Method and device for training segmentation model for segmenting blood vessel segment, and storage medium

Abstract

The invention discloses a method and equipment for training a segmentation model for segmenting a blood vessel segment, and a storage medium. The method comprises the following steps: acquiring multi-modal image data, including digital subtraction angiography data, CT angiography data and magnetic resonance angiography data; performing blood vessel segmentation based on the multi-modal image data and extracting a blood vessel surface to obtain blood vessel surface data; labeling positioning points and labeling blood vessel segments on the blood vessel surface data, and inputting the labeled blood vessel surface data into a positioning model for training; distance features on the blood vessel surface data are calculated according to the positioning point information; and inputting the blood vessel surface data, the distance features and the blood vessel segmentation labels into the segmentation model for blood vessel segmentation to obtain blood vessel segments, so as to train a segmentation model for segmenting the blood vessel segments. By using the scheme of the invention, the segmentation precision of the segmentation model can be improved so as to obtain a more accurate blood vessel segment segmentation result.

Description

Technical Field

[0001] The present application generally relates to the technical field of blood vessel segmentation, and more specifically, to a method for training a segmentation model for segmenting blood vessel segments, a method, a device and a computer-readable storage medium for segmenting blood vessel segments. Background Art

[0002] According to the needs of research and clinical treatment, it is usually necessary to segment vascular images into multiple vascular segments based on image processing and segmentation methods of medical image analysis to obtain vascular network information for disease diagnosis and prognosis. For example, by segmenting the intracranial arteries, it can help doctors accurately diagnose the location of arteriosclerosis and cerebrovascular disease, and make a pre-judgment of the prognosis. At present, for vascular segmentation, the vascular segments are directly labeled in the image data, and the labeled images are input into the segmentation model for vascular segmentation, and the segmentation accuracy is low.

[0003] In view of this, there is an urgent need to provide a solution for training a segmentation model for segmenting blood vessel segments, so as to improve the segmentation accuracy of the segmentation model and obtain a more accurate blood vessel segmentation result. Summary of the invention

[0004] In order to at least solve one or more of the technical problems mentioned above, the present application proposes a solution for training a segmentation model for segmenting blood vessel segments in multiple aspects.

[0005] In a first aspect, the present application provides a method for training a segmentation model for segmenting blood vessel segments, comprising: acquiring multimodal image data, wherein the multimodal image data comprises digital subtraction angiography data, CT vascular imaging data and magnetic resonance vascular imaging data; performing blood vessel segmentation and extracting blood vessel surfaces based on the multimodal image data to obtain blood vessel surface data; marking positioning points and blood vessel segments on the blood vessel surface data, and inputting the marked blood vessel surface data into the positioning model for training to output positioning point information; calculating distance features on the blood vessel surface data according to the positioning point information; and inputting the blood vessel surface data, the distance features and the blood vessel segmentation annotations into the segmentation model for blood vessel segmentation to obtain blood vessel segments, so as to train the segmentation model for segmenting blood vessel segments.

[0006] In some embodiments, the positioning points at least include connection points at key blood vessels in the blood vessel surface data; the positioning point information includes coordinate information of the positioning points or index information of the positioning points.

[0007] In some other embodiments, calculating the distance feature on the blood vessel surface data based on the positioning point information includes: calculating the projection values ​​of the vector formed between the target point and the target positioning point on the blood vessel surface data on the vectors formed between the target positioning point and the remaining positioning points based on the positioning point information, so as to calculate the distance feature on the blood vessel surface data.

[0008] In some other embodiments, calculating the distance feature on the blood vessel surface data based on the positioning point information also includes: calculating the projection value of the vector formed between the target point on the blood vessel surface data and the target positioning point on the target vector based on the positioning point information, so as to calculate the distance feature on the blood vessel surface data; wherein the target vector is perpendicular to the vectors formed between the target positioning point and the remaining positioning points.

[0009] In some other embodiments, it also includes: calculating the geodesic distance on the blood vessel surface data based on the positioning point information; adding the geodesic distance to the distance feature to obtain a new distance feature; and inputting the blood vessel surface data, the new distance feature and the blood vessel segmentation annotation into the segmentation model to perform blood vessel segmentation to obtain blood vessel segments, so as to train the segmentation model used to segment blood vessel segments.

[0010] In some other embodiments, calculating the geodesic distance on the blood vessel surface data based on the positioning point information includes: calculating the original geodesic distance from the target point on the blood vessel surface data to each positioning point; and / or calculating the shortest geodesic distance from the target point on the blood vessel surface data to each positioning point.

