Methods for predicting the risk of brain disease and methods for training brain disease risk analysis models.

Abstract

A method for predicting the risk of brain diseases according to an embodiment of the present disclosure includes: obtaining a brain disease risk analysis model that has completed learning; obtaining target shape information of the cerebrovascular vessels of a target patient; obtaining target blood flow information of the cerebrovascular vessels of the target patient; obtaining target patient information of the target patient; and inputting the target shape information, the target blood flow information, and the target patient information into the brain disease risk analysis model, and obtaining a brain disease risk value output through the brain disease risk analysis model.

Description

Technical Field

[0001] The present disclosure relates to a method for predicting the risk of brain diseases and a method for training a brain disease risk analysis model. Specifically, the present disclosure relates to a method for predicting the risk of brain diseases based on the shape information of cerebral blood vessels, blood flow information, and / or patient information, and a method for training a brain disease risk analysis model for predicting the risk of brain diseases.

Background Art

[0002] Various imaging techniques such as positron emission tomography (PET), functional magnetic resonance imaging (fMRI), magnetic resonance angiography (MRA), and computed tomography have been developed to visualize the three-dimensional structure of the brain and cerebral blood vessels. The development of such imaging techniques has enabled the measurement of cerebral blood flow throughout the brain, and due to the ability to visualize cerebral blood vessels and measure cerebral blood flow, research for predicting or diagnosing the risk of brain diseases based on the shape information of cerebral blood vessels and cerebral blood flow information has attracted attention.

[0003] Conventionally, brain diseases related to cerebral blood vessels have been diagnosed or predicted depending on medical databases for a patient's gender, disease, underlying disease, and race, etc., and the patient's symptoms. However, the conventional method only predicts the possibility of brain diseases based on the patient's symptoms and the patient's statistical data, and there is a limitation in that it cannot diagnose or predict brain diseases by comprehensively considering the shape information of cerebral blood vessels and cerebral blood flow information.

[0004] Therefore, there is a real need for the development of new technologies for predicting or analyzing the risk of brain diseases by comprehensively considering the shape information of cerebral blood vessels, cerebral blood flow information, and / or the patient's health information.

Summary of the Invention

Problems to be Solved by the Invention

[0005] One problem that this disclosure aims to solve is to provide a method for calculating the risk of brain disease based on cerebrovascular shape information, cerebral blood flow information, and / or patient information, and a method for training a brain disease risk analysis model.

[0006] The problems that this disclosure seeks to solve are not limited to those described above, and any problems not mentioned can be clearly understood by a person with ordinary skill in the art to which this disclosure pertains from this specification and the accompanying drawings. [Means for solving the problem]

[0007] This summary is provided to introduce the selection of concepts into simplified forms, which will be further explained in the detailed description below. This summary is not intended to identify the main or essential features of the requested subject, nor to help determine the scope of the requested subject.

[0008] A method for predicting the risk of brain disease according to one embodiment of the present disclosure may include the steps of: acquiring a brain disease risk analysis model that has been trained; acquiring target shape information of the cerebral blood vessels of a target patient; acquiring target blood flow information of the cerebral blood vessels of a target patient; acquiring target patient information of a target patient; and inputting the target shape information, the target blood flow information, and the target patient information into the brain disease risk analysis model and acquiring a brain disease risk value output through the brain disease risk analysis model.

[0009] The technical solutions provided in this disclosure are not limited to those described above, and any technical solutions not mentioned can be clearly understood by a person with ordinary skill in the art to which this disclosure pertains from this specification and the accompanying drawings. [Effects of the Invention]

[0010] According to a learning method for a brain disease risk analysis model as described in one embodiment of this disclosure, a brain disease risk analysis model can be provided that can automatically calculate or predict the brain disease risk based on cerebral blood vessel shape information data, cerebral blood flow information, and / or patient health information.

[0011] According to a brain disease risk prediction method according to one embodiment of the present disclosure, a brain disease risk personalized for a target patient can be calculated by considering the shape information of the target patient's cerebral blood vessels, blood flow information related to the velocity and pressure of the target patient's blood flow, and patient information related to the target patient's age, sex, underlying disease, race, etc.

[0012] The effects of this disclosure are not limited to those described above, and any effects not mentioned can be clearly understood by a person with ordinary skill in the art to which this disclosure pertains from this specification and the accompanying drawings. [Brief explanation of the drawing]

[0013] [Figure 1] This is a schematic diagram relating to a brain disease risk analysis device according to one embodiment of the present disclosure. [Figure 2] This is a schematic diagram illustrating the operation of a brain disease risk analysis device according to one embodiment of the present disclosure. [Figure 3] This is a diagram illustrating the operation of a brain disease risk analysis device according to one embodiment of the present disclosure. [Figure 4] This is a schematic diagram illustrating the operation of a brain disease risk analysis device for training a brain disease risk analysis model according to one embodiment of the present disclosure. [Figure 5] This diagram illustrates the process of training a brain disease risk analysis model according to one embodiment of the present disclosure. [Figure 6] This is a flowchart illustrating the learning method of a brain disease risk analysis model according to one embodiment of this disclosure. [Figure 7] These are diagrams illustrating the process of training a brain disease risk analysis model according to one embodiment of this disclosure, and the process of verifying the brain disease risk analysis model. [Figure 8]The drawings for explaining an aspect of training a brain disease risk analysis model according to an embodiment of the present disclosure and an aspect of predicting the brain disease risk using the learned brain disease risk analysis model. [Figure 9] The drawings for explaining an aspect of training a brain disease risk analysis model according to another embodiment of the present disclosure and an aspect of predicting the brain disease risk using the learned brain disease risk analysis model.

