similar to the embodiment described with reference to FIGS. 10 to 14, so a detailed description will be omitted.

[0173] An example of the correction of the reference boundary will be described below with reference to the drawings.

[0174] 17a and 17b show examples of how to calculate the modified reference boundary in this embodiment.

[0175] 17a and 17b, the image analysis device (2000) can modify the reference boundary. Specifically, the second controller (2002) can modify the reference boundary so that the modified reference boundary is located between an area labeled as a non-interest area and an area labeled as a reference area. Here, the reference boundary can be located so as to divide the distance between the reference area and the non-interest area at a predetermined ratio. Here, the predetermined ratio can be set in consideration of various factors. For example, the predetermined ratio can be determined in consideration of the distance between the non-interest area and the reference area. That is, the second controller (2002) can modify the reference boundary by reflecting various factors that can be considered in the medical image, including the distance between the reference area and the non-interest area, the position of the reference boundary overlapping with the non-interest area, and / or the shape of the reference area.

[0176] 17A, the distance between the non-interested region (gray matter) and the reference region (ventricle) is relatively close, and the reference boundary is relatively overlapped with a large portion of the non-interested region. In this case, the second controller can correct the reference boundary in consideration of the distance between the non-interested region and the reference region, so that the corrected reference boundary is located halfway between the gray matter region and the ventricle region (where the distance between the ventricle and the gray matter is divided 5:5).

[0177] 17b, the distance between the non-interested region (gray matter) and the reference region (ventricles) is relatively far, and the reference boundary overlaps a relatively small portion of the non-interested region. In this case, the second controller (2002) can modify the reference boundary so that the modified reference boundary is closer to the non-interested region.

[0178] In addition, according to the present embodiment, a reference boundary determined in one medical image may overlap with a plurality of regions of non-interest, i.e., for example, the situations shown in Fig. 17(a) and Fig. 17(b) may occur in one medical image.

[0179] In such a case, the second controller (2002) can modify the reference boundary such that the first and second sections overlapping the non-interest area in the reference boundary have different ratios. Exemplarily, the second controller (2002) can modify the reference boundary in the first section as shown in FIG. 17a, and in the second section as shown in FIG. 17b.

[0180] Expressed differently, the second controller (2002) can set the ratio of the distance from the reference region to the reference boundary to the distance from the reference boundary to the region of non-interest to a first value at a position where the distance between the region of non-interest and the reference region is a first distance, and can correct the reference boundary so that the ratio of the distance from the reference region to the reference boundary to the distance from the reference boundary to the region of non-interest to a second value at a position where the distance between the region of non-interest and the reference region is a second distance. Here, when the second distance is greater than the first distance, the reference boundary can be corrected so that the second value is smaller than the first value. For example, the farther the distance between the reference region and the region of non-interest is, the more the corrected reference boundary can be positioned toward the region of non-interest.

[0181] As a specific example, the second controller (2002) may modify the reference boundary so that the reference boundary divides the distance between the gray matter and the ventricle at a ratio of 5:5 in the case of the first region where the distance between the gray matter and the ventricle is relatively short, and may modify the reference boundary so that the reference boundary divides the distance between the gray matter and the ventricle at a ratio of 3:7 in the case of the second region where the distance between the gray matter and the ventricle is relatively far. Here, the ratio for dividing the distance between the gray matter and the ventricle may be preset.

[0182] The second controller (2002) can also modify the reference boundary only for the area where the reference boundary overlaps with the non-interest area, or can modify the entire reference boundary. In addition, the second controller (2002) can also modify the reference boundary for the added buffer area in addition to the area where the reference boundary overlaps with the gray matter area. This will be described in detail with reference to FIG. 18.

[0183] FIG. 18 shows an example of a partially modified reference boundary and a fully modified reference boundary according to an embodiment.

[0184] 18, the image analysis device 2000 can modify all or a part of the reference boundary. Specifically, the second controller 2002 can modify all or a part of the reference boundary in consideration of the relationship between the reference area and the non-interest area.

[0185] For example, if the distance between the reference region and the non-interest region is close only in a specific portion, the reference boundary is modified only in a portion of the reference boundary corresponding to the specific portion, and if the distance between the non-interest region and the reference region is similarly measured overall, the reference boundary can be modified generally. However, this is merely an example, and there may be various situations in which the reference boundary is modified overall or partially.

[0186] An example will now be described with reference to the drawings, in which a fully modified reference boundary and a partially modified reference boundary are shown.

[0187] An example in which the reference boundary is generally modified will be described first.

[0188] The image analysis device 2000 can globally correct the reference boundary. Specifically, the second controller 2002 can globally correct the reference boundary in consideration of the distance between the reference region and the non-interest region.

[0189] The second controller (2002) can modify the entire reference boundary in a region where the reference boundary and the region of non-interest overlap, corresponding to the modification of the reference boundary.

[0190] Alternatively, the second controller (2002) may modify the reference boundary taking into account the distance between the reference region and the region of non-interest in all sections of the reference boundary, not just in sections where the reference boundary and the region of non-interest overlap.

[0191] For example, it may be expressed that the second controller (2002) can generally modify the reference boundary to correspond to a position that divides the distance between the non-interest area and the reference area in the first domain, taking into account the distance between the non-interest area and the reference area in the first domain.

[0192] The second controller (2002) can also globally modify the reference boundary for all sections of the reference boundary, taking into account the distance between the general area of ​​the reference region and the area of ​​non-interest.

[0193] Also, the image analysis device 2000 can modify only a portion of the reference boundary. Specifically, the second controller 2002 can modify the reference boundary only for a portion of the section where the reference boundary and the region of non-interest overlap. Here, the second controller 2002 can modify the reference boundary even for a portion where a buffer section is added to the section where the reference boundary and the region of non-interest overlap, and can modify the reference boundary only for a section where the portion of the section is removed.

[0194] In other words, the second controller (2002) can determine the reference boundary for a section where the distance between the reference boundary and the non-interest area is closer than a predetermined distance such that the reference boundary and the reference area have a first distance, and can determine the reference boundary for a section where the distance between the reference boundary and the non-interest area is farther than the predetermined distance such that the reference boundary and the reference area have a second distance. Here, the predetermined distance can correspond to the distance between the reference boundary and the reference area described with reference to Figures 10 to 14.

[0195] Referring to the drawings, the partially corrected reference boundary, the globally corrected reference boundary 1, and the globally corrected reference boundary 2 are all shown to have similar shapes in the sections where the non-interest area (gray matter) and the reference area are close to each other. However, the globally corrected reference boundary 1 is often seen to be separated from the reference area by about the same distance even in sections where the distance between the non-interest area and the reference area is farther. In contrast, the globally corrected reference boundary 2 is determined so that the reference boundary is located at a position that divides the non-interest area and the reference area at a predetermined ratio in all sections of the reference boundary.

[0196] However, even in the case of a partially modified reference boundary, it is necessary to smoothly connect the modified area of ​​the reference boundary with the existing reference boundary so that the modified reference boundary maintains a soft shape.

[0197] Hereinafter, a connection relationship between an existing reference boundary and a modified reference boundary when a part of the reference boundary is modified will be described with reference to the drawings.

[0198] FIG. 19 shows an example of a specific modification process of a partially modified reference boundary according to an embodiment.

[0199] 19, the image analysis device 2000 can connect the modified reference boundary to the existing reference boundary. Specifically, the second controller 2002 can determine the range of the reference boundary to be modified by considering the distance between the reference area and the non-interest area, and can connect the existing reference boundary from a portion of the modified reference boundary.

[0200] First, the second controller (2002) can modify the reference boundary from a first point where the distance between the reference area and the non-interest area is equal to or less than the critical distance to a second point where the distance between the reference area and the non-interest area is equal to or less than the critical distance. However, this is merely an example, and the first point can be selected as any point where the reference boundary and the non-interest area overlap, or any point where the distance between the non-interest area and the reference area is equal to or less than a predetermined distance. In addition, the method of modifying the reference boundary can be the above-mentioned embodiment or any other method integrated with the concept of the present invention.

