Method for calculating cerebral blood flow data and method for training neural network model for calculating cerebral blood flow data

Abstract

A method for calculating cerebral blood flow data according to one embodiment of the present disclosure includes the steps of: acquiring a trained cerebral blood flow prediction model; acquiring an original medical image; acquiring shape data corresponding to a cerebral vascular region from the original medical image, the shape data being composed of point cloud-type data; acquiring a boundary region of the cerebral vascular region and acquiring boundary information corresponding to the boundary region; acquiring initial condition information or boundary condition information; and inputting the shape data, the boundary information, and the initial condition information or boundary condition information into the cerebral blood flow prediction model, and acquiring cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the cerebral blood flow prediction model.

Description

[Technical Field]

[0001] The present disclosure relates to a method for calculating cerebral blood flow data, and more particularly to a method and apparatus for training a neural network model for calculating cerebral blood flow data, and a method and apparatus for calculating cerebral blood flow data using a trained neural network model. [Background technology]

[0002] To visualize the three-dimensional structure of the brain and cerebral blood vessels, various imaging techniques have been developed, such as positron emission tomography (PET), functional magnetic resonance imaging (fMRI), magnetic resonance angiography (MRA), and computed tomography. These advances in imaging techniques have made it possible to measure cerebral blood flow throughout the brain.

[0003] However, conventional methods have been used to calculate cerebral blood flow using blood flow calculation software or arbitrary mathematical / statistical techniques. However, these methods have the limitation that new cerebral blood flow must be calculated each time new data is acquired using calculation software or mathematical / statistical techniques. Furthermore, conventional methods for calculating cerebral blood flow have technical barriers and are cumbersome, requiring the creation of a surface or solid model related to the shape of the brain in order to calculate cerebral blood flow.

[0004] For this reason, there is a need to develop new techniques for efficiently calculating cerebral blood flow. Summary of the Invention [Problem to be solved by the invention]

[0005] One problem to be solved by the present disclosure is to provide a method for calculating cerebral blood flow data and a method for training a neural network model for calculating cerebral blood flow data.

[0006] The problems to be solved by the present disclosure are not limited to the problems described above, and problems not mentioned will be clearly understood by a person having ordinary skill in the art to which the present disclosure pertains from this specification and the accompanying drawings. [Means for solving the problem]

[0007] This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to aid in determining the scope of the claimed subject matter.

[0008] A method for calculating cerebral blood flow data according to one embodiment of the present disclosure may include the steps of: acquiring a trained cerebral blood flow prediction model; acquiring an original medical image; acquiring shape data corresponding to a cerebral vascular region from the original medical image, wherein the shape data is configured as point cloud data; acquiring a boundary region of the cerebral vascular region and acquiring boundary information corresponding to the boundary region; acquiring initial condition information or boundary condition information; and inputting the shape data, the boundary information, and the initial condition information or boundary condition information into the cerebral blood flow prediction model, and acquiring cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the cerebral blood flow prediction model.

[0009] A method for calculating cerebral blood flow data according to another embodiment of the present disclosure may include the steps of: acquiring a trained cerebral blood flow prediction model; acquiring an original medical image; acquiring first shape data corresponding to a cerebral vascular region from the original medical image, the first shape data being composed of node information and connection information constituting a 1D network model; acquiring a boundary region of the cerebral vascular region and acquiring boundary information corresponding to the boundary region; acquiring initial condition information or boundary condition information; and inputting the shape data, the boundary information, and the initial condition information or boundary condition information into the cerebral blood flow prediction model, and acquiring cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the cerebral blood flow prediction model.

[0010] The technical solutions of the present disclosure are not limited to the above-mentioned technical solutions, and any unmentioned technical solutions will be clearly understood by a person skilled in the art to which the present disclosure pertains from this specification and the accompanying drawings. [Effects of the Invention]

[0011] According to a method for calculating cerebral blood flow data, a method for learning a cerebral blood flow prediction model, and a cerebral blood flow calculation device according to one embodiment of the present disclosure, it is possible to learn a cerebral blood flow prediction model that can calculate cerebral blood flow information from a point model or a 1D network model, rather than a surface model or a solid model.

[0012] According to a method for calculating cerebral blood flow data, a method for learning a cerebral blood flow prediction model, and a cerebral blood flow calculation device according to one embodiment of the present disclosure, a trained cerebral blood flow prediction model can be used to automatically calculate cerebral blood flow information from a point model or a 1D network model, rather than a surface model or a solid model.

[0013] The effects of the present disclosure are not limited to the effects described above, and unmentioned effects will be clearly understood by those having ordinary skill in the art to which the present disclosure pertains from this specification and the accompanying drawings. [Brief explanation of the drawings]

[0014] [Figure 1] 1 is a schematic diagram of a cerebral blood flow calculation device according to an embodiment of the present disclosure. FIG. [Figure 2] FIG. 1 is a schematic diagram illustrating an operation of a cerebral blood flow calculation device according to an embodiment of the present disclosure. [Figure 3] 1 is a diagram illustrating an aspect of calculating cerebral blood flow data using a first cerebral blood flow prediction model according to one embodiment of the present disclosure. [Figure 4] 1 is a diagram illustrating an aspect of training a first cerebral blood flow prediction model according to an embodiment of the present disclosure. [Figure 5] 10 is a diagram illustrating another aspect of training the first cerebral blood flow prediction model according to an embodiment of the present disclosure. [Figure 6] 1 is a diagram illustrating an aspect of training a first cerebral blood flow prediction model according to an embodiment of the present disclosure and an aspect of verifying the performance of the first cerebral blood flow prediction model. [Figure 7] 10 is a diagram illustrating an aspect of calculating cerebral blood flow data using a second cerebral blood flow prediction model according to another embodiment of the present disclosure. [Figure 8] 10 is a diagram illustrating an aspect of training a second cerebral blood flow prediction model according to another embodiment of the present disclosure. [Figure 9] 10 is a diagram illustrating another aspect of training a second cerebral blood flow prediction model according to another embodiment of the present disclosure. [Figure 10] 10 is a diagram illustrating an aspect of training a second cerebral blood flow prediction model according to another embodiment of the present disclosure and an aspect of verifying the performance of the second cerebral blood flow prediction model. [Figure 11] 1 is a flowchart illustrating a method for calculating cerebral blood flow data using a first cerebral blood flow prediction model according to one embodiment of the present disclosure. [Figure 12]10 is a flowchart illustrating a method for calculating cerebral blood flow data using a second cerebral blood flow prediction model according to another embodiment of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0015] The above-mentioned technical problems, features, and advantages of the present disclosure will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. However, since the present disclosure can be modified in various ways and can have various embodiments, the following will describe in detail specific embodiments by way of example in the drawings.

[0016] The same reference numerals will generally refer to the same elements throughout the specification. In addition, the same reference numerals will be used to describe elements having the same functions within the same concept shown in the drawings of each embodiment, and redundant descriptions thereof will be omitted.