Mode for Carrying Out the Invention

[0014] The above-described technical problems, features, and advantages of the present disclosure will become more apparent through the following detailed description related to the accompanying drawings. However, the present disclosure can be modified in various ways and can have various embodiments. Hereinafter, specific embodiments will be illustrated in the drawings and will be described in detail.

[0015] Throughout the specification, the same reference numerals generally indicate the same components. Also, components having the same functions within the scope of the same concept shown in the drawings of each embodiment will be described using the same reference numerals, and duplicate descriptions thereof will be omitted.

[0016] When it is determined that a specific description of a known function or configuration related to the present disclosure may unnecessarily obscure the gist of the present disclosure, the detailed description thereof will be omitted. Also, the numbers (for example, first, second, etc.) used in the description process of this specification are merely identification symbols for distinguishing one component from another component.

[0017] Also, the suffixes “module” and “unit” for the components used in the following embodiments are given or mixed only in consideration of the ease of preparing the specification, and do not have meanings or roles that are distinguishable from each other by themselves.

[0018] In the following embodiments, the singular expressions include plural expressions unless the context clearly indicates a different meaning.

[0019] In the following embodiments, terms such as "comprising" or "having" mean that the features or components described in the specification exist, and do not preclude in advance the possibility of adding one or more other features or components.

[0020] In the drawings, for the sake of convenience of explanation, the sizes of components may be exaggerated or reduced. For example, the sizes and thicknesses of each configuration shown in the drawings are arbitrarily shown for the convenience of explanation, and the present disclosure is not necessarily limited to what is shown.

[0021] If a certain embodiment can be embodied differently, the order of a specific process may be performed differently from the order described. For example, two processes described continuously may be performed substantially simultaneously, or may proceed in the order opposite to the described order.

[0022] In the following embodiments, when components or the like are connected, it includes not only the case where the components are directly connected, but also the case where components are interposed between the components and indirectly connected.

[0023] For example, when components or the like are electrically connected in this specification, it includes not only the case where the components or the like are directly electrically connected, but also the case where components or the like are interposed therebetween and indirectly electrically connected.

[0024] A method for predicting the risk of brain diseases according to an embodiment of the present disclosure may include: obtaining a brain disease risk analysis model for which learning has been completed; obtaining target shape information of the cerebral blood vessels of a target patient; obtaining target blood flow information of the cerebral blood vessels of the target patient; obtaining target patient information of the target patient; and inputting the target shape information, the target blood flow information, and the target patient information into the brain disease risk analysis model, and obtaining a brain disease risk value output through the brain disease risk analysis model.

[0025] According to one embodiment of the present disclosure, the step of acquiring the target shape information further includes the step of acquiring the shape information of a standard blood vessel; and the step of acquiring the difference between the target shape information and the shape information of the standard blood vessel; and the step of acquiring the brain disease risk value may further include the step of inputting the difference between the target shape information and the shape information of the standard blood vessel into the brain disease risk analysis model and acquiring the brain disease risk value output through the brain disease risk analysis model. According to one embodiment of the present application, the step of acquiring the target blood flow information further includes the step of acquiring the blood flow information of a standard blood vessel; and the step of acquiring the difference between the target blood flow information and the blood flow information of the standard blood vessel; and the step of acquiring the brain disease risk value may further include the step of inputting the difference between the target blood flow information and the blood flow information of the standard blood vessel into the brain disease risk analysis model and acquiring the brain disease risk value output through the brain disease risk analysis model.

[0026] According to one embodiment of the present disclosure, the brain disease risk analysis model may be trained on a first training dataset related to cerebrovascular shape information, including location information or diameter information of cerebrovascular vessels; a second training dataset related to cerebrovascular blood flow information, including blood flow velocity information or blood flow pressure information of cerebrovascular vessels; a third training dataset related to patient information, including patient age or gender; and a fourth training dataset related to brain disease information.

[0027] According to one embodiment of the present disclosure, the brain disease risk analysis model is configured to acquire the first training dataset, the second training dataset, the third training dataset, and the fourth training dataset, and to output a brain disease risk prediction value, and the brain disease risk analysis model can be trained to output the brain disease risk prediction value to approximate the brain disease information contained in the fourth training dataset.

[0028] According to one embodiment of the present disclosure, the brain disease risk analysis model can be trained on a first training dataset consisting of the difference between the shape information of a patient's cerebral blood vessels and the shape information of standard cerebral blood vessels; a second training dataset consisting of the difference between the blood flow information of a patient's cerebral blood vessels and the blood flow information of standard cerebral blood vessels; a third training dataset related to patient information, including the patient's age or gender; and a fourth training dataset related to brain disease information.

[0029] According to one embodiment of the present disclosure, the brain disease risk analysis model is configured to acquire the first training dataset, the second training dataset, the third training dataset, and the fourth training dataset, and to output a brain disease risk prediction value, and the brain disease risk analysis model can be trained to output the brain disease risk prediction value to approximate the brain disease information contained in the fourth training dataset.

[0030] According to one embodiment of the present disclosure, a computer-readable recording medium may be provided which contains a program for performing the method for predicting the risk of brain disease.

[0031] The following describes, with reference to Figures 1 to 9, a brain disease risk prediction method, a brain disease risk analysis model training method, and a brain disease risk analysis device (or brain disease risk analysis server, hereinafter referred to as the "brain disease risk analysis device") according to the embodiments of this disclosure.

[0032] Figure 1 is a schematic diagram relating to a brain disease risk analysis device according to one embodiment of the present disclosure.

[0033] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can train a brain disease risk analysis model. Furthermore, the brain disease risk analysis device 1000 can use the trained brain disease risk analysis model to calculate or predict the brain disease risk from medical images.