[0201] Thereafter, the second controller 2002 may connect the first or second point, which is the end point of the area where the reference boundary has been modified, to the third point, which is the section where the existing reference boundary begins again. Here, the distance between the third point and the reference area according to the existing reference boundary may be predetermined as described with reference to FIGS. 10 to 14.

[0202] The second controller (2002) may connect the first or second point with the third point by interpolating between the position value of the first or second point and the position value of the third point. Here, the length of the interpolation section between the second and third points may be set in various ways. For example, the interpolation section may be set in consideration of the distance between the non-interest area and the interest area.

[0203] A specific example will be described with reference to the drawings.

[0204] The second controller (2002) may partially modify the reference boundary for an area where the distance between the gray matter and the ventricles is close or where the existing reference boundary overlaps with the gray matter region. The second controller (2002) may interpolate between a first point located on one side of the modified reference boundary and a third point located on the existing reference boundary. Here, the second point may be selected as a point where the distance between the gray matter and the ventricles is equal to or greater than a critical distance. Alternatively, the second point may be selected as a point where the distance between the modified reference boundary and the region of non-interest is equal to or greater than a critical distance.

[0205] In this manner, the image analysis device (2000) according to the embodiment can set a reference boundary for calculating a disease index more clearly by smoothly connecting the existing reference boundary and the modified reference boundary, and the accuracy of the disease index obtained by the image analysis device (2000) can be improved due to the clearly set reference boundary.

[0206] If the disease index is calculated using the modified reference boundary, the image analysis device (2000) can provide auxiliary diagnostic information based on the disease index obtained using the modified reference boundary. Specifically, the second controller (2002) can obtain a disease index from the analysis result of the medical image using the modified reference boundary, and can provide auxiliary diagnostic information based on the obtained disease index. The details of the provided disease index or disease information are similar to those described in FIG. 14, so detailed description will be omitted.

[0207] As explained above, by using a modified reference boundary, a more accurate disease index can be obtained from a wider variety of images than when a uniform reference boundary is used.

[0208] In addition, although the above description has been given of modifying the reference boundary only when the reference boundary overlaps with the gray matter region, it is also possible to set the reference boundary from the beginning as described in Figures 15 to 19, and an appropriate reference boundary setting method can be selected according to the requirements of the field.

[0209] The above description of the present specification has primarily been directed to a method for analyzing a medical image.

[0210] In the following, a method for providing auxiliary diagnostic information from a plurality of medical images when the image analyzing device (2000) analyzes a plurality of medical images will be described.

[0211] FIG. 20 is a schematic flow chart of a method for the image analysis device (2000) to provide auxiliary diagnostic information from a plurality of medical images.

[0212] As shown in FIG. 20, the method for providing diagnostic auxiliary information from multiple medical images includes a step of acquiring multiple medical images (S3200), a step of acquiring disease indexes from the multiple medical images (S3400), and a step of providing diagnostic auxiliary information based on the multiple disease indexes (S3600).

[0213] First, the image analysis device (2000) can acquire a plurality of medical images (S3200). Specifically, the second controller (2002) can acquire the plurality of medical images from the image acquisition device (1000) through the second communication module (2800). Here, the plurality of medical images is a collection of tomographic images of an object to be imaged. For example, the plurality of medical images is a collection of MRI slice images of the object to be imaged.

[0214] Hereinafter, the image analysis device (2000) can acquire disease indexes from a plurality of medical images (S3400). Specifically, the second controller (2002) can acquire disease indexes for the plurality of medical images by using a program for acquiring disease indexes stored in the second memory (2400). Here, the second controller (2002) can acquire disease indexes for each of the plurality of medical images, or can acquire disease indexes for some of the plurality of medical images. The method of acquiring disease indexes is similar to that described above, and therefore a detailed description will be omitted.

[0215] If a plurality of disease indexes are acquired, the image analysis device (2000) can provide auxiliary diagnostic information based on the plurality of disease indexes (S3600). Specifically, the second controller (2002) can acquire and provide auxiliary diagnostic information by comprehensively considering the disease indexes related to the acquired plurality of medical images. Here, the second controller (2002) can provide auxiliary diagnostic information for all of the plurality of medical images for which disease indexes have been calculated, or can provide auxiliary diagnostic information for only some of the plurality of medical images for which disease indexes have been calculated. The method of providing auxiliary diagnostic information for a plurality of medical images by the image analysis device (2000) will be described in detail later.

[0216] Hereinafter, examples of a plurality of medical images and a method for providing auxiliary diagnostic information based on the plurality of medical images according to an embodiment will be described with reference to the accompanying drawings.

[0217] FIG. 21 illustrates an example of multiple medical images according to an embodiment.

[0218] 21, the image analysis device 2000 can obtain a plurality of cross-sectional images of the photographed object. Specifically, the second controller 2002 can obtain a plurality of medical images including a plurality of cross-sectional images of the photographed object through the second communication module 2800. Here, the plurality of cross-sectional images are composed of a set of slice images, as described above.

[0219] FIG. 21 illustrates an example of a number of medical images acquired with an MRI imaging device.

[0220] 21, the medical images include slice images on a plurality of planes parallel to a first axis of the object to be imaged. Also, each slice image may include information on the object to be imaged at a position spaced apart from each other by a predetermined distance. That is, each slice image includes different information on the object to be imaged. In this way, the medical images collectively include information on various cross sections of the object to be imaged, and the image analysis device (2000) may obtain auxiliary diagnostic information on the object to be imaged in general by analyzing the medical images.

[0221] FIG. 22 shows an example of slice images each containing other information and an example of how this provides auxiliary diagnostic information.

[0222] 22 shows an example of a plurality of medical images of a single object taken along different slices, in particular, a collection of various slice images of an MRI image of a human brain.

[0223] (a) shows an MRI slice image of an upper cross section of the brain, and (b) shows an MRI slice image of a lateral cross section of the brain.

[0224] Compared to (b), (a) is an image of a cross section of the upper brain region, and the size of the ventricles is smaller than that of (b). Also, the white matter and gray matter regions in (a) and (b) are depicted differently, and most importantly, they contain different regional information distinguished by WMH.

[0225] As described above, the disease index of this embodiment can be obtained based on the relationship between the region information of the reference region and the target region. Therefore, different disease indexes can be calculated from each of a plurality of medical images including cross sections at various positions of the photographed subject. As a specific example, when the upper cross section of the brain is photographed (a), a disease index of 3 grade is calculated, and when the lateral cross section of the brain is photographed (b), a disease index of 2 grade is calculated.

[0226] In this case, both auxiliary diagnostic information regarding (a) and auxiliary diagnostic information regarding (b) are obtained, and the image analysis device (2000) can obtain and provide auxiliary diagnostic information for all images or auxiliary diagnostic information for some images.

[0227] That is, the image analysis device (2000) according to the embodiment can calculate disease indexes for a plurality of medical images obtained by photographing various cross sections of a subject, and can provide auxiliary diagnostic information based on the calculated disease indexes.

[0228] In addition, the image analyzer (2000) can provide auxiliary diagnostic information for all medical images for which a disease index has been calculated, or can provide auxiliary diagnostic information for only some of the medical images.

[0229] The image analysis device (2000) can select some images from the medical images for which the disease indexes have been calculated, and provide auxiliary diagnostic information for the selected medical images. Specifically, the second controller (2002) can select images including predetermined information from the medical images for which the disease indexes have been calculated, and calculate disease indexes for the images including the predetermined information to provide auxiliary diagnostic information.

[0230] Here, there are various criteria for selecting a medical image for providing auxiliary diagnostic information. That is, the image analyzing device 2000 may select a medical image for providing auxiliary diagnostic information based on the type or grade of a disease index.