[0011] In some further embodiments, the positioning model and the segmentation model both include a point cloud processing model or an image processing model.

[0012] In a second aspect, the present application provides a method for segmenting blood vessel segments, comprising: acquiring image data to be segmented and extracting blood vessel surface data based on the image data to be segmented; and inputting the blood vessel surface data into a segmentation model trained by one or more embodiments of the first aspect to perform blood vessel segmentation to obtain a blood vessel segmentation result.

[0013] In a third aspect, the present application provides a device, comprising: a processor; and a memory, on which computer instructions for training a segmentation model for segmenting blood vessel segments are stored, and when the computer instructions are executed by the processor, multiple embodiments of the first aspect are implemented; or computer instructions for segmenting blood vessel segments are stored, and when the computer instructions are executed by the processor, one embodiment of the second aspect is implemented.

[0014] In a fourth aspect, the present application provides a computer-readable storage medium storing computer program instructions for training a segmentation model for segmenting a blood vessel segment, which, when executed by one or more processors, implements multiple embodiments of the first aspect; or storing computer instructions for segmenting a blood vessel segment, which, when executed by a processor, implements one embodiment of the second aspect.

[0015] Through the scheme for training the segmentation model for segmenting blood vessel segments provided above, the embodiment of the present application obtains blood vessel surface data by performing blood vessel segmentation on multimodal image data and extracting the blood vessel surface. First, the positioning model is trained based on the positioning point annotations on the blood vessel surface data, and then the distance feature is calculated according to the positioning point information. Then, the segmentation model is trained according to the blood vessel surface data, the distance feature and the blood vessel segmentation annotation to output the blood vessel segmentation result. Based on this, the embodiment of the present application quantifies the blood vessel feature information by adding the distance feature, thereby improving the recognition accuracy of the blood vessel segmentation and improving the segmentation accuracy of the blood vessel segmentation model. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present application will become easy to understand. In the accompanying drawings, several embodiments of the present application are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0017] Figure 1 is an exemplary flowchart illustrating a method for training a segmentation model for segmenting a blood vessel segment according to an embodiment of the present application;

[0018] Figure 2 is an exemplary flowchart showing the overall process of training a segmentation model for segmenting a blood vessel segment according to an embodiment of the present application;

[0019] Figure 3 is another exemplary flowchart showing the overall training of a segmentation model for segmenting a blood vessel segment according to an embodiment of the present application;

[0020] Figure 4 is an exemplary schematic diagram showing blood vessel surface data according to an embodiment of the present application;

[0021] Figure 5 is an exemplary schematic diagram showing a distance feature map obtained by calculating projection values ​​according to an embodiment of the present application;

[0022] Figure 6 is an exemplary schematic diagram showing a distance feature map obtained by calculating the geodesic distance according to an embodiment of the present application;

[0023] Figure 7 is an exemplary flowchart showing a method for segmenting a blood vessel segment according to an embodiment of the present application;

[0024] Figure 8 is a block diagram showing an exemplary structure of a device according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0026] It should be understood that the terms "include" and "comprising" used in the specification and claims of the present application indicate the presence of described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0027] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in this application specification and claims, unless the context clearly indicates otherwise, the singular forms of "a", "an" and "the" are intended to include plural forms. It should also be further understood that the term "and / or" used in this application specification and claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0028] As used in this specification and claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0029] The specific implementation of the present application is described in detail below with reference to the accompanying drawings.

[0030] Figure 1 FIG. 1 is an exemplary flowchart of a method 100 for training a segmentation model for segmenting a blood vessel segment according to an embodiment of the present application. The segmentation model includes a positioning model and a segmentation model. Figure 1As shown in , the method 100 includes: step S101: acquiring multimodal image data, wherein the multimodal image data includes digital subtraction angiography data, CT angiography data and magnetic resonance angiography data; step S102: performing blood vessel segmentation and extracting blood vessel surface based on the multimodal image data to obtain blood vessel surface data; step S103: marking positioning points and blood vessel segments on the blood vessel surface data, and inputting the marked blood vessel surface data into a positioning model for training to output positioning point information; step S104: calculating distance features on the blood vessel surface data according to the positioning point information; and step S105: inputting the blood vessel surface data, distance features and blood vessel segmentation annotations into a segmentation model for blood vessel segmentation to obtain blood vessel segments, so as to train the segmentation model for segmenting blood vessel segments.