[0017] If it is determined that a detailed description of a known function or configuration related to the present disclosure may unnecessarily obscure the gist of the present disclosure, the detailed description will be omitted. Furthermore, numbers (e.g., 1, 2, etc.) used in the description of this specification are merely identification symbols for distinguishing one component from another.

[0018] Furthermore, the suffixes "module" and "section" used in the following embodiments for components are given or used interchangeably solely for the sake of ease of writing the specification, and do not have any distinct meanings or roles in themselves.

[0019] In the following embodiments, the singular expression includes the plural expression unless the context clearly indicates otherwise.

[0020] In the following embodiments, terms such as "include" or "have" mean that the features or components described in the specification are present, and do not preclude the possibility that one or more other features or components may be added.

[0021] In the drawings, the size of components may be exaggerated or reduced for the sake of convenience of explanation. For example, the size and thickness of each component shown in the drawings are arbitrarily shown for the sake of convenience of explanation, and the present disclosure is not necessarily limited to what is shown in the drawings.

[0022] When an embodiment can be implemented differently, the order of certain processes may be performed differently than described. For example, two processes described in succession may be performed substantially simultaneously or may be performed in the reverse order from that described.

[0023] In the following embodiments, when elements are said to be connected, this includes not only the case where the elements are directly connected, but also the case where the elements are indirectly connected with another element interposed between them.

[0024] For example, when it is stated in this specification that components are electrically connected, this includes not only cases where the components are directly electrically connected, but also cases where the components are indirectly electrically connected through an intervening component.

[0025] A method for calculating cerebral blood flow data according to one embodiment of the present disclosure may include the steps of: acquiring a trained cerebral blood flow prediction model; acquiring an original medical image; acquiring shape data corresponding to a cerebral vascular region from the original medical image, wherein the shape data is configured as point cloud data; acquiring a boundary region of the cerebral vascular region and acquiring boundary information corresponding to the boundary region; acquiring initial condition information or boundary condition information; and inputting the shape data, the boundary information, and the initial condition information or boundary condition information into the cerebral blood flow prediction model, and acquiring cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the cerebral blood flow prediction model.

[0026] According to one embodiment of the present disclosure, the cerebral blood flow prediction model may be trained based on a learning dataset including first data related to shape data corresponding to a cerebral vascular region in the form of a point cloud, second data related to boundary information of the cerebral vascular region, third data related to initial condition information or boundary condition information, and fourth data related to a cerebral blood flow prediction value for cerebral blood flow velocity or pressure.

[0027] According to one embodiment of the present disclosure, the cerebral blood flow prediction model is configured to receive the training data set and output an output value from the training data set, and parameters included in the cerebral blood flow prediction model can be updated and trained to output the output value so as to approximate the cerebral blood flow predicted value.

[0028] According to one embodiment of the present disclosure, the fourth data is generated by interpolating the blood flow velocity or blood flow pressure calculated from the shape data corresponding to the cerebrovascular region into the shape data in the form of a point cloud, and the calculated blood flow velocity or blood flow pressure may be calculated using Computational Fluid Dynamics (CFD) based on a surface model generated from the shape data.

[0029] According to one embodiment of the present disclosure, the fourth data is generated by interpolating the blood flow velocity or blood flow pressure calculated from the shape data corresponding to the cerebrovascular region into the shape data in the form of a point cloud, and the calculated blood flow velocity or blood flow pressure can be calculated from a 1D network model generated from the shape data.

[0030] A method for calculating cerebral blood flow data according to another embodiment of the present disclosure may include the steps of: acquiring a trained cerebral blood flow prediction model; acquiring an original medical image; acquiring first shape data corresponding to a cerebral vascular region from the original medical image, the first shape data being composed of node information and connection information constituting a 1D network model; acquiring a boundary region of the cerebral vascular region and acquiring boundary information corresponding to the boundary region; acquiring initial condition information or boundary condition information; and inputting the shape data, the boundary information, and the initial condition information or boundary condition information into the cerebral blood flow prediction model, and acquiring cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the cerebral blood flow prediction model.

[0031] According to another embodiment of the present disclosure, the cerebral blood flow prediction model may be trained based on a learning dataset including first data consisting of node information and connectivity information constituting a 1D network model, second data related to boundary information of the cerebral vascular region, third data related to initial condition information or boundary condition information, and fourth data related to predicted values of cerebral blood flow for cerebral blood flow velocity or pressure.

[0032] According to another embodiment of the present disclosure, the cerebral blood flow prediction model is configured to receive the training data set and output an output value from the training data set, and parameters included in the cerebral blood flow prediction model may be updated and trained to output the output value so as to approximate the cerebral blood flow predicted value.

[0033] According to another embodiment of the present disclosure, the fourth data is generated by interpolating the blood flow velocity or blood flow pressure calculated from the second shape data in the form of a point cloud corresponding to the cerebrovascular region to the first shape data in the form of a 1D network model, and the calculated blood flow velocity or blood flow pressure may be calculated using Computational Fluid Dynamics (CFD) based on a surface model generated from the second shape data.

[0034] According to another embodiment of the present disclosure, the fourth data may be generated by calculating a predicted cerebral blood flow value related to the velocity or pressure of blood flow from the first shape data.

[0035] According to one embodiment of the present disclosure, a computer-readable recording medium having a program recorded thereon for executing the cerebral blood flow data calculation method can be provided.

[0036] Below, with reference to Figures 1 to 12, we will explain a method for calculating cerebral blood flow data, a method for learning a cerebral blood flow prediction model, and a cerebral blood flow calculation device (or cerebral blood flow calculation server, hereinafter referred to as the cerebral blood flow calculation device) according to an embodiment of the present disclosure.

[0037] FIG. 1 is a schematic diagram of a cerebral blood flow calculation device according to an embodiment of the present disclosure.

[0038] The cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure can train a cerebral blood flow prediction model, and can calculate or predict cerebral blood flow data from original medical images using the trained cerebral blood flow prediction model.

[0039] The cerebral blood flow calculation device 1000 according to one embodiment of the present disclosure may include a transceiver unit 1100, a memory 1200, and a processor 1300.

[0040] The transceiver 1100 can communicate with any external device, including a medical imaging device (e.g., an MRA device, an MRI device, a CT device, a PET device, etc.). For example, the cerebral blood flow calculation device 1000 can acquire medical images captured by a medical imaging device through the transceiver 1100. The cerebral blood flow calculation device 1000 can also acquire any execution data for executing a trained cerebral blood flow prediction model through the transceiver 1100. Here, the execution data may encompass any appropriate data for executing the cerebral blood flow prediction model, including structural information, hierarchical information, a calculation library, and a parameter set related to weights included in the cerebral blood flow prediction model. The cerebral blood flow calculation device 1000 can also transmit or output cerebral blood flow data acquired through the cerebral blood flow prediction model to any external device, including a user terminal, through the transceiver 1100.