[0231] For example, the image analysis apparatus (2000) may provide auxiliary diagnostic information for a medical image that is the basis for calculating a disease index level that indicates the most seriousness of the disease progression among the medical images in which the disease index is calculated. As another example, the image analysis apparatus (2000) may provide auxiliary diagnostic information for a medical image that is the basis for calculating a disease index level that indicates the most improvement in the disease state among the medical images in which the disease index is calculated. As yet another example, the image analysis apparatus (2000) may provide auxiliary diagnostic information for a medical image that is the basis for calculating a disease index level that indicates the average (or median) of the disease index calculated from a plurality of medical images among the medical images in which the disease index is calculated.

[0232] As another example, the image analysis device 2000 may provide auxiliary diagnostic information for a medical image in which a first disease index is measured to be high, or for a medical image in which a second disease index is measured to be high, or may provide auxiliary diagnostic information by combining these. It is also possible that the image analysis device 2000 may select a medical image for providing auxiliary diagnostic information, taking into account the type and grade of the disease index.

[0233] In this manner, since other auxiliary diagnostic information may be calculated depending on the type and grade of the disease index, the image analysis device (2000) can provide auxiliary diagnostic information suited to the patient by taking into consideration the characteristics of the photographed subject (e.g., mainly a patient) and the type and grade of the disease index and selecting a medical image for providing auxiliary diagnostic information.

[0234] In another embodiment, the image analysis device (2000) may select candidate images for calculating a disease index from a plurality of medical images, and provide diagnostic auxiliary information based on the disease index obtained from the selected candidate images.

[0235] Hereinafter, a method for providing auxiliary diagnostic information based on a disease index obtained from a candidate image by the image analyzing device (2000) will be described with reference to the drawings.

[0236] FIG. 23 is a flow chart that generally illustrates a method for obtaining auxiliary diagnostic information from a candidate image.

[0237] Referring to FIG. 23, a method for the image analysis device (2000) to obtain diagnostic auxiliary information from a candidate image may include a step of obtaining a plurality of medical images (S1202), a step of segmenting the plurality of medical images (S1212), a step of determining a candidate image from the plurality of segmented images (S1222), a step of obtaining a disease index for the selected candidate image (S1402), and a step of providing diagnostic auxiliary information based on the obtained disease index (S1602).

[0238] First, the image analysis device (2000) can acquire (S1202) a plurality of medical images from the image acquisition device (1000). The step of acquiring a plurality of medical images (S1202) is similar to the explanation in FIG. 20, so a detailed explanation will be omitted.

[0239] Thereafter, the image analysis device 2000 can segment the medical images (S1212). The segmentation of the medical images can be performed in a similar manner to that described above, and therefore a detailed description thereof will be omitted.

[0240] Once segmentation is performed, the image analysis device (2000) can determine candidate images from the segmented medical images (S1222). More specifically, the second controller (2002) can determine candidate images that satisfy a predetermined criterion from the segmented medical images according to a predetermined criterion. A detailed description of the candidate images will be given below with reference to the drawings.

[0241] When the candidate image is determined, the image analysis device (2000) can obtain a disease index from the candidate image (S1402). Specifically, the second controller (2002) can obtain a disease index for all or a part of the candidate image by using a program for disease index calculation stored in the second memory (2002). The method of obtaining the disease index can be performed in a similar manner to the above description, and therefore a detailed description will be omitted.

[0242] Once the disease index is acquired, the image analysis device (2000) can acquire and provide auxiliary diagnostic information based on the acquired disease index (S1602). Specifically, the second controller (2002) can provide auxiliary diagnostic information based on the disease index acquired from one or more candidate images. Here, the method of the second controller (2002) providing auxiliary diagnostic information from a plurality of candidate images can be the method of providing auxiliary diagnostic information from a plurality of medical images as described in Figures 20 to 22.

[0243] Hereinafter, an example in which a candidate image is extracted from a plurality of medical images and an example in which a disease index is calculated from the extracted candidate image to provide auxiliary diagnostic information will be described with reference to the drawings.

[0244] FIG. 24 shows an example of candidate images according to an embodiment.

[0245] 24, the image analysis device 2000 can select candidate images. Specifically, the second controller 2002 can select one or more candidate images from among a plurality of medical images acquired from the image acquisition device 1000. Here, the candidate images are images that satisfy a predetermined criterion from among the plurality of medical images.

[0246] Here, the predetermined criterion may be determined to correspond to the disease index to be calculated. As an example, the predetermined criterion is whether or not the reference region is inclusive. That is, the second controller (2002) may select an image including information on the reference region as a candidate image from among the plurality of medical images. As another example, the predetermined criterion is whether or not the object region is inclusive. That is, the second controller (2002) may select an image including information on the object region as a candidate image from among the plurality of medical images. Also, the second controller (2002) may select a medical image including both the object region and the reference region as a candidate image.

[0247] Also, the candidate image may be determined as one or more images included in a candidate image range as shown in the drawings. Here, the candidate image range refers to a range from a first medical image satisfying a predetermined condition to a second medical image satisfying a predetermined condition among tomographic medical images having a continuous order obtained by photographing the same photographed object on various planes relative to one axis. Here, the range from the first medical image to the second medical image may include a plurality of medical images sequentially located between the first medical image and the second medical image.

[0248] That is, to describe a specific example with reference to the drawings, when the reference region is the ventricular region, among a plurality of medical images taken in successive cross sections of a transverse plane relative to the saggital axis, the range from a first medical image located at the bottom among the medical images including the ventricles to a second medical image located at the top among the medical images including the ventricles can be set as the candidate image range.

[0249] In another embodiment, the second controller (2002) may select a candidate image or range of candidate images based on whether the number of pixels corresponding to a predefined condition contained in the medical image is below or above a threshold value.

[0250] Specifically, the second controller (2002) can analyze the segmentation result of the medical image to determine whether the number of pixels labeled as ventricles is less than or equal to a threshold value. If the number of pixels labeled as ventricles included in the medical image is equal to or greater than the threshold value, the second controller (2002) can determine the medical image as a candidate image.

[0251] In addition, when the second controller (2002) sets the range of candidate images, if the analysis result of two adjacent medical images shows that the number of pixels labeled as the ventricles in one image is less than or equal to a critical value and the number of pixels labeled as the ventricles in the other image is greater than or equal to the critical value, the other image can be set as the boundary of the range of candidate images.

[0252] FIG. 25 illustrates an example of providing diagnostic auxiliary information from a candidate image according to an embodiment.

[0253] 25, an image analyzing device 2000 can provide auxiliary diagnostic information from a candidate image. Specifically, a second controller 2002 can obtain a disease index from the candidate image and provide auxiliary diagnostic information based on the obtained disease index.

[0254] The method for calculating the disease index is as described above, and therefore a detailed description thereof will be omitted.

[0255] As shown in FIG. 25, (a) is a medical image that is not a candidate image, and (b) is a medical image that has been determined as a candidate image.

[0256] In calculating the disease index according to this embodiment, it is important that a specific region of the photographed subject is included in the medical image. For example, the method of obtaining the disease index according to the embodiment can be derived based on the correlation between the ventricles and the WMH, but in this case, if the ventricles are not included in the medical image, it is difficult to calculate the disease index, and therefore accurate diagnostic auxiliary information may not be provided. That is, in the case of (a) of FIG. 25, the medical image is taken of the upper end of the brain, and the ventricle region is not included in the medical image, so the method of calculating the disease index according to the embodiment may not be applicable.

[0257] In the case of (b), the ventricular region is included in the medical image, and the second controller (2002) can analyze the medical image such as (b) to calculate a disease index, and can obtain diagnostic auxiliary information based on the calculated disease index.

[0258] Furthermore, when analyzing all medical images of the photographed subject, the image analyzer (2000) may have an excessive amount of calculations, slowing down the calculation speed, and thus causing errors during the calculation process.

[0259] However, according to the method of providing diagnostic auxiliary information using candidate images of this embodiment, after segmenting multiple images, candidate images for calculating the disease index are first determined, and the disease index is calculated only for the determined candidate images, thereby reducing the amount of calculation, thereby increasing the calculation speed and improving the accuracy of the calculation.