[0031] First, at step S101, multimodal image data is collected. The multimodal image data may include digital subtraction angiography data, CT angiography data, and magnetic resonance angiography data. Among them, digital subtraction angiography data is the gold standard for screening and diagnosis of various vascular diseases, while CT angiography data and magnetic resonance angiography data can be used for screening and follow-up of vascular diseases. The embodiment of the present application can greatly improve the performance of the segmentation model by using the three modal data for training the segmentation model.

[0032] Next, at step S102, blood vessel segmentation is performed based on the multimodal image data and the blood vessel surface is extracted to obtain blood vessel surface data. In some embodiments, image processing methods or deep learning methods may be used to perform blood vessel segmentation, wherein the aforementioned image processing methods may include but are not limited to image edge detection and image morphology methods. The aforementioned deep learning method may be based on an existing trained segmentation model to perform blood vessel segmentation to obtain a blood vessel segmentation result, and the present application does not impose any restrictions on this. In other embodiments, based on the aforementioned blood vessel segmentation results, blood vessel surface data may be extracted by using, for example, the Marching Cubes algorithm.

[0033] Further, at step S103, the vascular surface data is labeled with positioning points and vascular segments, and the labeled vascular surface data is input into the positioning model for training to output the positioning point information. In some implementation scenarios, the aforementioned dynamic positioning points and vascular segmentation annotations may be manually or automatically annotated. In some embodiments, the aforementioned positioning model may be, for example, a point cloud processing model (e.g., PointNet) or an image processing model (e.g., a Unet model, an FPN model, or a PSPNet model).

[0034] In some embodiments, the positioning points may include at least connection points at key blood vessels in the blood vessel surface data, and the positioning point information may include coordinate information of the positioning points or index information of the positioning points. That is, when marking the positioning points, they can be marked by coordinates (x, y, z) or by indexes (for example, P1, P2, P3, ..., Pn). In some implementation scenarios, when marking blood vessel segments, each blood vessel segment can be marked by, for example, 1, 2, 3, ..., n. If the pixel point is a background pixel, it is marked as 0. It can be understood that the aforementioned positioning points can be added or deleted according to the specific blood vessel segmentation task, and the present application does not limit the number of positioning points.

[0035] As an example, in the carotid system of the intracranial artery, the aforementioned positioning point may be the connection point of the left internal carotid artery (Internal Carotid Artery, "ICA") with the left middle cerebral artery (Middle Cerebral Artery, "MCA") and the left anterior cerebral artery (Anterior Cerebral Artery, "ACA"); the connection point of the left ICA with the left MCA and the left ACA; and the connection point of the right ICA with the right MCA and the right ACA. The aforementioned connections are all located at key positions of the Circle of Willis (also known as the Circle of the Cerebral Arteries), which is conducive to locating the vascular segments of the carotid system. In the vertebral-basilar artery system, the aforementioned positioning points may include, for example, the connection point of the basilar artery (Basilar Artery, "BA") with the left posterior communicating artery (Posterior Communicating Artery, "PCA") and the right PCA, which is located at a key position of the vertebral-basilar artery system, which is conducive to locating the vascular segments of the vertebral-basilar artery system.

[0036] Based on the annotated positioning points, the positioning point information can be output by inputting them into the positioning model for training. For example, if the annotated positioning point information is coordinate information, the input of the positioning model is the three coordinate values ​​(x, y, z) of each point. If the annotated positioning point information is the index of the connection point, the input of the positioning model is the category of the connection to which each point belongs. For example, if the connection point where the left ICA connects with the MCA and ACA is recorded as P1, the connection point where the right ICA connects with the MCA and ACA is recorded as P2, and the connection point where the BA connects with the left and right PCA is recorded as P3. In this scenario, the category of the connection to which each point belongs is 0, P1, P2, or P3.

[0037] Based on the positioning point information obtained above, at step S104, the distance feature on the blood vessel surface data is calculated according to the positioning point information. In some embodiments, the projection values ​​of the vector formed between the target point and the target positioning point on the blood vessel surface data on the vectors formed between the target positioning point and the remaining positioning points are calculated respectively according to the positioning point information to calculate the distance feature on the blood vessel surface data.

[0038] For example, in an exemplary scenario, assuming that the target point on the blood vessel surface data is denoted as X, and taking the target positioning point as the connection point P3 where BA and the left and right PCAs are connected as an example, the projection value of the vector P3X formed between the target point X and the target positioning point P3 on the vector P3P1 formed between the target positioning point P3 and the positioning point P1 can be calculated to obtain a feature map, and the projection value of the vector P3X formed between the target point X and the target positioning point P3 on the vector P3P2 formed between the target positioning point P3 and the positioning point P2 can be calculated to obtain a feature map to obtain a distance feature. Similarly, the projection values ​​of the vector P1X on the vector P1P2 and the vector P1P3, and the projection values ​​of the vector P2X on the vector P2P1 and the vector P2P3 can be calculated.