[0041] The cerebral blood flow calculation device 1000 can connect to a network via the transceiver 1100 to transmit and receive various data. The transceiver 1100 can be broadly classified into a wired type and a wireless type. Since the wired type and the wireless type each have advantages and disadvantages, the cerebral blood flow calculation device 1000 may be provided with both a wired type and a wireless type. Here, in the case of the wireless type, a communication method based on a WLAN (Wireless Local Area Network) such as Wi-Fi can be mainly used. Alternatively, in the case of the wireless type, a communication method based on cellular communication, for example, LTE or 5G, 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.

[0042] The memory 1200 can store various information. Various data can be temporarily or semi-permanently stored in the memory 1200. Examples of the memory 1200 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). The memory 1200 can be provided in a form built into the cerebral blood flow calculation device 1000 or in a removable form. The memory 1200 can store various data necessary for the operation of the cerebral blood flow calculation device 1000, including an operating system (OS) for driving the cerebral blood flow calculation device 1000 and programs for operating each component of the cerebral blood flow calculation device 1000.

[0043] The processor 1300 can control the overall operation of the cerebral blood flow calculation device 1000. For example, the processor 1300 can control the overall operation of the cerebral blood flow calculation device 1000, including an operation of acquiring 3D shape data from an original medical image, an operation of acquiring boundary information, an operation of acquiring initial conditions or boundary conditions, an operation of training a cerebral blood flow prediction model, and / or an operation of calculating cerebral blood flow data through the trained cerebral blood flow prediction model, etc. Specifically, the processor 1300 can load and execute programs for the overall operation of the cerebral blood flow calculation device 1000 from the memory 1200. The processor 1300 can be embodied as an application processor (AP), a central processing unit (CPU), or similar devices, depending on hardware, software, or a combination thereof. In this case, the hardware can be provided in the form of an electronic circuit that processes electrical signals and performs control functions, and the software can be provided in the form of a program or code that drives a hardware circuit.

[0044] FIG. 2 is a schematic diagram illustrating an operation of a cerebral blood flow calculation device according to an embodiment of the present disclosure.

[0045] A cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure can calculate cerebral blood flow data from 3D shape data using a trained cerebral blood flow prediction model. Specifically, the cerebral blood flow calculation device 1000 can detect an area corresponding to cerebral blood vessels from pixel information of an original medical image and acquire shape data of the detected cerebral blood vessel area. Here, pixel information may refer to coordinate information consisting of x-, y-, and z-coordinates of the original medical image and gray level information. Furthermore, the cerebral blood flow calculation device 1000 can be configured to input 3D shape data corresponding to the cerebral blood vessel area into the trained cerebral blood flow prediction model. At this time, the cerebral blood flow calculation device 1000 can acquire cerebral blood flow data output through the cerebral blood flow prediction model.

[0046] According to one embodiment, the cerebral blood flow calculation device 1000 can calculate cerebral blood flow data based on 3D shape data configured in the form of a point cloud. This will be described in more detail with reference to FIGS. 3 to 6.

[0047] According to another embodiment, the cerebral blood flow calculation device 1000 can calculate cerebral blood flow data based on 3D shape data configured in the form of a 1D network that connects points (or nodes) included in coordinate information corresponding to the cerebral blood vessel region. This will be described in more detail with reference to FIGS. 7 to 10.

[0048] Hereinafter, various operations of the cerebral blood flow calculation device 100 according to an embodiment of the present disclosure, a method for calculating cerebral blood flow data, and a method for training a cerebral blood flow prediction model will be described in detail with reference to Figures 3 to 6. Figure 3 is a diagram illustrating how cerebral blood flow data is calculated using a first cerebral blood flow prediction model according to an embodiment of the present disclosure. According to this embodiment, cerebral blood flow data can be calculated from 3D shape data in the form of a point cloud through the first cerebral blood flow prediction model that has completed training.

[0049] A cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure can acquire an original medical image and obtain coordinate information of a region of interest (e.g., a region corresponding to a cerebral blood vessel) from pixel information of the original medical image. For example, the cerebral blood flow calculation device 1000 can acquire a region of interest using an image segmentation technique and obtain coordinate information corresponding to the region of interest. For example, the cerebral blood flow calculation device 1000 can obtain coordinate information of the region of interest (e.g., a region corresponding to a cerebral blood vessel) from pixel information of the medical image whose gray level information is less than or equal to a predetermined value. As described above, the coordinate information may refer to the x-, y-, and z-coordinates of pixels constituting the original medical image. Furthermore, the cerebral blood flow calculation device 1000 can acquire 3D shape data (e.g., referred to as point information or a point model) in the form of a point cloud based on the coordinate information of the region of interest.

[0050] A cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure can acquire a boundary region of a region of interest (e.g., a region corresponding to a cerebral blood vessel) and acquire boundary information corresponding to the boundary region (e.g., coordinate information of the boundary region). Specifically, the cerebral blood flow calculation device 1000 can acquire nodes corresponding to the boundary region of the region of interest (e.g., a cerebral blood vessel region) using an image segmentation technique and acquire the boundary information. As a result, the cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure can acquire three-dimensional initial conditions and / or boundary conditions associated with the boundary region.

[0051] The cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure can calculate cerebral blood flow data using a trained first cerebral blood flow prediction model. Specifically, the cerebral blood flow calculation device 1000 inputs 3D shape data (e.g., a points model), boundary information, and / or 3D initial condition information and boundary condition information into the trained first cerebral blood flow prediction model, and can obtain cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the trained first cerebral blood flow prediction model. As described below, the first cerebral blood flow prediction model is trained to output cerebral blood flow predicted values related to cerebral blood flow velocity and pressure based on point information, boundary information, and / or 3D initial condition information and boundary condition information. Therefore, the trained first cerebral blood flow prediction model can output cerebral blood flow data related to cerebral blood flow velocity and pressure from 3D shape data of a cerebral vascular region in the form of a point cloud, boundary information of the cerebral vascular region, and / or initial condition information or boundary condition information.

[0052] Hereinafter, a method for training a first cerebral blood flow prediction model and a method for generating a training dataset according to an embodiment of the present disclosure will be described in more detail with reference to Figures 4 and 5. According to an embodiment of the present disclosure, the training dataset for training the first cerebral blood flow prediction model includes data related to a cerebral blood flow predicted value.

[0053] According to one aspect, the predicted cerebral blood flow value can be calculated using a surface model and / or a solid model, which will be described in more detail with reference to FIG.

[0054] In another aspect, the predicted cerebral blood flow value can be calculated using a 1D network model, which will be described in more detail with reference to FIG.

[0055] FIG. 4 is a diagram illustrating an aspect of training a first cerebral blood flow prediction model according to an embodiment of the present disclosure.