[0260] FIG. 26 is another implementation of providing auxiliary diagnostic information according to an embodiment.

[0261] 26, an image analyzing apparatus 2000 according to an embodiment can provide auxiliary diagnostic information for at least one image among the candidate images. Specifically, a second controller 2002 can provide auxiliary diagnostic information for at least one image that satisfies a predetermined condition among the medical images selected as the candidate images.

[0262] Here, the predetermined condition may be determined taking into consideration the disease index. As an example, the predetermined condition may be a disease index value indicating the most severe disease progression. That is, the second controller (2002) may select a particular medical image, the disease index of which indicates the most severe disease progression among the disease indexes of the candidate images, as a medical image for providing auxiliary diagnostic information.

[0263] In addition, when multiple disease indexes can be calculated from a medical image, the predetermined condition may be determined by taking into consideration all of the disease indexes. That is, the second controller (2002) may select all of the specific candidate images for each of the multiple disease indexes as medical images for providing diagnostic auxiliary information, or may select only specific candidate images for some of the multiple disease indexes as medical images for providing diagnostic auxiliary information.

[0264] To explain a specific example with reference to the drawings, the second controller (2002) can provide auxiliary diagnostic information related to a medical image that satisfies a predetermined condition related to a first disease index among a plurality of medical images through the display module (2600). Here, the predetermined condition is whether or not the grade of the first disease index is calculated to be the highest among a plurality of candidate images. In addition, the second controller (2002) can provide auxiliary diagnostic information related to a medical image that satisfies a predetermined condition related to a second disease index among a plurality of medical images through the display module (2600), and can also provide auxiliary diagnostic information for all medical images that satisfy the predetermined conditions related to the first disease index and the second disease index.

[0265] As described above, the method of determining a candidate image for calculating a disease index from a plurality of medical images and the method of calculating a disease index related to one medical image have been mainly described.

[0266] However, in order to calculate a more accurate disease index, a plurality of medical images may be used to calculate the disease index, which will be described with reference to FIGS.

[0267] FIG. 27 is a flow chart illustrating a method for obtaining a disease index based on multiple images according to an embodiment.

[0268] Referring to FIG. 27, a method for obtaining a disease index based on multiple images according to an embodiment includes a step of obtaining multiple medical images (S4200), a step of performing segmentation on the multiple medical images (S4400), a step of obtaining target area information of a first image (S4600), and a step of obtaining a disease index of a second image based on the target area information of the first image (S4800).

[0269] First, the steps of acquiring a plurality of medical images (S4200) and performing segmentation on the plurality of medical images (S4400) can be performed similarly to the operations described above, and therefore a detailed description thereof will be omitted.

[0270] stomach After performing segmentation on the medical image, the image analysis device (2000) can obtain information on a target area for the first image (S4600). Specifically, the second controller (2002) can obtain information on a target area corresponding to a disease index to be calculated. Here, the target area means an area that is essential for calculating a specific disease index. The specific details will be described later.

[0271] If the information on the target area is acquired, the image analysis device (2000) can calculate a disease index of the second image based on the target area information of the first image. Specifically, the second controller (2002) can calculate a disease index related to the second image based on the target area information acquired from the first image and area information related to the photographed object included in the second image. An example of this will be described later.

[0272] Hereinafter, a method for calculating a disease index of a second image by considering information of a first image will be described with reference to a specific example with reference to the drawings.

[0273] FIG. 28 is an example showing a process of calculating a disease index of a second image by considering information of a first image according to an embodiment of the present invention.

[0274] According to this embodiment, the image analysis device (2000) can directly calculate the disease index of the second image from the first image. Specifically, the second controller (2002) can calculate the disease index of the second image based on the relationship between the first region information included in the first image and the second region information included in the second image. Here, the first region information is information about a reference region included in the first image, and the second region information is information about a target region included in the second image.

[0275] To calculate the disease index, the medical image may not contain information about the reference region, but may contain information about the target region, in which case the requirements for calculating the disease index are not sufficient, making it difficult to calculate the disease index, but it may be necessary to calculate the disease index for such medical images in some cases.

[0276] For example, referring to FIG 28, the first slice image includes information on the WMH but does not include information on the ventricles. In this case, it is difficult to derive a disease index derived from the relationship between the ventricles and the WMH from the first slice image. However, since the WMH is likely to contain important information on the disease, it is desirable to calculate a disease index for an image in which the WMH is found in order to provide more accurate diagnostic auxiliary information.

[0277] In this case, the second controller (2002) can calculate the disease index of the second image based on the relationship between the first image and the second image and taking into account information about the first image. Here, the relationship between the first image and the second image can be acquired in advance. As an example, the relationship between the first image and the second image is the distance between the slice images.

[0278] Further, specific examples will be described with reference to the drawings.

[0279] The second controller (2002) may first determine a first image that includes a target region for calculating a disease index in the medical image but does not include a reference region as a result of segmentation of the multiple medical images. For example, the second controller (2002) may select a first image that includes only information about the WMHs and does not include information about the ventricles in the medical image, even though information about the ventricles and the WMHs is required to calculate a disease index according to an embodiment.

[0280] Next, the second controller (2002) can select a second image including a reference region not included in the first image from among other medical images in the vicinity of the first image, where the second image is an image immediately adjacent to the first image, but is not limited thereto. For example, the second controller (2002) can select a second image including a ventricular region for calculating a disease index from among medical images located in the periphery of the first image (e.g., a distance from the first image in the saggital axis within a predetermined distance).

[0281] When the second image is selected, the second controller (2002) can calculate a disease index based on the relationship between the reference region included in the second image and the target region included in the first image. For example, the second controller (2002) can calculate a disease index in the second image based on distance information from the ventricular region included in the second image to the WMH included in the first image and region information of the WMH included in the second image. Here, the distance between the first image and the second image can be obtained and stored in advance.

[0282] In addition, here, a criterion for deriving a relationship between a reference region included in the second image and a target region included in the first image can be determined in various ways. As an example, the second controller (2002) can calculate a disease index based on a distance from a center of a ventricular region included in the second image to a WMH region included in the first image. As another example, the second controller (2002) can calculate a disease index based on a distance from a ventricular region that is closest to the WMH region included in the first image among the ventricular regions included in the second image to the WMH region included in the first image. It can be understood by a person skilled in the art that various methods other than the exemplified method can be integrated into the concept of the present invention as long as the criterion is capable of calculating the distance from the ventricular region included in the second image to the WMH region included in the first image.

[0283] Hereinafter, a method for calculating a disease index of a second image indirectly from a first image will be described with reference to the drawings.

[0284] FIG. 29 shows another example of a process for calculating a disease index of a second image by taking into account information of a first image according to this embodiment.

[0285] According to this embodiment, the image analysis device (2000) can indirectly calculate the disease index of the second image from the first image. Specifically, the second controller (2002) can calculate the disease index of the second image based on the relationship between the reference cell in the second image, which is indirectly obtained from information on the reference region included in the first image, and the target region included in the second image.

[0286] A specific example will be described with reference to the drawings.

[0287] The second controller (2002), as described in FIG. 27, can first determine a first image as a result of segmentation of a plurality of medical images, the first image including a target region for calculating a disease index within the medical image but not including a reference region.

[0288] For example, the second controller (2002) may select a first image in which the medical image contains only information about the WMHs and no information about the ventricles, even though information about the ventricles and WMHs is required to calculate a disease index according to an embodiment.

[0289] Next, the second controller (2002) can select a second image from other medical images adjacent to the first image that includes a reference region not included in the first image. For example, the second controller (2002) can select a second image from medical images located near the first image that includes a ventricular region for calculating a disease index.

[0290] When the second image is selected, the second controller (2002) can determine a reference cell on the first image that corresponds to the reference area included in the second image based on information about the reference area included in the second image. For example, the second controller (2002) can determine a reference cell on the first image that corresponds to the ventricular area included in the second image based on information about the ventricular area included in the second image.