[0034] A brain disease risk analysis device 1000 according to one embodiment of the present disclosure may include a transmitting / receiving unit 1100, a memory 1200, and a processor 1300.

[0035] The transmitting / receiving unit 1100 can communicate with any external device, including medical imaging devices (e.g., MRA devices, MRI devices, CT devices, PET devices, etc.). For example, the brain disease risk analysis device 1000 can acquire medical images taken from a medical imaging device through the transmitting / receiving unit 1100. The brain disease risk analysis device 1000 can also acquire any execution data through the transmitting / receiving unit 1100 to run a brain disease risk analysis model that has been trained. Here, execution data may include any appropriate data for running the brain disease risk analysis model, including structural information, hierarchical information, computational libraries, and parameter sets related to weighted values ​​included in the brain disease risk analysis model. Furthermore, the brain disease risk analysis device 1000 can transmit or output the brain disease risk analysis results acquired through the brain disease risk analysis model to any external device, including a user terminal, through the transmitting / receiving unit 1100.

[0036] The brain disease risk analysis device 1000 can connect to a network via the transmitting / receiving unit 1100 to send and receive various data. The transmitting / receiving unit 1100 can broadly include wired and wireless types. Since wired and wireless types each have their advantages and disadvantages, the brain disease risk analysis device 1000 may be equipped with both wired and wireless types simultaneously, depending on the circumstances. In the case of the wireless type, communication methods such as Wi-Fi and other WLAN (Wireless Local Area Network) series can be mainly used. Alternatively, in the case of the wireless type, cellular communication, such as LTE or 5G series communication methods, can be used. However, the wireless communication protocol is not limited to the examples given above, and any appropriate wireless communication method can be used. In the case of the wired type, LAN (Local Area Network) and USB (Universal Serial Bus) communication are typical examples, but other methods are also possible.

[0037] Memory 1200 can store various types of information. Various types of data can be stored in Memory 1200 temporarily or semi-permanently. Examples of Memory 1200 include hard disk drives (HDD), solid state drives (SSD), flash memory, read-only memory (ROM), and random access memory (RAM). Memory 1200 can be provided either built into the brain disease risk analysis device 1000 or as a removable component. Memory 1200 can store various types of data necessary for the operation of the brain disease risk analysis device 1000, including operating systems (OS) for driving the device and programs for operating each component of the device.

[0038] The processor 1300 can control the overall operation of the brain disease risk analysis device 1000. For example, the processor 1300 can control the overall operation of the brain disease risk analysis device 1000, including operations such as acquiring pixel information of the original medical image (described later), acquiring coordinate information from the pixel information, generating shape information of cerebral blood vessels, acquiring fluid flow prediction values ​​(or blood flow information), training a brain disease risk analysis model, and / or predicting brain disease risk. Specifically, the processor 1300 can load and execute a program for the overall operation of the brain disease risk analysis device 1000 from the memory 1200. The processor 1300 can be embodied in hardware, software, or a combination thereof as an AP (Application Processor), CPU (Central Processing Unit), or similar device. In this case, hardware-wise it may be provided in the form of an electronic circuit that processes electrical signals to perform control functions, and software-wise it may be provided in the form of a program or code that drives the hardware circuit.

[0039] The following describes in detail the various operations of the brain disease risk analysis device 1000 according to one embodiment of this disclosure, with reference to Figures 2 and 3. Figure 2 is a schematic diagram illustrating the operation of the brain disease risk analysis device 1000 according to one embodiment of this disclosure. Figure 3 is a diagram illustrating the operation of the brain disease risk analysis device 1000 according to one embodiment of this disclosure.

[0040] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can acquire original medical images. Specifically, the brain disease risk analysis device 1000 can acquire original medical images (e.g., MRA images, MRI images, and / or CT images, etc.) acquired through a medical image acquisition device. At this time, the brain disease risk analysis device 1000 can acquire pixel information from the original medical image. Here, pixel information may encompass coordinate information consisting of the x, y, and z coordinates of the original medical image and gray level information. On the other hand, the brain disease risk analysis device 1000 can acquire patient information stored on an external server (e.g., a hospital server) or database, and acquire pixel information of the original medical image from the original medical image included in the patient information.

[0041] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can acquire coordinate information corresponding to vascular regions from the pixel information of an original medical image. Specifically, the brain disease risk analysis device 1000 can acquire coordinate information related to a region of interest (for example, a region corresponding to cerebral blood vessels) from pixel information in which the gray level information contained in the pixel information of the original medical image is smaller than or equal to a predetermined value. Here, coordinate information may mean the x, y, and z coordinates of pixels constituting the original medical image, as described above. On the other hand, coordinate information may be referred to as point information or point cloud in that it relates to information for a specific point in the original medical image.

[0042] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can generate cerebral blood vessel shape information based on coordinate information corresponding to the vascular region.

[0043] As an example, the brain disease risk analysis device 1000 can perform the operation of generating a surface model based on coordinate information corresponding to a vascular region. Specifically, the brain disease risk analysis device 1000 can utilize computer vision segmentation techniques to generate a surface model showing the shape information of cerebral blood vessels from coordinate information corresponding to a region of interest. For example, the brain disease risk analysis device 1000 can be implemented to generate a surface model for a region of interest based on coordinates that constitute a single surface among the coordinate information selected from pixel information. On the other hand, the original medical image is composed of a set of 2D images and can contain 3D coordinate information. In this case, the brain disease risk analysis device 1000 can generate a 3D surface model showing the shape information of a region of interest (e.g., cerebral blood vessels) based on the 3D coordinate information.