[0039] In other embodiments, the projection value of the vector formed between the target point and the target positioning point on the target vector may be calculated based on the positioning point information to calculate the distance feature on the blood vessel surface data; wherein the target vector is a vector perpendicular to the target positioning point and the remaining positioning points. For example, if the target vector perpendicular to the vector P3P1 and the vector P3P2 is recorded as P3P1×P3P2, the projection value of the vector P3X on the target vector P3P1×P3P2 may also be calculated as the distance feature.

[0040] Further, at step S105, the vascular surface data, distance features and vascular segmentation annotations are input into the segmentation model for vascular segmentation to obtain vascular segments, so as to train the segmentation model for segmenting vascular segments. Specifically, the vascular surface data and distance features can be used as input data, and the vascular segmentation annotations can be used as labels to train the segmentation model for segmenting vascular segments. In some implementation scenarios, the aforementioned segmentation model can also be, for example, a point cloud processing model (such as PointNet) or an image processing model (such as a Unet model, an FPN model or a PSPNet model). In the implementation scenario, each pixel is classified according to the vascular segmentation k=1,2,3,…,n to obtain a vascular segmentation result.

[0041] In combination with the above description, it can be seen that the embodiment of the present application first trains the positioning model based on the positioning point annotations on the blood vessel surface data to obtain the positioning point information, then calculates the distance feature according to the positioning point information, and then trains the segmentation model according to the blood vessel surface data, the distance feature and the blood vessel segmentation annotation to obtain the blood vessel segmentation result. Based on this, the blood vessel features are quantified by adding the distance feature to improve the accuracy of the segmentation model in performing blood vessel segmentation.

[0042] Figure 2 FIG. 1 is an exemplary flowchart showing the overall process of training a segmentation model for segmenting a blood vessel segment according to an embodiment of the present application. It should be understood that Figure 2 is the above Figure 1 A specific implementation of method 100, so the above Figure 1 The description also applies to Figure 2 .

[0043] like Figure 2 As shown in , at step S201, digital subtraction angiography data, CT angiography data and magnetic resonance angiography data are collected. Then, at step S202, blood vessel segmentation is performed based on the aforementioned three multimodal imaging data, and at step S203, the blood vessel surface is extracted to obtain blood vessel surface data. In some embodiments, blood vessel segmentation can be performed by, for example, edge detection, morphology or an existing segmentation model to obtain a blood vessel segmentation result. Blood vessel surface data is extracted by, for example, the MarchingCubes algorithm. Further, at step S204, positioning points and blood vessel segments are marked. For specific marking details, please refer to the above Figure 1 In step S205, the annotated blood vessel surface data is input into a positioning model for training, and positioning point information is output. In some embodiments, the positioning model may be, for example, a PointNet or PSPNet model.

[0044] Based on the above positioning point information, at step S206, multiple projection values ​​are calculated to obtain distance features. Specifically, the projection values ​​of the vector formed between the target point on the blood vessel surface data and the target positioning point on the vectors formed between the target positioning point and the remaining positioning points are calculated respectively; the projection value of the vector formed between the target point on the blood vessel surface data and the target positioning point on the target vector can also be calculated according to the positioning point information. Then, at step S207, the blood vessel surface data, distance features and blood vessel segmentation annotations are input into the segmentation model for training, so as to realize the training of the segmentation model for segmenting blood vessel segments.

[0045] In some embodiments, the geodesic distance on the blood vessel surface data can also be calculated according to the positioning point information, and the geodesic distance is added to the distance feature to obtain a new distance feature, and then the blood vessel surface data, the new distance feature and the blood vessel segmentation annotation are input into the segmentation model for blood vessel segmentation to obtain blood vessel segments, so as to train the segmentation model for segmenting blood vessel segments. That is, the distance feature is increased by calculating the geodesic distance, and the geodesic distance can better describe the connection relationship between different blood vessel segments, and can contain richer feature information, thereby further improving the accuracy of the segmentation model.

[0046] In some embodiments, the original geodesic distance from the target point on the blood vessel surface data to each positioning point and / or the shortest geodesic distance from the target point on the blood vessel surface data to each positioning point can be calculated. As an example, for a target point X on the blood vessel surface data, the original geodesic distance from the target point X to the positioning points P1, P2, and P3 can be calculated respectively. The shortest geodesic distance from the target point X to the positioning points P1, P2, or P3 can also be calculated.