[0291] If the reference cell is determined, the second controller (2002) can calculate a disease index based on the reference cell included in the first image and information on the target region. That is, for example, the second controller (2002) can calculate a disease index by considering the reference cell of the second region corresponding to the ventricular region included in the first image as the ventricle, deriving the relationship with the WMH included in the first image. It is natural that a disease index calculation method similar to that described above can be applied to the method of calculating a disease index based on the relationship between the reference cell and the WMH.

[0292] As described above, according to this embodiment, the image analysis device (2000) can calculate a disease index using a plurality of medical images to obtain an accurate disease index, and can provide auxiliary diagnostic information based on the calculated disease index.

[0293] In addition, the diagnostic auxiliary information providing system (10000) according to the embodiment can calculate a disease index even in a 3D medical model obtained by 3D modeling a medical image, and provide diagnostic auxiliary information.

[0294] Hereinafter, a method for providing auxiliary diagnostic information based on a 3D medical model will be described with reference to FIGS.

[0295] FIG. 30 is a flowchart illustrating an example of a method for providing diagnostic auxiliary information based on a 3D medical model according to an embodiment.

[0296] According to this embodiment, the method for providing diagnostic auxiliary information based on a 3D medical model performed by the image analysis device (2000) includes a step of acquiring a plurality of medical images (S5200), a step of performing segmentation on the plurality of medical images (S5400), a step of performing 3D modeling based on the segmented medical images (S5600), and a step of acquiring a disease index based on the 3D modeling (S5800).

[0297] The step of acquiring a plurality of medical images (S5200) and the step of performing segmentation on the plurality of medical images (S5400) may be performed in a manner similar to that described above, and detailed description thereof will be omitted.

[0298] When the segmentation is performed on the plurality of medical images, the image analysis device (2000) can obtain a 3D medical model based on the plurality of segmented medical images (S5600). Specifically, the second controller (2002) can generate a 3D medical model by processing a plurality of medical images obtained by photographing the object at successive cross sections. Here, the 3D medical model is generated based on the plurality of segmented medical images and can include 3D pixel information reflecting a plurality of brain region information. That is, the brain can be segmented into a plurality of brain regions in three dimensions. Also, the reference boundary described above can be formed in three dimensions.

[0299] When the 3D medical model is acquired, the image analysis device (2000) can acquire a disease index based on the acquired 3D medical model and provide auxiliary diagnostic information. Specifically, the second controller (2002) can calculate a disease index from the 3D medical model using a program for acquiring a disease index stored in the second memory (2400) and acquire auxiliary diagnostic information from the calculated disease index.

[0300] An example of calculating a disease index from a 3D medical model to provide auxiliary diagnostic information will be described below with reference to the drawings.

[0301] FIG. 31 shows an example of a process for calculating a disease index from a 3D medical model according to this embodiment, and FIG. 32 shows another example of a process for calculating a disease index according to this embodiment.

[0302] Referring to FIG. 31, the image analysis device (2000) can obtain a disease index based on region information included in a 3D medical model. Specifically, the second controller (2002) can calculate a disease index based on one or more specific region information of an imaged object included in a 3D medical model. Here, the one or more specific regions can include a target region and a reference region as described above. Specifically, the target region or the reference region is a cell (e.g., a voxel) labeled with the target region or the reference region expressed in three dimensions. Also, here, the second controller (2002) can set a criterion for deriving a relationship between regions included in one or more imaged objects.

[0303] As an example, the second controller (2002) can calculate a disease index based on the ventricular region and the WMH region included in the 3D medical model. Here, the second controller (2002) can set a reference boundary to derive a relationship between the ventricular region and the WMH region, similar to the above example. That is, the second controller (2002) can set a set of cells that is a preset distance away from the ventricular region in three dimensions as the reference boundary. Also, similar to the reference boundary set in the medical image described above, the second controller (2002) can set or modify the reference boundary in various ways.

[0304] The second controller (2002) can calculate a disease index based on the reference boundary formed in three dimensions and taking into consideration the relationship with the WMH region. Here, the relationship between the reference boundary and the WMH is similar to that described above. In addition, by performing three-dimensional 3D modeling on a two-dimensional medical image, information on the region of the photographed object, for example, region information on the target region, becomes a three-dimensional physical quantity. That is, the second controller (2002) can also obtain three-dimensional information such as the thickness of a specific region included in the photographed object, which cannot be obtained from a two-dimensional medical image, through 3D modeling. As another example, information on the WMH is the volume (or number) corresponding to a cell tagged in the WMH.

[0305] 32, the image analysis device 2000 can calculate a disease index based on a plurality of planes included in the 3D medical model to provide auxiliary diagnostic information. Specifically, the second controller 2002 can calculate a disease index based on one or more planes including information on at least one specific region of the photographed object included in the 3D medical model to provide auxiliary diagnostic information.

[0306] By performing 3D modeling, it is possible to easily derive the relationship between specific regions for calculating the disease index. That is, as described above, when calculating the disease index by analyzing the slice image, if the slice image does not contain information about the required region included in the photographed object for calculating the disease index, not only is it difficult to obtain the disease index, but the accuracy of the obtained disease index may be low.

[0307] However, when performing 3D modeling to calculate the disease index, the accuracy of the disease index can be improved by determining a plane so that all areas required for calculating the disease index are included, and then calculating the disease index for that plane.

[0308] A specific example will be described with reference to the drawings.

[0309] Referring again to FIG. 32, the second controller (2002) can set the first plane to correspond to a preset criterion so that the ventricular region and the first WMH region are both included, and calculate a disease index for the set first plane. Here, there are various criteria for setting the first plane. As an example, the first plane is a plane set in consideration of the distance between the target region and the reference region. As a more specific example, the first plane is a plane set to include a straight line that minimizes the distance between the WMH region and the ventricular region.

[0310] As another example, the second controller (2002) can set the second plane taking into account the number of target regions included in the plane and calculate the disease index for the set second plane. For example, the second controller (2002) can set the second plane to include the largest number of target regions in addition to including information about the reference region. The second controller (2002) can also set the second plane to maximize the number of labeled pixels in the target regions.

[0311] That is, the second controller (2002) can set the first plane to include the first object region and the reference region according to various criteria, and can set the second plane to include the second object region and the reference region. At this time, the first plane and the second plane can form a predetermined angle. In this way, the image analysis device (2000) can derive the relationship between the reference region and the object region even for a plane that forms a predetermined angle with the plane including the medical image that is the basis of the 3D medical model, and can calculate a disease index or provide auxiliary diagnostic information.

[0312] The above describes examples of methods for obtaining a disease index from one or more medical images and providing diagnostic auxiliary information based on the obtained disease index, which are executed by the diagnostic auxiliary information providing system (10000) according to various embodiments.

[0313] As described above, according to the method for providing diagnostic auxiliary information executed by the diagnostic auxiliary information providing system (10000) of this embodiment, by calculating a disease index based on an automated algorithm or program having clear and uniform criteria from medical images, subjective human judgment can be eliminated and a more objective disease index can be calculated, thereby providing more accurate diagnostic auxiliary information.

[0314] Hereinafter, a method for the diagnostic auxiliary information providing system (10000) according to the embodiment to perform segmentation on a specific region contained in a medical image from a medical image will be described.

[0315] According to this embodiment, the diagnostic auxiliary information providing system (10000) can segment a medical image. Specifically, the image analyzing device (2000) can acquire a medical image from the image acquiring device (1000) and segment the acquired medical image. More specifically, the second controller (2002) can segment the medical image using an algorithm for image segmentation stored in the second memory (2400).

[0316] The following description will be given with reference to the drawings.

[0317] FIG. 33 generally illustrates the medical image segmentation operation performed by the video analyzer according to the present embodiment.

[0318] According to this embodiment, the image analysis device 2000 can receive input data and output output data. Specifically, the second controller 2002 can input the medical image acquired from the image acquisition device 1000 as input data to an algorithm for image segmentation and obtain output data in which the medical image is segmented.

[0319] The image analysis device 2000 according to the embodiment can use various algorithms for image segmentation.