[0044] As an example, the brain disease risk analysis device 1000 can perform the operation of generating a 1D model based on coordinate information corresponding to the vascular region. Specifically, the brain disease risk analysis device 1000 can generate a 1D model showing the shape information of the region of interest (e.g., cerebral blood vessels) based on coordinate information corresponding to the vascular region. Here, the 1D model refers to a model in which points corresponding to the aforementioned coordinate information of the region of interest are connected in an arbitrary manner.

[0045] For example, the brain disease risk analysis device 1000 can generate a mesh model based on coordinate information corresponding to the vascular region. For instance, the brain disease risk analysis device 1000 can generate a mesh model from a surface model generated based on coordinate information.

[0046] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can acquire fluid flow prediction values ​​based on coordinate information corresponding to the vascular region and shape information of cerebral blood vessels. Specifically, the brain disease risk analysis device 1000 can calculate fluid flow prediction values ​​(or blood flow information) related to blood flow velocity and / or blood flow pressure based on coordinate information corresponding to the vascular region and shape information of cerebral blood vessels. Furthermore, the brain disease risk analysis device 1000 can acquire boundary conditions of cerebral blood vessels based on shape information of cerebral blood vessels, and calculate or acquire blood flow information related to blood flow velocity and / or blood flow pressure based on boundary conditions.

[0047] For example, the brain disease risk analysis device 1000 can calculate fluid flow predictions from cerebral blood vessel geometry data (e.g., mesh model and / or surface model) using Computational Fluid Dynamics (CFD) or Computational Aided Engineering (CAE). For example, the brain disease risk analysis device 1000 can calculate fluid flow predictions from cerebral blood vessel geometry data (e.g., 1D model) through a formula that assumes the fluid flow is Poiseuille Flow.

[0048] On the other hand, the brain disease risk analysis device 1000 according to one embodiment of the present disclosure can learn a fluid flow model and perform operations to predict or calculate fluid flow values ​​through the fluid flow model.

[0049] For example, the brain disease risk analysis device 1000 can generate a one-dimensional model from a surface model using computer vision skeleton techniques, and then generate a fluid flow model from the generated one-dimensional model. Specifically, the brain disease risk analysis device 1000 can train a first neural network that generates a fluid flow model based on the one-dimensional model by applying an artificial intelligence algorithm, and then obtain the fluid flow model from the one-dimensional model through the trained first neural network.

[0050] As another example, the brain disease risk analysis device 1000 can generate a fluid flow model based on coordinate information. Specifically, the brain disease risk analysis device 1000 can train a second neural network that generates a fluid flow model based on coordinate information by applying an artificial intelligence algorithm, and can obtain the fluid flow model from the coordinate information through the trained second neural network.

[0051] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can perform the operation of training a brain disease risk analysis model. Specifically, the brain disease risk analysis device 1000 can train a brain disease risk analysis model based on cerebrovascular shape information, fluid flow prediction values ​​(or blood flow information), and / or patient information (e.g., patient's age, sex, brain disease risk, etc.). Subsequently, the brain disease risk analysis device 1000 according to one embodiment of this disclosure can use the trained brain disease risk analysis model to obtain patient-specific brain disease risk prediction results, and perform the operation of transmitting or outputting patient-specific brain disease risk prediction results. The method for training the brain disease risk analysis model will be described in detail later in relation to Figures 4 to 9.

[0052] In the following, with reference to Figures 4 and 5, we will specifically describe the method for learning a brain disease risk analysis model according to one embodiment of this application. Figure 4 is a schematic diagram illustrating the operation of a brain disease risk analysis device 1000 for learning a brain disease risk analysis model according to one embodiment of this disclosure. Figure 5 is a diagram illustrating the process of learning a brain disease risk analysis model according to one embodiment of this disclosure.

[0053] A brain disease risk analysis device 1000 according to one embodiment of the present disclosure can perform the operation of training a brain disease risk analysis model for calculating brain disease risk from medical images. Specifically, the brain disease risk analysis device 1000 can perform the operation of training a brain disease risk analysis model for calculating and quantifying brain disease risk based on cerebrovascular shape information, cerebrovascular blood flow information, and patient information acquired from medical images. More specifically, the brain disease risk analysis device 1000 can train a brain disease risk analysis model using a first training dataset X related to vascular shape information (xn) including vascular location information and vascular diameter information, a second training dataset Y related to blood flow information (vn) including blood flow velocity information and blood flow pressure information, a third training dataset R consisting of patient information (rn) related to the patient's age, gender, and / or presence or absence of brain disease, and a fourth training dataset related to the patient's brain disease risk (yn), and acquire the trained brain disease risk analysis model.

[0054] A brain disease risk analysis device 1000 according to one embodiment of the present disclosure can acquire a first training dataset related to cerebrovascular shape information (e.g., location information of blood vessels and / or diameter information of blood vessels) and train a brain disease risk analysis model based on the first training dataset.

[0055] For example, the first training dataset may consist of vascular shape information for each patient. For instance, the first training dataset may include the first shape information for the first patient, the second shape information for the second patient, and / or the Nth shape information for the Nth patient.

[0056] As another example, the first training dataset may consist of the difference between the shape information of a patient's blood vessels and the shape information of a standard blood vessel. For example, the first training dataset may consist of the difference between the first shape information of the blood vessels of the first patient and the shape information of a standard blood vessel, the difference between the second shape information of the blood vessels of the second patient and the shape information of a standard blood vessel, and / or the difference between the Nth shape information of the blood vessels of the Nth patient and the shape information of a standard blood vessel. In this case, the shape information of the standard blood vessels may be the shape information of a standard blood vessel corresponding to the patient's sex and age.

[0057] However, the above is merely an example, and the brain disease risk analysis model could be trained using a first training dataset appropriately configured in any way to achieve the objective of training a brain disease risk analysis model for predicting the risk of brain disease.