[0056] The first cerebral blood flow prediction model may be trained based on a first learning dataset including: first data (Node Info(x, y, z) in FIG. 4) related to three-dimensional shape data of the cerebral vascular region in the form of a point cloud; second data (Boundary info in FIG. 4) related to boundary information of the cerebral vascular region; third data (initial / boundary values (bcv) in FIG. 4) related to initial condition information and boundary condition information; and fourth data (Flow data(v, p) in FIG. 4) related to blood flow velocity and pressure. Specifically, the cerebral blood flow calculation device 1000 can generate the first data, second data, and third data from an original medical image (raw image) in the same manner as described in FIG. 3.

[0057] Furthermore, the cerebral blood flow calculation device 1000 may calculate a predicted cerebral blood flow value using a surface model and / or a solid model, and may interpolate the predicted cerebral blood flow value to 3D shape data in the form of a point cloud to obtain fourth data related to the velocity and pressure of blood flow (Flow data (v, p) in FIG. 4). Specifically, the cerebral blood flow calculation device 1000 may generate a surface model or solid model representing the shape of cerebral blood vessels from coordinate information corresponding to a region of interest (e.g., a cerebral blood vessel region) using a computer vision segmentation technique. For example, the cerebral blood flow calculation device 1000 may be embodied to generate a surface model for a region of interest based on coordinates constituting a surface among the coordinate information. Meanwhile, an original medical image may be composed of a collection of 2D images and may contain 3D information. At this time, the cerebral blood flow calculation device 1000 can generate a three-dimensional surface model that indicates the shape of the region of interest (for example, the cerebral blood vessel region) based on the three-dimensional coordinate information.

[0058] Furthermore, the cerebral blood flow calculation device 1000 can acquire a boundary region of a region of interest (e.g., a cerebral blood vessel region) from a surface model (or a solid model) and acquire boundary information corresponding to the boundary region (e.g., coordinate information of the boundary region or node information of the boundary region). Specifically, the cerebral blood flow calculation device 1000 can acquire nodes corresponding to the boundary region of a region of interest (e.g., a cerebral blood vessel region) using any segmentation technique and acquire the boundary information. Consequently, the cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure can acquire three-dimensional initial conditions and / or boundary conditions associated with the boundary region.

[0059] In addition, the cerebral blood flow calculation device 1000 can calculate a predicted cerebral blood flow value for cerebral blood flow velocity or pressure from a surface model (or solid model) using Computational Fluid Dynamics (CFD) or Computational Aided Engineering (CAE). At this time, the cerebral blood flow calculation device 1000 can interpolate the calculated predicted cerebral blood flow value to three-dimensional shape data in the form of a point cloud to generate fourth data (Flow data (v, p) in FIG. 4) related to blood flow velocity and pressure.

[0060] Furthermore, the cerebral blood flow calculation device 1000 can input a first learning dataset including first data related to three-dimensional shape data in the form of a point cloud (Node info(x, y, z) in FIG. 4), second data related to boundary information (bci in FIG. 4), third data related to initial conditions and boundary conditions (bcv in FIG. 4), and fourth data related to a predicted cerebral blood flow value ((v, p) in FIG. 4)) into a first cerebral blood flow prediction model to obtain an output value output through the first cerebral blood flow prediction model. At this time, the cerebral blood flow calculation device 1000 can compare the output value with the predicted cerebral blood flow included in the fourth data and update the parameters included in the first cerebral blood flow prediction model so that the first cerebral blood flow prediction model outputs an output value that approximates the predicted cerebral blood flow value.

[0061] FIG. 5 is a diagram illustrating another aspect of training the first cerebral blood flow prediction model according to an embodiment of the present disclosure.

[0062] As described in Figure 4, the first cerebral blood flow prediction model can be trained based on a first learning dataset including first data related to three-dimensional shape data for the cerebral vascular region in the form of a point cloud (Node Info(x, y, z) in Figure 5), second data related to boundary information of the cerebral vascular region (Boundary info(bci) in Figure 5), third data related to initial condition information and boundary condition information (initial / boundary values(bcv) in Figure 5), and fourth data related to blood flow velocity and pressure (Flow data(v, p) in Figure 5).

[0063] Meanwhile, according to one aspect, the cerebral blood flow calculation device 1000 can calculate a predicted cerebral blood flow value using a 1D network model and interpolate the predicted cerebral blood flow value to 3D shape data in the form of a point cloud to obtain fourth data (Flow data (v, p) in FIG. 5) related to the velocity and pressure of blood flow. Specifically, the cerebral blood flow calculation device 1000 can generate a 1D network model representing the shape of cerebral blood vessels from coordinate information corresponding to a region of interest (e.g., a cerebral blood vessel region). Here, the 1D network model refers to a model in which points corresponding to coordinate information of a segmented region of interest (e.g., a cerebral blood vessel region) are connected in an arbitrary manner, and the 1D network model can be composed of node information and connectivity information indicating connections between the nodes.

[0064] Furthermore, the cerebral blood flow calculation device 1000 can acquire a boundary region of a region of interest (e.g., a cerebral blood vessel region) from the 1D network model and acquire boundary information corresponding to the boundary region (e.g., coordinate information of the boundary region). Specifically, the cerebral blood flow calculation device 1000 can acquire nodes corresponding to the boundary region of a region of interest (e.g., a cerebral blood vessel region) using an image segmentation technique and acquire the boundary information. Consequently, the cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure can acquire three-dimensional initial conditions and / or boundary conditions associated with the boundary region.

[0065] In addition, the cerebral blood flow calculation device 1000 can calculate a predicted cerebral blood flow value for cerebral blood flow velocity or pressure from a 1D network model using any appropriate mathematical technique. For example, the cerebral blood flow calculation device 1000 can be configured to calculate a predicted cerebral blood flow value for cerebral blood flow velocity or pressure through a governing equation, assuming that cerebral blood flow is Poiseuille flow. In this case, the cerebral blood flow calculation device 1000 can interpolate the calculated predicted cerebral blood flow value to 3D shape data in the form of a point cloud to generate fourth data (Flow data(v, p) in FIG. 5) related to blood flow velocity and pressure.

[0066] 4, the cerebral blood flow calculation device 1000 can input a first learning dataset including first data related to 3D shape data in the form of a point cloud, second data related to boundary information, third data related to initial conditions and boundary conditions, and fourth data related to a predicted cerebral blood flow value into a first cerebral blood flow prediction model and obtain an output value output through the first cerebral blood flow prediction model. At this time, the cerebral blood flow calculation device 1000 can compare the output value with the predicted cerebral blood flow included in the fourth data and update parameters included in the first cerebral blood flow prediction model so that the first cerebral blood flow prediction model outputs an output value that approximates the predicted cerebral blood flow value.