[0320] As an example, an algorithm for image segmentation can be provided in a machine learning model. A representative example of a machine learning model is an artificial neural network. Specifically, a representative example of an artificial neural network is a deep learning series artificial neural network including an input layer that accepts data input, an output layer that outputs the results, and a hidden layer that processes data between the input layer and the output layer.

[0321] Specific examples of artificial neural networks include convolutional neural networks, recurrent neural networks, and deep neural networks. In this specification, the term "artificial neural network" should be interpreted in a comprehensive sense to include the above-mentioned artificial neural networks, various other forms of artificial neural networks, and combinations of these forms of artificial neural networks, and must be a deep running series.

[0322] Furthermore, the machine learning model does not necessarily have to be in the form of an artificial neural network model, but may also include other techniques such as nearest neighbor algorithm (KNN), random forest, support vector machine (SVM), principal component analysis (PCA), etc., and may also include ensembles of the above-mentioned techniques or combinations thereof in various ways. Meanwhile, in the embodiments described with a focus on an artificial neural network, it is clear that the artificial neural network may be replaced with a different machine learning model unless otherwise specified.

[0323] Moreover, in this specification, the algorithm for image segmentation is not necessarily limited to a machine learning model, that is, the algorithm for image segmentation may include various judgment / decision algorithms other than the machine learning model.

[0324] Therefore, it should be understood that in this specification, an algorithm for image segmentation has a comprehensive meaning that includes all types of algorithms that analyze a medical image and distinguish regions contained within the medical image.

[0325] An embodiment of the artificial neural network model will now be described with reference to FIGS.

[0326] FIG. 34 illustrates an example implementation of an artificial neural network model according to an embodiment.

[0327] Referring to FIG. 34, an image analysis device 2000 according to an embodiment of the present invention can utilize U-net, an artificial neural network for image segmentation.

[0328] U-net, which is often used for image segmentation, can be constructed with an architecture that includes a contraction path and an expansion path.

[0329] Specifically, the U-net contraction path can be configured to perform two convolutions and max pooling consecutively, and image-related features can be extracted in the U-net contraction path.

[0330] However, since the contraction path also reduces the size of the feature map, the U-net can be configured to additionally include an expansion path to restore the size of the feature map.

[0331] The extension path of the U-net can be configured to perform up-convolution and two-degree convolution consecutively, at which time the magnitude of the image and feature map can be extracted in the extension path of the U-net.

[0332] Additionally, the U-net can be architectured to concatenate same-level feature maps, providing location information about the feature to the contraction path and the expansion path.

[0333] At this time, based on the label difference between the input image label and the output segmentation map, the parameters or weights of at least one node in the layer including the U-net are adjusted so that the label difference between the input image label and the output segmentation map is minimized.

[0334] FIG. 35 illustrates another implementation of an artificial neural network model according to an embodiment.

[0335] 35, the image analysis apparatus (2000) according to the embodiment can utilize U-net++ in an artificial neural network for image segmentation. U-net++ is an artificial neural network model using the dense block idea of ​​DenseNet to improve the performance of U-net, and is different from U-net in that a convolution layer exists in the skip path to bridge the semantic gap between the feature maps of the encoder and the decoder, and a tense skip connection exists in the skip path to improve the gradient flow.

[0336] Specifically, the image analysis device 2000 can be embodied to input an input image to an input layer of a U-net++ neural network model and obtain label information output through an output layer. In this case, the image analysis device 2000 can adjust the parameters or weights of at least one node in a hidden layer included in the U-net++ based on the difference between the label information included in the input image and the label information output from the neural network model.

[0337] More specifically, the second controller (2002) is embodied to repeatedly perform the operation of adjusting the parameters or weights of at least one of the above-mentioned nodes, and can obtain the parameters or weights of the nodes that minimize the difference between the label information included in the input image and the label information output from the neural network model.

[0338] FIG. 36 illustrates a segmentation result using an artificial neural network model according to an embodiment.

[0339] According to an embodiment, the image analysis device (2000) can segment the medical image based on characteristics of the medical image, and in particular, the second controller (2002) can segment the medical image to correspond to the conditions under which the medical image was acquired.

[0340] Here, the condition under which the medical image is acquired may vary depending on the setting parameters of the image acquisition device 1000. As an example, the condition under which the medical image is acquired is a preset magnetic state parameter value of the image acquisition device 1000. Specifically, when the image acquisition device 1000 is embodied as an MRI device, a T1-weighted image or a T2-weighted image may be acquired depending on the setting of the TR / TE value, and the second controller 2002 may segment the image to correspond to the acquired image. Here, the characteristics according to the acquisition conditions of the medical image are as described above.

[0341] A specific example will be described with reference to the drawings.

[0342] Referring to FIG. 36, segmentation results of a T1-weighted image and a T2-FLAIR image acquired from the same subject are shown.

[0343] According to an embodiment, the second controller (2002) can segment the T1-weighted image. Here, the T1-weighted image can be segmented to determine anatomical characteristics. As shown in the drawing, the segmentation result of the T1-weighted image shows a clearer segmentation of wrinkles in the gray matter outer shell region than the segmentation result of the T2-FLAIR image.

[0344] Additionally, according to an embodiment, the second controller (2002) can segment the T2-FLAIR image, where the T2-FLAIR image can be segmented to determine pathological characteristics. As shown in the drawings, the segmentation results of the T2-FLAIR image are shown to contain more information about the WMH region than the segmentation results of the T1-weighted image.

[0345] In this way, even if medical images are acquired from the same photographed subject, the characteristics (e.g., anatomical characteristics, pathological characteristics) contained in the medical image may vary depending on the acquisition conditions of the image acquisition device (1000), and the information observable in the medical image may also vary depending on the characteristics of the medical image.

[0346] However, if desired, more accurate diagnostic aids can be provided if various medical image characteristics are taken into account in a comprehensive manner. Therefore, the following describes a process for segmenting a medical image to include multiple characteristics.

[0347] FIG. 37 illustrates an example of a medical image that has been segmented to include multiple features according to an embodiment.

[0348] 37, the image analysis device 2000 according to the embodiment can segment a medical image so that at least two or more characteristics that change depending on the acquisition condition are included in one medical image. Specifically, the second controller 2002 can segment the first image based on the first image so that the characteristics of the second image are included.

[0349] A specific example will be described with reference to the drawings.

[0350] A second controller (2002) can input a first medical image having a first characteristic as input data to the artificial neural network to obtain output data labeled with the first characteristic and a second characteristic related to the second medical image, where the artificial neural network can learn based on a learning set related to the second image having the second characteristic, for example, where the first characteristic is an anatomical or structural characteristic such that the first image is a T1-weighted image, and the second characteristic is a pathological characteristic such that the second image is a T2-FLAIR image.

[0351] As a more specific example, in conjunction with (a), the second controller (2002) can input a first image (in the illustrated example, a T2-FLAIR image) containing pathological characteristics as input data to an artificial neural network according to an embodiment.

[0352] Here, if an artificial neural network trained only on a learning set relating to the first image is used, the second controller (2002) will output only segmentation results including only the first characteristics (i.e., pathological characteristics) along with (b).

[0353] However, the artificial neural network according to an embodiment of the present invention can learn using a learning set based on both the first image related to the first characteristic and the second image related to the second characteristic, whereby the second controller (2002) can input the first image to the artificial neural network and obtain a segmentation result for the first image that reflects both the first characteristic and the second characteristic, together with (c).

[0354] In this manner, even when a medical image containing one characteristic is segmented, information on multiple characteristics can be obtained together, making it possible to obtain various disease-related information from a single medical image, and the image analysis device (2000) according to the embodiment can provide more accurate and various auxiliary diagnostic information from the medical image.

[0355] Hereinafter, a learning process of an artificial neural network according to an embodiment will be described with reference to the drawings.

[0356] FIG. 38 is a flowchart illustrating the learning process of an artificial neural network according to an embodiment.