[0058] A brain disease risk analysis device 1000 according to one embodiment of the present disclosure can acquire a second training dataset related to cerebral blood flow information (e.g., blood flow velocity information, blood flow pressure information) and train a brain disease risk analysis model based on the second training dataset.

[0059] For example, the second training dataset may consist of blood flow information for each patient's blood vessels. For instance, the second training dataset could include the first blood flow information for the first patient, the second blood flow information for the second patient, and / or the Nth blood flow information for the Nth patient.

[0060] As another example, the second training dataset may consist of the difference between the blood flow information of a patient's blood vessels and the blood flow information of a standard blood vessel. For example, the second training dataset may consist of the difference between the first blood flow information of the first patient's blood vessels and the blood flow information of a standard blood vessel, the difference between the second blood flow information of the second patient's blood vessels and the blood flow information of a standard blood vessel, and / or the difference between the Nth blood flow information of the Nth patient's blood vessels and the blood flow information of a standard blood vessel. In this case, the standard blood vessels may be the standard blood vessels corresponding to the patient's sex and age.

[0061] However, the above is merely an example, and the brain disease risk analysis model could be trained using a second training dataset appropriately configured in any way to achieve the objective of training a brain disease risk analysis model for predicting the risk of brain disease.

[0062] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can acquire a third training dataset related to patient information and use the third training dataset to train a brain disease risk analysis model. Specifically, the third training dataset may consist of patient data for at least one of the following: patient age, patient sex, patient underlying disease, or patient race. Specifically, the brain disease risk analysis device 1000 can acquire a third training dataset based on patient information acquired from an external server (e.g., a hospital server).

[0063] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can acquire a fourth training dataset related to brain disease risk information and train a brain disease risk analysis model using the fourth training dataset. For example, the brain disease risk analysis device 1000 can assign brain disease risk information (e.g., brain disease probability value) based on location information included in blood vessel shape information and brain disease information included in patient information. More specifically, the brain disease risk analysis device 1000 can detect brain regions associated with brain diseases based on brain disease information included in patient information, and acquire a fourth training dataset by assigning brain disease risk information (e.g., brain disease probability value) to location information corresponding to the detected brain regions. As an example, the brain disease risk analysis device 1000 can calculate the probability value of brain disease occurring for each brain region based on patient information, and assign the calculated probability value (e.g., a value between 0 and 1) to each brain region using location information included in blood vessel shape information. For example, in the case of a patient with stroke disease, the brain disease risk analysis device 1000 can detect the brain region where a stroke actually occurred based on the patient information and assign stroke risk information (e.g., risk level 1) indicating that a stroke occurred to the location information corresponding to the detected brain region. For example, the brain disease risk analysis device 1000 can detect brain regions where a stroke has not occurred based on the patient information and assign stroke risk information (e.g., risk level 0) indicating that a stroke has not occurred to the location information corresponding to the detected brain region.

[0064] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can train a brain disease risk analysis model based on a first training dataset related to cerebrovascular morphology information, a second training dataset related to cerebrovascular blood flow information, a third training dataset related to patient information, and / or a fourth training dataset related to brain disease-specific risk information.

[0065] As an example, the brain disease risk analysis device 1000 can train a brain disease risk analysis model based on the patient's cerebrovascular data (e.g., cerebrovascular shape information or cerebrovascular blood flow information). This will be described in more detail in Figure 8.

[0066] As another example, the brain disease risk analysis device 1000 can train a brain disease risk analysis model based on a patient's cerebrovascular data (e.g., cerebrovascular shape information or cerebrovascular blood flow information) and standard cerebrovascular data (e.g., standard cerebrovascular shape information or standard cerebrovascular blood flow information). Specifically, the brain disease risk analysis device 1000 can train a brain disease risk analysis model based on the difference between a patient's cerebrovascular data (e.g., cerebrovascular shape information or cerebrovascular blood flow information) and standard cerebrovascular data (e.g., standard cerebrovascular shape information or standard cerebrovascular blood flow information). For example, the brain disease risk analysis device 1000 can train a brain disease risk analysis model using a training dataset related to the difference between a patient's cerebrovascular shape information and standard cerebrovascular shape information. For example, the brain disease risk analysis device 1000 can train a brain disease risk analysis model using a training dataset related to the difference between a patient's cerebrovascular blood flow information and standard cerebrovascular blood flow information. This will be described in more detail in Figure 9.

[0067] On the other hand, Figures 4 and 5 describe the brain disease risk analysis model as a single model. However, this is merely for explanatory purposes and should not be interpreted restrictively. It is possible to configure separate models to be trained for each training dataset, and to integrate the calculation of brain disease risk through a structure in which some layers of each model are shared.

[0068] For example, the first brain disease risk analysis model may be configured to obtain a first training dataset and a fourth training dataset related to the risk of each brain disease obtained from patient information, and to output a predicted brain disease risk value. In this case, the first brain disease risk analysis model may be trained by updating any parameters included in the first brain disease risk analysis model based on the predicted brain disease risk value and the fourth training dataset related to the risk of each brain disease.

[0069] For example, a second brain disease risk analysis model may be configured to obtain a second training dataset and a fourth training dataset related to the risk of each brain disease obtained from patient information, and to output a predicted brain disease risk value. In this case, the second brain disease risk analysis model may be trained by updating any parameters included in the second brain disease risk analysis model based on the predicted brain disease risk value and the fourth training dataset related to the risk of each brain disease.

[0070] In this case, some layers of the first brain disease risk analysis model and some layers of the second brain disease risk analysis model may be configured to have a structure that is shared with each other, and through this, it will be possible to predict the risk of brain disease by taking into account all the morphological information and blood flow information of the cerebral blood vessels.