[0067] FIG. 6 is a diagram illustrating a phase of training a first cerebral blood flow prediction model according to an embodiment of the present disclosure and a phase of verifying the performance of the first cerebral blood flow prediction model.

[0068] The cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure may perform an operation of preprocessing first data related to 3D shape data (e.g., Node Info(x, y, z) of a region of interest) included in a first training dataset, second data related to boundary information (bci) of the region of interest, third data related to initial or boundary conditions (bcv), and / or fourth data related to predicted cerebral blood flow values (v, p). For example, the cerebral blood flow calculation device 1000 may perform normalization on each of the first data, second data, third data, and / or fourth data included in the first training dataset.

[0069] Furthermore, the cerebral blood flow calculation device 1000 according to an embodiment of the present disclosure may perform an operation of dividing a preprocessed first learning dataset into a training dataset and a validation dataset. For example, the cerebral blood flow calculation device 1000 may randomly divide the preprocessed first learning dataset into a training dataset and a validation dataset. At this time, the cerebral blood flow calculation device 1000 may train a first cerebral blood flow prediction model using the training dataset through a machine learning algorithm (ML algorithm). Furthermore, the cerebral blood flow calculation device 1000 may test or verify the performance of the learned first cerebral blood flow prediction model using the validation dataset. For example, the cerebral blood flow calculation device 1000 may calculate the performance of the learned first cerebral blood flow prediction model, compare the calculated performance with a predetermined target performance, and perform learning of the first cerebral blood flow prediction model until the calculated performance exceeds the predetermined target performance, thereby obtaining an optimized first cerebral blood flow prediction model.

[0070] Hereinafter, various operations of the cerebral blood flow calculation device 100 according to another embodiment of the present disclosure, a method for calculating cerebral blood flow data, and a method for training a cerebral blood flow prediction model will be described in detail with reference to Figures 7 to 10. Figure 7 is a diagram illustrating the manner in which cerebral blood flow data is calculated using a second cerebral blood flow prediction model according to another embodiment of the present disclosure. According to this embodiment, cerebral blood flow data can be calculated from a 1D network model through the second cerebral blood flow prediction model that has completed training.

[0071] A cerebral blood flow calculation device 1000 according to another embodiment of the present disclosure can acquire 3D shape data in the form of a 1D network model from an original medical image. For example, the cerebral blood flow calculation device 1000 can acquire a region of interest (e.g., a cerebral blood vessel region) from the original medical image using an image segmentation technique and acquire coordinate information corresponding to the region of interest in the form of a point cloud. The cerebral blood flow calculation device 1000 can then connect points included in the coordinate information corresponding to the cerebral blood vessel region to generate a 1D network model composed of node information corresponding to the region of interest and connectivity information indicating the connectivity between the nodes. The cerebral blood flow calculation device 1000 can then acquire node information corresponding to the region of interest (e.g., a cerebral blood vessel region) and connectivity information indicating the connectivity between the nodes from the generated 1D network model.

[0072] Furthermore, the cerebral blood flow calculation device 1000 according to another embodiment of the present disclosure can obtain a boundary region of a region of interest and obtain boundary information corresponding to the boundary region, as described with reference to Fig. 3. The cerebral blood flow calculation device 1000 can also obtain three-dimensional initial conditions and / or boundary conditions associated with the boundary region.

[0073] A cerebral blood flow calculation device 1000 according to another embodiment of the present disclosure can calculate cerebral blood flow data using a trained second cerebral blood flow prediction model. Specifically, the cerebral blood flow calculation device 1000 can input 3D shape data in the form of a 1D network model, boundary information, and / or 3D initial condition information and boundary condition information to the trained second cerebral blood flow prediction model, and obtain cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the trained second cerebral blood flow prediction model. As described below, the second cerebral blood flow prediction model is trained to output cerebral blood flow prediction values related to cerebral blood flow velocity and pressure based on the node information and connection information, boundary information, and / or three-dimensional initial condition information and boundary condition information that constitute the 1D network model.Therefore, once training is complete, the second cerebral blood flow prediction model can output cerebral blood flow data related to cerebral blood flow velocity and pressure from the three-dimensional shape data for the cerebral vascular region that is related to the node information and connection information that constitute the 1D network model, the boundary information of the cerebral vascular region, and / or the initial condition information and / or boundary condition information.

[0074] Hereinafter, a method for training a second cerebral blood flow prediction model and a method for generating a training dataset according to another embodiment of the present disclosure will be described in more detail with reference to Figures 8 and 9. According to this embodiment, the training dataset for training the second cerebral blood flow prediction model includes data related to a cerebral blood flow predicted value.

[0075] According to one aspect, the predicted cerebral blood flow value can be calculated using a surface model and / or a solid model, which will be described in more detail with reference to FIG.

[0076] In another aspect, the predicted cerebral blood flow value can be calculated using a 1D network model, which will be described in more detail with reference to FIG.

[0077] FIG. 8 is a diagram illustrating one aspect of training a second cerebral blood flow prediction model according to another embodiment of the present disclosure.

[0078] The second cerebral blood flow prediction model may be trained based on a second learning dataset including: first data (Node / Connectivity Info(x, y, z, c) in FIG. 8)) consisting of node information and connectivity information constituting the 1D network model, which is 3D shape data in the form of a 1D network model for the cerebral blood vessel region; second data (Boundary info (bci) in FIG. 8) related to boundary information of the cerebral blood vessel region; third data (initial / boundary values (bcv) in FIG. 8) related to initial condition information and boundary condition information; and fourth data (Flow data (v, p) in FIG. 8) related to blood flow velocity and pressure. Specifically, the cerebral blood flow calculation device 1000 may generate the first data by acquiring node information and connectivity information of the 1D network model from the original medical image (raw image), generate the second data by acquiring boundary information of the cerebral blood vessel region, and generate the third data by acquiring the three-dimensional initial and boundary conditions, as described above.

[0079] Furthermore, as described above with reference to FIG. 4, the cerebral blood flow calculation device 1000 can calculate a predicted cerebral blood flow value using a surface model and / or a solid model, and can interpolate the predicted cerebral blood flow value to three-dimensional shape data in the form of a 1D network model to obtain fourth data related to the velocity and pressure of blood flow (Flow data (v, p) in FIG. 8). Specifically, the cerebral blood flow calculation device 1000 can generate a surface model or solid model representing the shape of cerebral blood vessels from coordinate information corresponding to a region of interest (e.g., a cerebral blood vessel region) using a computer vision segmentation technique. In addition, the cerebral blood flow calculation device 1000 can obtain a boundary region of the region of interest (e.g., a cerebral blood vessel region) from the surface model (or solid model) and obtain boundary information corresponding to the boundary region (e.g., coordinate information of the boundary region). Specifically, the cerebral blood flow calculation device 1000 can acquire boundary information by acquiring nodes corresponding to the boundary region of a region of interest (e.g., a cerebral blood vessel region) using any segmentation technique. Consequently, the cerebral blood flow calculation device 1000 can acquire three-dimensional initial conditions and / or boundary conditions related to the boundary region. The cerebral blood flow calculation device 1000 can also calculate a predicted cerebral blood flow value for cerebral blood velocity or pressure from a surface model (or solid model) using Computational Fluid Dynamics (CFD) or Computational Aided Engineering (CAE). The cerebral blood flow calculation device 1000 can then interpolate the calculated predicted cerebral blood flow value into three-dimensional shape data in the form of a 1D network model to generate fourth data (Flow data(v, p) in FIG. 8) related to blood flow velocity and pressure.