[0357] The artificial neural network according to the embodiment may be trained using training data. Here, the artificial neural network according to the embodiment may be trained through various devices capable of driving an artificial neural network algorithm. As an example, the artificial neural network may be trained through a video analysis device (2000). For convenience of explanation, the present specification has described the artificial neural network being trained through the video analysis device (2000), but it should be understood that the present specification is not limited thereto and the artificial neural network may be trained through other devices for driving an artificial neural network.

[0358] Referring to FIG. 39, the learning process of the artificial neural network according to the embodiment includes a step of segmenting a first image including a first characteristic (S6000), a step of aligning segmentation information of the first image based on a second image including a second characteristic (S6200), a step of performing primary learning using the aligned image and the second image (S6400), a step of morphologically correcting the second image including the output result of the primary learning (S6600), and a step of performing secondary learning using the second image including the morphologically corrected segmentation result (S6800).

[0359] First, the image analysis device (2000) can segment a first image including a first characteristic (S6000). Specifically, the second controller (2002) can segment the first image to include information about the first characteristic acquired under a first acquisition condition. In other words, the image acquisition device (1000) can segment the first image including the first characteristic acquired when set to a first parameter.

[0360] After segmenting the first image, the image analysis device (2000) can match the segmented first image and the second image based on the second image (S6200). Specifically, the second controller (2002) can match the first image segmented to include the first characteristic on the second image including the second characteristic to obtain a second image including segmented information regarding the first characteristic. In other words, the first image segmented to include the first characteristic can be matched on the second image including the second characteristic obtained when set as the second parameter of the image acquisition device (1000). This will be described in detail later.

[0361] After the first image and the second image are matched, the image analysis device (2000) can perform primary learning using the matched second image and the unmatched second image (S6400). Specifically, the second controller (2002) can train the artificial neural network using the second image that is matched to reflect the first characteristic with the second image, so that the artificial neural network according to the embodiment outputs a segmentation result including the first characteristic from the second image.

[0362] That is, the artificial neural network can learn using a learning set including a second image labeled with a first characteristic and an original second image. Thus, the artificial neural network can receive the second image as input data and output the second image labeled with the first characteristic. Here, although the learning set has been described as including the original second image, a processed second image other than the original second image can be used. As an example, the second image is a second image segmented to include the second characteristic. Alternatively, the second image may be a second image on which other preprocessing processes have been performed. However, for convenience of explanation, the following description of this specification will mainly describe the artificial neural network learning using a running set based on the original second image.

[0363] After the artificial neural network has been primarily trained, the image analysis device (2000) can perform morphological correction on the output image of the primarily trained artificial neural network (S6600). Specifically, the second controller (2002) can input the original second image data to the artificial neural network and perform morphological correction on the second image segmented to include the second characteristic output, in order to improve the accuracy of the segmentation result.

[0364] The primarily trained artificial neural network may not accurately include information about the first characteristic. Therefore, the image analysis apparatus (2000) according to the embodiment may improve the computational accuracy of the artificial neural network by morphologically correcting the primary output data of the artificial neural network based on the first image including the first characteristic, so that the artificial neural network can more accurately distinguish information about the first characteristic on the second image.

[0365] When the morphological correction of the primary output data of the artificial neural network is completed, the image analysis device (2000) can secondarily train the artificial neural network based on the morphologically corrected second image (S6800). Specifically, the second controller (2002) can train the artificial neural network according to the embodiment based on a learning set including the second image that has been morphologically corrected and labeled with the first characteristic. That is, the artificial neural network according to the embodiment can perform segmentation so that information related to the first characteristic is more accurately reflected from the second image by learning using a learning set including the second image that has been labeled to reflect the morphologically corrected first characteristic.

[0366] Hereinafter, an example of a learning process of an artificial neural network according to an embodiment will be described with reference to the drawings.

[0367] FIG. 39 illustrates an example of a matching process between a first image and a second image according to an embodiment.

[0368] Referring to Fig. 39, the image analysis device (2000) according to the embodiment can match a first image and a second image. Specifically, the second controller (2002) can obtain a matching matrix between the first image and the second image. Here, the matching matrix means a conversion function between the first image and the second image. That is, the matching matrix according to the embodiment can be expressed as a function that indicates a correlation between a first point included in the first image and a second point included in the second image between the first image and the second image of the same photographed object. Here, the first point and the second point may mean a pixel or a set of pixels.

[0369] The image analysis device (2000) according to the embodiment can process the second image to correspond to the segmentation result of the first image based on the first image segmented for the first characteristic using a matching matrix. That is, the second controller (2002) can segment the second image such that the segmentation result of the first image with respect to the first characteristic is reflected in the second image. Here, as described above, the second image can be pre-segmented to include the second characteristic. That is, the second controller (2002) can segment the second image such that both the first characteristic and the second characteristic are reflected.

[0370] As a more specific example, the first image may be a T1-weighted image and the second image may be a T2-FLAIR image, that is, the image analysis device 2000 may first match the T1-weighted image and the T2-FLAIR image to obtain a matching matrix.

[0371] Also, the image analysis device (2000) can segment the T1-weighted image so that the anatomical and structural characteristics of the T1-weighted image are reflected. For example, the second controller (2002) can process the T1-weighted image so that a specific region of the brain can be distinguished. More specifically, the second controller (2002) can segment the brain so that organs contained in the brain, such as the cerebrum, cerebellum, diencephalon, and hippocampus, can segment the T1-weighted image so that organs contained in the brain, such as a seahorse, can segment the brain so that each part and position of the brain, such as the temporal lobe, frontal lobe, and occipital lobe, can be distinguished, or a combination of these can be segmented.

[0372] The image analyzer (2000) can then use the matching matrix to match the segmented T1-weighted image to the T2-FLAIR image so that the segmentation results of the T1-weighted image are reflected in the T2-FLAIR image.

[0373] Thus, the image analysis device (2000) can obtain a T2-FLAIR image segmented to reflect anatomical characteristics. Specifically, the second controller (2002) can segment the T2-FLAIR image so that the above-mentioned anatomical characteristics are reflected in a T2-FLAIR image in which WMHs observable in a medical image or white matter and gray matter, which are materials constituting the brain, are distinguished. The T2-FLAIR image segmented to reflect the above-mentioned anatomical characteristics can be used for the primary learning of the artificial neural network as described above.

[0374] FIG. 40 shows an example of morphological modification according to an embodiment.

[0375] 40, the image analysis apparatus 2000 according to the embodiment can morphologically modify a segmentation result for a medical image. Specifically, the second controller 2002 can morphologically modify the segmentation result that reflects a first characteristic included in the second image. The second image including the morphologically modified segmentation result can more accurately reflect the first characteristic.

[0376] A specific example will be described with reference to the drawings.

[0377] As described above, the artificial neural network can perform primary learning using the T2-FLAIR image segmented to reflect anatomical characteristics and the original T2-FLAIR image. As a result of the primary learning, the artificial neural network can receive the T2-FLAIR image as input and output a T2-FLAIR image segmented to reflect the characteristics (anatomical or structural characteristics) of the T1-weighted image. The T2-FLAIR image, which is the output result of the artificial neural network, is segmented to reflect the characteristics of the T1-weighted image, but is incomplete in some areas and therefore requires correction.

[0378] In case (a), the first learning result of the artificial neural network according to the embodiment is shown in which the contour of the segmented area is segmented to incompletely reflect the anatomical characteristics, whereas in case (b), the first output of the morphologically corrected artificial neural network is shown in which the contour of the segmented area is corrected to better reflect the anatomical characteristics.

[0379] More specifically, in (a), some gray matter regions are not differentiated, the boundaries of each region are partially cut off, and there is also noise due to incorrect labeling of regions due to incomplete learning.

[0380] However, the morphologically corrected medical images show that the grey matter regions are well differentiated and that the boundaries of each region are clearly defined to distinguish the specific region.

[0381] Here, according to the present embodiment, the second controller (2002) can perform morphological correction based on the characteristics of the brain structure. In general, the brain regions including the ventricles must be connected in 26 different ways in three dimensions. Therefore, the second controller (2002) can reduce noise contained in the segmented medical image based on the connection relationship between each brain region.