[0071] The following will describe a method for learning a brain disease risk analysis model and a method for predicting brain disease risk according to one embodiment of this disclosure, with reference to Figures 6 to 9. Figure 6 is a flowchart illustrating a method for learning a brain disease risk analysis model according to one embodiment of this disclosure. In describing the method for learning a brain disease risk analysis model, some embodiments that overlap with the descriptions in Figures 2 to 5 may be omitted. However, this is for the sake of convenience of explanation and should not be interpreted restrictively.

[0072] A method for learning a brain disease risk analysis model according to one embodiment of this disclosure may include the steps of: acquiring a first learning dataset related to cerebrovascular morphology information (S1100); acquiring a second learning dataset related to cerebrovascular blood flow information (S1200); acquiring a third learning dataset related to patient information (S1300); acquiring a fourth learning dataset related to the risk of brain disease (S1400); and training a brain disease risk analysis model using the first, second, third, or fourth learning datasets (S1500).

[0073] In the step of acquiring a first training dataset related to cerebrovascular shape information (S1100), the brain disease risk analysis device 1000 can acquire a first training dataset X related to cerebrovascular shape information (e.g., vessel location information, vessel diameter information). As mentioned above, as an example, the first training dataset may consist of cerebrovascular shape information for each patient. For example, the first training dataset X may include the first shape information for the first patient, the second shape information for the second patient, and / or the Nth shape information for the Nth patient. As another example, the first training dataset may consist of the difference between the patient's cerebrovascular shape information and the standard cerebrovascular shape information. For example, the first training dataset may consist of the difference between the first shape information of the first patient's vessels and the standard cerebrovascular shape information, the difference between the second shape information of the second patient's vessels and the standard cerebrovascular shape information, and / or the difference between the Nth shape information of the Nth patient's vessels and the standard cerebrovascular shape information.

[0074] In the step of acquiring a second training dataset related to cerebral vascular blood flow information (S1200), the brain disease risk analysis device 1000 can acquire a second training dataset related to cerebral vascular blood flow information (e.g., blood flow velocity information, blood flow pressure information). As an example, the second training dataset may consist of blood flow information for each patient's blood vessels. For example, the second training dataset may include the first blood flow information for the first patient, the second blood flow information for the second patient, and / or the Nth blood flow information for the Nth patient. As another example, the second training dataset may consist of the difference between the blood flow information for the patient's blood vessels and the blood flow information for a standard blood vessel. For example, the second training dataset may consist of the difference between the first blood flow information for the first patient's blood vessels and the blood flow information for a standard blood vessel, the difference between the second blood flow information for the second patient's blood vessels and the blood flow information for a standard blood vessel, and / or the difference between the Nth blood flow information for the Nth patient's blood vessels and the blood flow information for a standard blood vessel.

[0075] In the step of acquiring a third training dataset related to patient information (S1300), the brain disease risk analysis device 1000 can acquire a third training dataset related to patient information (e.g., patient's age, patient's sex, patient's underlying medical conditions, and / or patient's race).

[0076] In the stage of acquiring a fourth training dataset related to the risk of brain disease (S1400), the brain disease risk analysis device 1000 can acquire a fourth training dataset related to the risk of brain disease. As mentioned above, the fourth training dataset may be related to brain disease risk information (e.g., brain disease probability value) assigned based on the positional information included in the vascular shape information and the brain disease information included in the patient information. Specifically, the brain disease risk analysis device 1000 can calculate the probability value of brain disease occurring for each brain region based on the patient information, and acquire a fourth training dataset by assigning the calculated probability value (e.g., a value between 0 and 1) to each brain region using the positional information included in the vascular shape information.

[0077] In the stage (S1500) of training the brain disease risk analysis model using the first, second, third, or fourth training dataset, the brain disease risk analysis device 1000 can train the brain disease risk analysis model based on the first training dataset related to cerebrovascular shape information, the second training dataset related to cerebrovascular blood flow information, the third training dataset related to patient information, and / or the fourth training dataset related to the risk of brain diseases. Specifically, the brain disease risk analysis device 1000 can train the brain disease risk analysis model based on the predicted values ​​output through the brain disease risk analysis model and the risk of brain diseases included in the fourth training dataset. For example, the brain disease risk analysis device 1000 can train the brain disease risk analysis model by updating the parameters included in the brain disease risk analysis model so that the predicted values ​​output through the brain disease risk analysis model approximate the risk of brain diseases included in the fourth training dataset.

[0078] Refer to Figure 7. Figure 7 is a diagram illustrating the training and verification of a brain disease risk analysis model according to one embodiment of the present disclosure. The brain disease risk analysis device 1000 according to one embodiment of the present disclosure can perform operations to preprocess training data consisting of a first training dataset, a second training dataset, a third training dataset, and / or a fourth training dataset. For example, the brain disease risk analysis device 1000 can perform normalization on each data point included in the training data consisting of the first training dataset, the second training dataset, the third training dataset, and / or the fourth training dataset.

[0079] In addition, the brain disease risk analysis device 1000 according to one embodiment of this application can perform the operation of dividing preprocessed learning data into a training dataset and a test dataset (or validation dataset). For example, the brain disease risk analysis device 1000 can randomly divide preprocessed learning data into a training dataset and a validation dataset. At this time, the brain disease risk analysis device 1000 can train a brain disease risk analysis model using the training dataset through any machine learning algorithm (ML Algorithm). At this time, the brain disease risk analysis device 1000 can test or validate the performance of the learned brain disease risk analysis model using the validation dataset. For example, the brain disease risk analysis device 1000 can calculate the performance of the learned brain disease risk analysis model, compare the calculated performance with a predetermined target performance, and perform the aforementioned learning of the brain disease risk analysis model until the calculated performance is greater than the predetermined target performance, thereby obtaining an optimized brain disease risk analysis model.