[0080] In this case, the cerebral blood flow calculation device 1000 can input a second learning dataset including first data consisting of node information and connectivity information constituting a 1D network model, second data related to boundary information, third data related to initial conditions and boundary conditions, and fourth data related to a predicted cerebral blood flow value into the second cerebral blood flow prediction model, and obtain an output value output through the second cerebral blood flow prediction model.In this case, the cerebral blood flow calculation device 1000 can compare the output value with the predicted cerebral blood flow included in the fourth data, and update parameters included in the second cerebral blood flow prediction model so that the second cerebral blood flow prediction model outputs an output value that approximates the predicted cerebral blood flow value.

[0081] FIG. 9 is a diagram illustrating another aspect of training a second cerebral blood flow prediction model according to another embodiment of the present disclosure.

[0082] Similar to the description in FIG. 8, the second cerebral blood flow prediction model can be trained based on a second learning dataset including first data related to three-dimensional shape data for the cerebral vascular region in the form of a 1D network model (Node / Connectivity Info(x, y, z, c) in FIG. 9), second data related to boundary information of the cerebral vascular region (Boundary info(bci) in FIG. 9), third data related to initial condition information and boundary condition information (initial / boundary values(bcv) in FIG. 9), and fourth data related to blood flow velocity and pressure (Flow data(v, p) in FIG. 9).

[0083] Meanwhile, according to one aspect, the cerebral blood flow calculation device 1000 can calculate a predicted cerebral blood flow value using a 1D network model to obtain fourth data related to blood flow velocity and pressure (Flow data (v, p) in FIG. 9). As an example, the cerebral blood flow calculation device 1000 can calculate a predicted cerebral blood flow value for cerebral blood flow velocity or pressure from the 1D network model using any appropriate mathematical technique. For example, the cerebral blood flow calculation device 1000 can be configured to assume cerebral blood flow as Poiseuille flow and calculate a predicted cerebral blood flow value for cerebral blood flow velocity or pressure through a governing equation.

[0084] 8, the cerebral blood flow calculation device 1000 can input a second learning data set including first data consisting of node information and connectivity information constituting a 1D network model, second data related to boundary information, third data related to initial conditions and boundary conditions, and fourth data related to a predicted cerebral blood flow value into a second cerebral blood flow prediction model and obtain an output value output through the second cerebral blood flow prediction model. At this time, the cerebral blood flow calculation device 1000 can compare the output value with the predicted cerebral blood flow included in the fourth data and update the parameters included in the second cerebral blood flow prediction model so that the second cerebral blood flow prediction model outputs an output value that approximates the predicted cerebral blood flow value.

[0085] FIG. 10 is a diagram illustrating a phase of training a second cerebral blood flow prediction model according to another embodiment of the present disclosure and a phase of verifying the performance of the second cerebral blood flow prediction model.

[0086] A cerebral blood flow calculation device 1000 according to another embodiment of the present disclosure may perform an operation of preprocessing first data related to 3D shape data (e.g., Node / Connectivity Info (x, y, z, c) of a region of interest), second data related to boundary information (bci) of the region of interest, third data related to initial or boundary conditions (bcv), and / or fourth data related to predicted cerebral blood flow values (v, p) included in a second training dataset. For example, the cerebral blood flow calculation device 1000 may perform normalization on each of the first data, second data, third data, and / or fourth data included in the second training dataset.

[0087] Furthermore, the cerebral blood flow calculation device 1000 according to another embodiment of the present disclosure may perform an operation of dividing a preprocessed second learning dataset into a training dataset and a validation dataset. For example, the cerebral blood flow calculation device 1000 may randomly divide the preprocessed second learning dataset into a training dataset and a validation dataset. The cerebral blood flow calculation device 1000 may then train a second cerebral blood flow prediction model using the training dataset through a machine learning algorithm (ML algorithm). The cerebral blood flow calculation device 1000 may then test or verify the performance of the learned second cerebral blood flow prediction model using the validation dataset. For example, the cerebral blood flow calculation device 1000 may calculate the performance of the learned second cerebral blood flow prediction model, compare the calculated performance with a predetermined target performance, and perform training of the second cerebral blood flow prediction model until the calculated performance exceeds the predetermined target performance, thereby obtaining an optimized second cerebral blood flow prediction model.

[0088] Hereinafter, a method for calculating cerebral blood flow data according to an embodiment of the present disclosure will be described in more detail with reference to Figure 11. In describing the method for calculating cerebral blood flow data, the details described with reference to Figures 3 to 6 may be omitted. However, this is merely for the convenience of explanation, and the omissions should not be interpreted as restrictive.

[0089] A method for calculating cerebral blood flow data according to one embodiment of the present disclosure includes a step of acquiring a first cerebral blood flow prediction model that has completed learning (S1100), a step of acquiring an original medical image (S1200), a step of acquiring shape data composed of point cloud-type data corresponding to the cerebral vascular region (S1300), a step of acquiring a boundary region of the cerebral vascular region and acquiring boundary information corresponding to the boundary region (S1400), a step of acquiring initial condition information or boundary condition information (S1500), and a step of acquiring cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure through the first cerebral blood flow prediction model (S1600).

[0090] In the step of acquiring the trained first cerebral blood flow prediction model (S1100), the cerebral blood flow calculation device 1000 can acquire execution data for executing the trained first cerebral blood flow prediction model. Here, the execution data refers to any data necessary for properly executing the first cerebral blood flow prediction model, including structural information, calculation information, hierarchical information, and / or parameter information. Meanwhile, the first cerebral blood flow calculation device 1000 can be trained based on a first training dataset, as described above, including first data related to shape data corresponding to a cerebral vascular region in the form of a point cloud, second data related to boundary information of the cerebral vascular region, third data related to initial condition information or boundary condition information, and fourth data related to a predicted cerebral blood flow value for cerebral blood flow velocity or pressure. Specifically, the first cerebral blood flow prediction model can be configured to receive the first training dataset and output an output value from the first training dataset, and can be trained by updating parameters included in the first cerebral blood flow prediction model to output an output value that approximates the predicted cerebral blood flow value included in the fourth data.