[0382] Also, here, the ventricles are erroneously distinguished from other non-tissues and erroneously labeled as non-tissues. In this case, to compensate for the erroneous labeling of the ventricular regions, the second controller (2002) can label the ventricular regions using a fill-hole method.

[0383] In this manner, an artificial neural network according to an embodiment can be retrained by inputting modified morphological segmentation results and outputting results that more accurately reflect the characteristics one wishes to include.

[0384] FIG. 41 is a flowchart illustrating a deploying process using an artificial neural network of a video analysis device according to an embodiment.

[0385] Referring to FIG. 41, a deployment process using an artificial neural network of an image analysis device (2000) according to an embodiment includes a step of inputting a first image including a first characteristic (S7000), and a step of outputting the first image processed to include a segmentation result including the first characteristic and a second characteristic (S7200).

[0386] First, the image analysis device (2000) can input a first image including a first characteristic to the artificial neural network (S7000). Specifically, the second controller (2002) can input the first image including the first characteristic to the artificial neural network that has been fully trained and stored in the second memory (2400). Here, the artificial neural network has been fully trained through the above-mentioned training process.

[0387] When a first image is input, the image analysis device (2000) can obtain, as an output result of the artificial neural network, a first image that has been processed to include a first characteristic associated with the first image and a second characteristic associated with a second image of a different format from the first image (S7200). Specifically, the second controller (2002) can obtain a first image that has been segmented so that the first characteristic associated with the first image and the second characteristic associated with the second image of a different format from the first image are comprehensively distinguished.

[0388] In this way, by using an artificial neural network that has been fully trained according to the learning process of the artificial neural network according to the embodiment, the image analysis device (2000) can obtain a medical image that is segmented to include both the first characteristic and a second characteristic different from the first characteristic even if a medical image including only the first characteristic is input to the artificial neural network. As a result, various information for analyzing a disease for which related information is to be obtained can be obtained, and various information related to the disease can be provided to the user by analyzing the various information.

[0389] FIG. 42 shows an example of the final output result of an artificial neural network according to an embodiment.

[0390] 42, the image analysis device 2000 can input a first image to an artificial neural network to obtain a first image that has been processed to include various characteristics. Specifically, the second controller 2002 can input the first image reflecting the first characteristic to the artificial neural network to obtain a first image that has been segmented to more fully reflect characteristics different from the first characteristic.

[0391] To explain a specific example with reference to the drawings, the image analysis device (2000) can input a T2-FLAIR image into an artificial neural network to obtain an image processed to reflect the characteristics of a T1-weighted image.

[0392] Specifically, the second controller (2002) can input a T2-FLAIR image to a fully trained artificial neural network based on a learning set reflecting the characteristics of a T1-weighted image and the characteristics of a T2-FLAIR image, and obtain a T2-FLAIR image segmented to reflect the characteristics of the T1-weighted image. Here, it is natural that the artificial neural network is trained to perform segmentation by reflecting the existing characteristics of the T2-FLAIR image. That is, the artificial neural network according to the present embodiment can input a T2-FLAIR image as input data and obtain an image segmented to include not only anatomical or structural characteristics that are easily observable in a T1-weighted image, but also all pathological characteristics that are easily observable in a T2-FLAIR image.

[0393] To describe a more specific example, the final output of the artificial neural network according to the embodiment may be segmented to include all of the characteristics of the T1-weighted image, along with the organs and locations of brain regions contained therein, based on a T2-FLAIR image in which pathological characteristics such as WMHs and brain materials such as white matter and gray matter are clearly distinguished.

[0394] The method according to the embodiment may be embodied in the form of program instructions executable by various computer means and recorded on a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be those specifically designed and constructed for the embodiments, or those publicly available to those skilled in the art of 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, magneto-optical media such as floptical disks, and hardware devices specifically configured to store and execute program instructions, such as ROMs, RAMs, flash memories, 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 created by a compiler. The above-mentioned hardware devices may be configured to operate as one or more software modules to perform the operations of the embodiments, and may function in the same manner.

[0395] Although the embodiments of the present invention have been described above by way of limited examples and drawings, those skilled in the art may make various modifications and variations based on the above description. For example, the described techniques may be performed in an order different from that described, and / or the components of the described systems, structures, devices, circuits, etc. may be combined or combined in a manner different from that described, or may be replaced or substituted by other components or equivalents, and still achieve suitable results.

[0396] Accordingly, other implementations, other embodiments, and equivalents of the claims are within the scope of the following claims.

Claims

1. A method for providing auxiliary diagnostic information based on a plurality of images, the method being performed by an apparatus for providing auxiliary diagnostic information, comprising: obtaining a first slice image and a second slice image, wherein the first and second slice images relate to a brain, each including a plurality of cells; labeling the cells included in the first and second slice images with feature values ​​representing one of white matter, ventricles, and white matter hyperintensity (WMH); providing diagnostic auxiliary information including a disease index calculated based on a distance between a first cell included in the first slice image and a second cell included in the second slice image; The first cell is selected as one of a plurality of cells labeled with a feature value representing a ventricle, and the second cell is selected as one of a plurality of cells labeled with a feature value representing a white matter hyperintensity signal.

13. A method for providing auxiliary diagnostic information based on a plurality of images, comprising:

2. The distance between the first cell and the second cell is determined taking into consideration the distance between the first slice image and the second slice image.

2. A method for providing auxiliary diagnostic information based on a plurality of images as claimed in claim 1.

3. The first cell is located in the center of the area formed by the labeled cells in the ventricle.

3. A method for providing auxiliary diagnostic information based on a plurality of images as claimed in claim 2.

4. The first cell is the cell that is located at the shortest distance from the white matter hyperintensity signal among the cells labeled in the ventricle.

3. A method for providing auxiliary diagnostic information based on a plurality of images as claimed in claim 2.

5. The first slice image and the second slice image are two images selected from a plurality of end surface images taken in a continuous order for one object to be photographed.

2. A method for providing auxiliary diagnostic information based on a plurality of images as claimed in claim 1.

6. The first slice image and the second slice image are two adjacent images among the plurality of continuous end surface images.

6. A method for providing auxiliary diagnostic information based on a plurality of images as claimed in claim 5.

7. 1. An apparatus for providing diagnostic auxiliary information, comprising: a communication module for acquiring a plurality of slice images associated with the brain, the slice images including a first slice image and a second slice image including a plurality of cells; a controller for analyzing the plurality of slice images to obtain auxiliary diagnostic information; the controller labels the cells included in the first and second slice images with feature values ​​representing any one of white matter, ventricles, and white matter hyperintensity (WMH), and provides auxiliary diagnostic information including a disease index calculated based on a distance between a first cell included in the first slice image and a second cell included in the second slice image; The first cell is selected as one of a plurality of cells labeled with a feature value representing a ventricle, and the second cell is selected as one of a plurality of cells labeled with a feature value representing a white matter hyperintensity signal. A diagnostic auxiliary information providing device comprising:

8. The distance between the first and second cells is: is determined taking into consideration a pre-acquired distance between the first slice image and the second slice image.

8. The diagnostic auxiliary information providing device according to claim 7.

9. The first cell is located in the center of the area formed by the labeled cells in the ventricle.

8. The diagnostic auxiliary information providing device according to claim 7.

10. The first cell is the cell that is located at the shortest distance from the white matter hyperintensity signal among the cells labeled in the ventricle.

8. The diagnostic auxiliary information providing device according to claim 7.

11. The first slice image and the second slice image are two images selected from a plurality of end surface images taken in a continuous order for one object to be photographed.

8. The diagnostic auxiliary information providing device according to claim 7.

12. The first slice image and the second slice image are two adjacent images among the plurality of continuous end surface images.

12. The diagnostic auxiliary information providing device according to claim 11.

13. A computer-readable recording medium having stored thereon a program for executing the method according to any one of claims 1 to 6.

Images