[0080] Figure 8 is a diagram illustrating the process of training a brain disease risk analysis model according to one embodiment of this disclosure and the process of predicting brain disease risk using the trained brain disease risk analysis model.

[0081] A brain disease risk analysis device 1000 according to one embodiment of this disclosure can train a brain disease risk analysis model f based on a first training dataset X related to the shape information of the patient's blood vessels obtained from the patient's blood vessel image, a second training dataset V related to the patient's blood flow information, a third training dataset R related to patient information, and a fourth training dataset Y1 related to brain disease-specific risk information obtained from patient information. Specifically, the brain disease risk analysis device 1000 can input the first training dataset X, the second training dataset V, the third training dataset R, and the fourth training dataset Y1 into the brain disease risk analysis model f and obtain a brain disease risk prediction value output through the brain disease risk analysis model f. At this time, the brain disease risk analysis device 1000 can update any parameters included in the brain disease risk analysis model based on the brain disease risk prediction value and brain disease-specific risk information. Specifically, the brain disease risk analysis device 1000 compares the predicted brain disease risk value with the brain disease-specific risk information Y1, updates arbitrary parameters included in the brain disease risk analysis model f so that the predicted brain disease risk value approximates the brain disease-specific risk information, trains the brain disease risk analysis model f, and obtains the trained brain disease risk analysis model f''.

[0082] Furthermore, a brain disease risk analysis device 1000 according to one embodiment of this disclosure can predict brain disease risk using a trained brain disease risk analysis model f''. Specifically, the brain disease risk analysis device 1000 can be configured to acquire target shape information TX, target blood flow information TV, and target patient information TR from the blood vessel image of the target patient to be analyzed, and to input the target shape information TX, target blood flow information TV, and target patient information TR into the trained brain disease risk analysis model f''. At this time, the brain disease risk analysis device 1000 can acquire a brain disease risk value Y output through the trained brain disease risk analysis model f''. At this time, as described above, the trained brain disease risk analysis model f'' has been trained to output brain disease-specific risk information Y1 based on blood vessel shape information X, blood vessel blood flow information V, and patient information R, so it can output a brain disease risk value Y based on the target shape information TX, target blood flow information TV, and target patient information TR of the target patient.

[0083] Figure 9 is a diagram illustrating how to train a brain disease risk analysis model according to another embodiment of this disclosure and how to predict brain disease risk using the trained brain disease risk analysis model.

[0084] In another embodiment of the brain disease risk analysis device 1000 of this disclosure, a brain disease risk analysis model f can be trained based on a first training dataset X2 consisting of the difference between the shape information of the patient's blood vessels obtained from the patient's vascular image and the shape information of standard blood vessels, a second training dataset V2 consisting of the difference between the patient's blood flow information and the blood flow information of standard blood vessels, a third training dataset R related to patient information, and a fourth training dataset Y1 related to brain disease risk information obtained from patient information. Specifically, the brain disease risk analysis device 1000 can input the first training dataset X2, the second training dataset V2, the third training dataset R, and the fourth training dataset Y1 into the brain disease risk analysis model f and obtain a predicted brain disease risk value output through the brain disease risk analysis model f. At this time, the brain disease risk analysis device 1000 can update any parameters included in the brain disease risk analysis model based on the predicted brain disease risk value and the brain disease risk information Y1. Specifically, the brain disease risk analysis device 1000 compares the predicted brain disease risk value with the brain disease-specific risk information Y1, updates arbitrary parameters included in the brain disease risk analysis model f so that the predicted brain disease risk value approximates the brain disease-specific risk information Y1, and trains the brain disease risk analysis model f, thereby obtaining the trained brain disease risk analysis model f''.

[0085] Furthermore, a brain disease risk analysis device 1000 according to another embodiment of this disclosure can predict brain disease risk using a trained brain disease risk analysis model f''. Specifically, the brain disease risk analysis device 1000 may be configured to acquire the difference TX2 between target shape information obtained from the vascular image of the target patient and the shape information of a standard vascular vessel, the difference TV2 between the target blood flow information of the target patient and the blood flow information of a standard vascular vessel, and the target patient information TR of the target patient, and to input the difference TX2 between the target shape information and the shape information of a standard vascular vessel, the difference TV2 between the target blood flow information and the blood flow information of a standard vascular vessel, and the target patient information TR into a trained brain disease risk analysis model f''. At this time, the brain disease risk analysis device 1000 can obtain the brain disease risk value Y output through the brain disease risk analysis model f'' that has completed training. At this time, the brain disease risk analysis model f'' that has completed training has been trained to output brain disease risk information based on the first training dataset consisting of the difference X2 between the shape information of blood vessels and the shape information of standard blood vessels, the second training dataset consisting of the difference V2 between the blood flow information of blood vessels and the blood flow information of standard blood vessels, and patient information R, as described above, so it can output the brain disease risk value Y based on the difference TX2 between the shape information of the target and the shape information of standard blood vessels, the difference TV2 between the blood flow information of the target and the blood flow information of standard blood vessels, and the target patient information TR.

[0086] The trained brain disease risk analysis model f'' can output a personalized, patient-specific brain disease risk assessment for each patient, taking into account the patient's cerebrovascular morphology, blood flow velocity, blood pressure-related blood flow information, and patient information such as age, sex, underlying diseases, and race.