[0091] According to one aspect, the fourth data may be generated by interpolating a blood flow velocity or a blood flow pressure calculated from shape data in the form of a point cloud corresponding to the cerebral vascular region to the shape data in the form of a point cloud. Specifically, the blood flow velocity or the blood flow pressure may be calculated using Computational Fluid Dynamics (CFD) based on a surface model generated from the shape data in the form of a point cloud.

[0092] In another aspect, the blood flow velocity or blood flow pressure associated with the fourth data can be calculated using any mathematical technique (e.g., a Governing Equation) from a 1D network model generated from shape data in the form of a point cloud.

[0093] In the step of acquiring the original medical image (S1200), the cerebral blood flow calculation device 1000 can acquire the original medical image through the transmitting / receiving unit 1100.

[0094] In the step of acquiring shape data composed of point cloud data corresponding to the cerebral blood vessel region (S1300), the cerebral blood flow calculation device 1000 can acquire a region of interest (e.g., the cerebral blood vessel region) using an image segmentation technique and acquire coordinate information corresponding to the region of interest. Furthermore, the cerebral blood flow calculation device 1000 can acquire 3D shape data composed of point cloud data (i.e., points model) based on the coordinate information of the region of interest.

[0095] In the step of acquiring a boundary region of the cerebral blood vessel region and acquiring boundary information corresponding to the boundary region (S1400), the cerebral blood flow calculation device 1000 can acquire the boundary region of the region of interest (e.g., a region corresponding to the cerebral blood vessel) and acquire boundary information corresponding to the boundary region (e.g., coordinate information of the boundary region). Specifically, the cerebral blood flow calculation device 1000 can acquire nodes corresponding to the boundary region of the region of interest (e.g., the cerebral blood vessel region) by utilizing an image segmentation technique and acquire the boundary information.

[0096] In the step of acquiring initial condition information or boundary condition information (S1500), the cerebral blood flow calculation device 1000 can acquire three-dimensional initial conditions and / or boundary conditions related to the boundary region.

[0097] In the step of acquiring cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure through the first cerebral blood flow prediction model (S1600), the cerebral blood flow calculation device 1000 inputs shape data, boundary information, and initial condition information or boundary condition information to the first cerebral blood flow prediction model, and can acquire cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the first cerebral blood flow prediction model. Since the first cerebral blood flow prediction model is trained to output cerebral blood flow predicted values related to cerebral blood flow velocity and pressure based on point information, boundary information, and / or three-dimensional initial condition information and boundary condition information, the first cerebral blood flow prediction model after completion of training can output cerebral blood flow data related to cerebral blood flow velocity and pressure from three-dimensional shape data of the cerebral blood vessel region in the form of a point cloud, boundary information of the cerebral blood vessel region, and / or initial condition information and / or boundary condition information.

[0098] Hereinafter, a method for calculating cerebral blood flow data according to another embodiment of the present disclosure will be described in more detail with reference to Fig. 12. In describing the method for calculating cerebral blood flow data, the details described with reference to Figs. 7 to 10 may be omitted. However, this is merely for the convenience of explanation, and the omissions should not be interpreted as restrictive.

[0099] A method for calculating cerebral blood flow data according to another embodiment of the present disclosure includes a step of acquiring a second cerebral blood flow prediction model that has completed learning (S2100), a step of acquiring an original medical image (S2200), a step of acquiring shape data composed of node information and connection information that constitute a 1D network model corresponding to the cerebral vascular region (S2300), a step of acquiring a boundary region of the cerebral vascular region and acquiring boundary information corresponding to the boundary region (S2400), a step of acquiring initial condition information or boundary condition information (S2500), and a step of acquiring cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure through the second cerebral blood flow prediction model (S2600).

[0100] In the step of acquiring the trained second cerebral blood flow prediction model (S2100), the cerebral blood flow calculation device 1000 can acquire execution data for executing the trained second cerebral blood flow prediction model. Here, the execution data refers to any data necessary for properly executing the second cerebral blood flow prediction model, including calculation information, hierarchical information, and / or parameter information. Meanwhile, the second cerebral blood flow calculation device 1000 can be trained based on a second training dataset, which includes first data consisting of node information and connectivity information constituting a 1D network model, second data related to boundary information of the cerebral vascular region, third data related to initial condition information or boundary condition information, and fourth data related to predicted cerebral blood flow values for cerebral blood flow velocity or pressure, as described above. Specifically, the second cerebral blood flow prediction model is configured to receive the second training dataset and output an output value from the second training dataset. The parameters included in the second cerebral blood flow prediction model can be updated and trained to output an output value that approximates the predicted cerebral blood flow value included in the fourth data.

[0101] According to one aspect, the fourth data may be generated by interpolating a blood flow velocity or a blood flow pressure calculated from shape data in the form of a point cloud corresponding to the cerebral vascular region to shape data in the form of a 1D network model. Specifically, the blood flow velocity or the blood flow pressure may be calculated using Computational Fluid Dynamics (CFD) based on a surface model generated from the shape data in the form of a point cloud.

[0102] In another aspect, the blood flow velocity or blood flow pressure associated with the fourth data can be calculated using any mathematical technique (e.g., a Governing Equation) from shape data composed of node information and connection information in the form of a 1D network model.

[0103] In the step of acquiring the original medical image (S2200), the cerebral blood flow calculation device 1000 can acquire the original medical image through the transmitting / receiving unit 1100.

[0104] In the step S2300 of acquiring shape data composed of node information and connectivity information constituting a 1D network model corresponding to the cerebral blood vessel region, the cerebral blood flow calculation device 1000 may acquire a region of interest (e.g., the cerebral blood vessel region) from the original medical image using an image segmentation technique and acquire coordinate information corresponding to the region of interest in the form of a point cloud. At this time, the cerebral blood flow calculation device 1000 may connect points included in the coordinate information corresponding to the cerebral blood vessel region to generate a 1D network model composed of node information corresponding to the region of interest and connectivity information indicating the connectivity between the nodes.

[0105] In the step of acquiring a boundary region of the cerebral blood vessel region and acquiring boundary information corresponding to the boundary region (S2400), the cerebral blood flow calculation device 1000 can acquire the boundary region of the region of interest (e.g., a region corresponding to the cerebral blood vessel) and acquire boundary information corresponding to the boundary region (e.g., coordinate information of the boundary region). Specifically, the cerebral blood flow calculation device 1000 can acquire nodes corresponding to the boundary region of the region of interest (e.g., the cerebral blood vessel region) by utilizing an image segmentation technique and acquire the boundary information.

[0106] In the step of acquiring initial condition information or boundary condition information (S1500), the cerebral blood flow calculation device 1000 can acquire three-dimensional initial conditions and / or boundary conditions related to the boundary region.