[0087] On the other hand, Figure 8 mainly illustrates a method for predicting the risk of brain disease based on a patient's cerebrovascular data (e.g., cerebrovascular shape information or cerebrovascular blood flow information), while Figure 9 mainly illustrates a method for predicting the risk of brain disease based on the difference between a patient's cerebrovascular data and standard cerebrovascular data. However, these are merely examples and can be selected, combined, or modified in any appropriate way.

[0088] According to a learning method for a brain disease risk analysis model as described in one embodiment of this disclosure, a brain disease risk analysis model can be provided that can automatically calculate or predict the brain disease risk based on cerebrovascular shape information, cerebrovascular blood flow information, and / or patient information.

[0089] The various operations of the brain disease risk analysis device 1000 described above may be stored in the memory 1200 of the brain disease risk analysis device 1000, and the processor 1300 of the brain disease risk analysis device 1000 may be provided to perform the operations stored in the memory 1200.

[0090] The features, structures, and effects described in the embodiments above are included in at least one embodiment of this disclosure and are not necessarily limited to just one embodiment. Furthermore, the features, structures, and effects exemplified in each embodiment can be combined or modified and implemented in other embodiments by a person with ordinary skill in the art to which the embodiment belongs. Therefore, content related to such combinations and modifications should be interpreted as being included within the scope of this disclosure.

[0091] Furthermore, although the above description has focused on embodiments, these are merely illustrative and do not limit the disclosure. A person with ordinary skill in the art to which this disclosure belongs will understand that various modifications and applications not exemplified above are possible without departing from the essential characteristics of these embodiments. In other words, each component specifically shown in the embodiments can be modified and implemented. Differences related to such modifications and applications should be interpreted as being within the scope of this disclosure as defined by the attached claims. [Industrial applicability]

[0092] The methods described above for predicting the risk of brain disease, the methods for training brain disease risk analysis models, and brain disease risk analysis devices can be applied to the healthcare service sector, which provides medical services.

Claims

1. In a method for predicting the risk of brain disease, the brain disease risk analysis device The stage of obtaining a completed brain disease risk analysis model; The stage of acquiring target shape information for the cerebral blood vessels of the target patient; The stage of obtaining target blood flow information for the cerebral blood vessels of the target patient; The stage of obtaining patient information for the target patients; and The step includes inputting the target shape information, the target blood flow information, and the target patient information into the brain disease risk analysis model and obtaining a brain disease risk value output through the brain disease risk analysis model; The aforementioned target blood flow information is obtained through a learned fluid flow model. The aforementioned fluid flow model is, A one-dimensional model representing the target shape information is generated based on coordinate information corresponding to the vascular region, an artificial intelligence algorithm is applied to train a first neural network that generates a fluid flow model based on the one-dimensional model, and the fluid flow model is obtained from the one-dimensional model through the first neural network after the training is complete, or A second neural network is trained to generate a fluid flow model based on the coordinate information by applying an artificial intelligence algorithm, and the fluid flow model is obtained from the coordinate information through the trained second neural network. A method for predicting the risk of brain disease, obtained through [means / method].

2. The step of acquiring the aforementioned target shape information is: Steps to acquire standard blood vessel shape information; and The step of obtaining the difference between the target shape information and the standard blood vessel shape information; further includes, The step of obtaining the aforementioned brain disease risk value is: A method for predicting the risk of brain disease according to claim 1, further comprising the steps of: inputting the difference between the target shape information and the standard blood vessel shape information into the brain disease risk analysis model; and obtaining a brain disease risk value output through the brain disease risk analysis model.

3. The step of acquiring the target blood flow information is as follows: The step of obtaining blood flow information from a standard vessel; and The step of obtaining the difference between the target blood flow information and the standard blood vessel blood flow information; further includes, The step of obtaining the aforementioned brain disease risk value is: A method for predicting the risk of brain disease according to claim 2, further comprising the steps of: inputting the difference between the target blood flow information and the standard blood vessel blood flow information into the brain disease risk analysis model; and obtaining a brain disease risk value output through the brain disease risk analysis model.

4. The aforementioned brain disease risk analysis model is, A method for predicting the risk of brain disease according to claim 1, which is trained on a first training dataset related to cerebrovascular shape information including location information or diameter information of cerebrovascular vessels, a second training dataset related to cerebrovascular blood flow information including blood flow velocity information or blood flow pressure information of cerebrovascular vessels, a third training dataset related to patient information including patient age or gender, and a fourth training dataset related to brain disease information.

5. The aforementioned brain disease risk analysis model is, The system is configured to acquire the first training dataset, the second training dataset, the third training dataset, and the fourth training dataset, and to output a predicted brain disease risk value. The aforementioned brain disease risk analysis model is, A method for predicting the risk of brain disease according to claim 4, wherein the model is trained to output a predicted brain disease risk value that approximates the brain disease information included in the fourth training dataset.

6. The aforementioned brain disease risk analysis model is, A method for predicting the risk of brain disease according to claim 3, which is trained on a first training dataset consisting of the difference between the shape information of a patient's cerebral blood vessels and the shape information of standard cerebral blood vessels; a second training dataset consisting of the difference between the blood flow information of a patient's cerebral blood vessels and the blood flow information of standard cerebral blood vessels; a third training dataset related to patient information including the patient's age or the patient's gender; and a fourth training dataset related to brain disease information.

7. The aforementioned brain disease risk analysis model is, The system is configured to acquire the first training dataset, the second training dataset, the third training dataset, and the fourth training dataset, and to output a predicted brain disease risk value. The aforementioned brain disease risk analysis model is, A method for predicting the risk of brain disease according to claim 6, which is trained to output a predicted brain disease risk value that approximates the brain disease information included in the fourth training dataset.

8. A computer-readable recording medium having a program stored on it that causes a computer to perform the method described in any one of claims 1 to 7.

Images