[0107] In the step of acquiring cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure through the second cerebral blood flow prediction model (S2600), the cerebral blood flow calculation device 1000 inputs shape data composed of node information and connection information, boundary information, and initial condition information or boundary condition information to the second cerebral blood flow prediction model, and can acquire cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the second cerebral blood flow prediction model. Since the second cerebral blood flow prediction model is trained to output cerebral blood flow predicted values related to cerebral blood flow velocity and pressure based on the node information, connection information, boundary information, and / or three-dimensional initial condition information and boundary condition information constituting the 1D network, the second cerebral blood flow prediction model after completion of training can output cerebral blood flow data related to cerebral blood flow velocity and pressure from the shape data of the cerebral blood vessel region composed of the node information and connection information of the 1D network model, the boundary information of the cerebral blood vessel region, and / or the initial condition information and / or boundary condition information.

[0108] According to a method for calculating cerebral blood flow data, a method for learning a cerebral blood flow prediction model, and a cerebral blood flow calculation device according to one embodiment of the present disclosure, it is possible to learn a cerebral blood flow prediction model that can calculate cerebral blood flow information from a point model or a 1D network model, rather than a surface model or a solid model.

[0109] According to a method for calculating cerebral blood flow data, a method for learning a cerebral blood flow prediction model, and a cerebral blood flow calculation device according to one embodiment of the present disclosure, a trained cerebral blood flow prediction model can be used to automatically calculate cerebral blood flow information from a point model or a 1D network model, rather than a surface model or a solid model.

[0110] Furthermore, according to the method for calculating cerebral blood flow data, the method for learning a cerebral blood flow prediction model, and the cerebral blood flow calculation device according to one embodiment of the present disclosure, the various operations of the cerebral blood flow calculation device 1000 described above may be stored in the memory 1200 of the cerebral blood flow calculation device 1000, and the processor 1300 of the cerebral blood flow calculation device 1000 may be provided to perform the operations stored in the memory 1200.

[0111] The features, structures, effects, etc. described in the above embodiments are included in at least one embodiment of the present invention and are not necessarily limited to only one embodiment. Furthermore, the features, structures, effects, etc. exemplified in each embodiment can be combined or modified in other embodiments by a person skilled in the art to which the embodiments belong. Therefore, content related to such combinations and modifications should be interpreted as being included within the scope of the present invention.

[0112] Furthermore, although the above description focuses on the embodiments, these are merely examples and do not limit the present invention. Those skilled in the art will recognize that various modifications and applications not exemplified above are possible within the scope of the essential characteristics of the present invention. In other words, each component specifically illustrated in the embodiments can be modified and implemented. Differences related to such modifications and applications should be construed as being included within the scope of the present invention as defined by the appended claims. [Industrial Applicability]

[0113] The method for calculating cerebral blood flow data, the method for learning a cerebral blood flow prediction model, and the cerebral blood flow calculation device as described above can be applied to the medical service field that provides medical services.

Claims

1. A method for calculating cerebral blood flow data using a cerebral blood flow calculation device, comprising: A step of obtaining a trained cerebral blood flow prediction model; obtaining original medical images; acquiring shape data corresponding to a cerebrovascular region from the original medical image, the shape data being configured as point cloud data; obtaining a boundary area of the cerebrovascular region and obtaining boundary information corresponding to the boundary area; obtaining initial or boundary condition information; and A method for calculating cerebral blood flow data, comprising: inputting the shape data, the boundary information, and initial condition information or boundary condition information into the cerebral blood flow prediction model, and obtaining cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the cerebral blood flow prediction model.

2. The cerebral blood flow prediction model is 2. The method for calculating cerebral blood flow data according to claim 1, wherein the method is trained based on a learning dataset including first data related to shape data corresponding to a cerebral vascular region in the form of a point cloud, second data related to boundary information of the cerebral vascular region, third data related to initial condition information or boundary condition information, and fourth data related to predicted values of cerebral blood flow for cerebral blood flow velocity or pressure.

3. The cerebral blood flow prediction model is configured to receive the training data set and output values from the training data set; The method for calculating cerebral blood flow data according to claim 2, wherein parameters included in the cerebral blood flow prediction model are updated and trained so as to output the output value so as to approximate the cerebral blood flow predicted value.

4. The fourth data is The blood flow velocity or blood flow pressure calculated from the shape data corresponding to the cerebral blood vessel region is interpolated to the shape data in the form of a point cloud, 4. The method for calculating cerebral blood flow data according to claim 3, wherein the calculated blood flow velocity or blood flow pressure is calculated using Computational Fluid Dynamics (CFD) based on a surface model generated from the shape data.

5. The fourth data is The blood flow velocity or blood flow pressure calculated from the shape data corresponding to the cerebral blood vessel region is interpolated to the shape data in the form of a point cloud, The method for calculating cerebral blood flow data according to claim 3 , wherein the calculated blood flow velocity or blood flow pressure is calculated from a 1D network model generated from the shape data.

6. A method for calculating cerebral blood flow data using a cerebral blood flow calculation device, comprising: A step of obtaining a trained cerebral blood flow prediction model; obtaining original medical images; obtaining first shape data corresponding to a cerebrovascular region from the original medical image, the first shape data being composed of node information and connection information constituting a 1D network model; obtaining a boundary area of the cerebrovascular region and obtaining boundary information corresponding to the boundary area; obtaining initial or boundary condition information; and A method for calculating cerebral blood flow data, comprising: inputting the first shape data, the boundary information, and initial condition information or boundary condition information into the cerebral blood flow prediction model, and obtaining cerebral blood flow data related to cerebral blood flow velocity or cerebral blood flow pressure output through the cerebral blood flow prediction model.

7. The cerebral blood flow prediction model is 7. The method for calculating cerebral blood flow data according to claim 6, wherein the method is trained based on a learning dataset including: first data consisting of node information and connection information constituting a 1D network model; second data related to boundary information of the cerebral vascular region; third data related to initial condition information or boundary condition information; and fourth data related to predicted values of cerebral blood flow for cerebral blood flow velocity or pressure.

8. The cerebral blood flow prediction model is configured to receive the training data set and output values from the training data set; The method for calculating cerebral blood flow data according to claim 7, wherein parameters included in the cerebral blood flow prediction model are updated and trained so as to output the output value so as to approximate the cerebral blood flow predicted value.

9. The fourth data is The blood flow velocity or blood flow pressure calculated from the second shape data in the form of a point cloud corresponding to the cerebral blood vessel region is interpolated to the first shape data in the form of a 1D network model, 9. The method for calculating cerebral blood flow data according to claim 8, wherein the calculated blood flow velocity or blood flow pressure is calculated using Computational Fluid Dynamics (CFD) based on a surface model generated from the second shape data.

10. The fourth data is The method for calculating cerebral blood flow data according to claim 8 , wherein a predicted cerebral blood flow value related to a blood flow velocity or a blood flow pressure is calculated from the first shape data to generate the cerebral blood flow data.

11. A computer-readable recording medium having recorded thereon a program for causing a computer to execute the method according to any one of claims 1 to 10.

Images