[0047] Figure 3 is another exemplary flowchart showing the overall training of a segmentation model for segmenting a blood vessel segment according to an embodiment of the present application. It should be understood that Figure 3 is the above Figure 1 Another specific implementation of the method 100, so the above Figure 1 The description also applies to Figure 3 .

[0048] like Figure 3 As shown in , at step S301, digital subtraction angiography data, CT angiography data and magnetic resonance angiography data are collected. Then, at step S302, blood vessel segmentation is performed based on the aforementioned three multimodal imaging data, and at step S303, the blood vessel surface is extracted to obtain blood vessel surface data. In some embodiments, blood vessel segmentation can be performed by, for example, edge detection, morphology or an existing segmentation model to obtain a blood vessel segmentation result. The blood vessel surface data is extracted by, for example, the MarchingCubes algorithm. Further, at step S304, the positioning points and the blood vessel segments are marked. At step S305, the marked blood vessel surface data is input into the positioning model for training, and the positioning point information is output.

[0049] Based on the above positioning point information, at step S306, multiple projection values ​​are calculated to obtain a distance feature. Then, at step S307, the geodesic distance is calculated, and the geodesic distance is added to the distance feature to obtain a new distance feature. Further, at step S308, the blood vessel surface data, the new distance feature and the blood vessel segmentation annotation are input into the segmentation model for training, so as to realize the training of the segmentation model for segmenting the blood vessel segment.

[0050] Figure 4 FIG. 2 is an exemplary schematic diagram showing blood vessel surface data according to an embodiment of the present application. Figure 4 Figures (a) and (b) in the figure respectively show the top view and front view of the blood vessel surface data. The figure shows the connection point P1 where the left ICA connects with the MCA and ACA, the connection point P2 where the right ICA connects with the MCA and ACA, and the connection point P3 where the BA connects with the left and right PCAs. That is, the positioning points of the embodiment of the present application.

[0051] Figure 5 FIG. 1 is an exemplary schematic diagram showing a distance feature map obtained by calculating projection values ​​according to an embodiment of the present application. Figure 5 Figures (a), (b) and (c) in the figure respectively show the feature map corresponding to the projection value of vector P3X on vector P3P1, the feature map corresponding to the projection value of P3X on P3P2, and the feature map corresponding to the projection value of P3X on target vector P3P1*P3P2.

[0052] Figure 6 FIG. 1 is an exemplary schematic diagram showing a distance feature map obtained by calculating the geodesic distance according to an embodiment of the present application. Figure 6 Figures (a), (b), (c) and (d) in the figure respectively show the feature maps corresponding to the original geodesic distances from the target point X to the positioning points P1, P2, and P3, and the distance maps corresponding to the shortest geodesic distances from the target point X to the positioning points P1, P2 or P3.

[0053] Figure 7 FIG. 7 is an exemplary flowchart of a method 700 for segmenting a blood vessel segment according to an embodiment of the present application. Figure 7 As shown in , the method 700 includes: step S701: acquiring image data to be segmented and extracting blood vessel surface data based on the image data to be segmented; step S702: inputting the blood vessel surface data into the trained segmentation model to perform blood vessel segmentation to obtain a blood vessel segmentation result.

[0054] In step S701, the image data to be segmented is obtained and the blood vessel surface data is extracted based on the image data to be segmented. Then, in step S702, the blood vessel surface data is input into the trained segmentation model to perform blood vessel segmentation to obtain a blood vessel segmentation result. As mentioned above, the segmentation model includes a positioning model and a segmentation model. In the process of segmenting the blood vessel segment, the blood vessel surface data first obtains the positioning point through the positioning model, and then completes the blood vessel segmentation through the segmentation model to obtain the blood vessel segmentation result. The blood vessel segmentation result corresponds to the blood vessel segment category, such as 1, 2, 3, ..., n.

[0055] Based on the above, it can be seen that the embodiment of the present application obtains blood vessel surface data by performing blood vessel segmentation on multimodal image data and extracting blood vessel surface, trains a positioning model based on positioning points on the blood vessel surface data, and calculates distance features to quantify feature information to train a segmentation model, thereby improving the recognition accuracy of blood vessel segmentation and the segmentation accuracy of the blood vessel segmentation model. Furthermore, the embodiment of the present application obtains a more accurate blood vessel segmentation result by adding distance features to further improve the segmentation accuracy of the segmentation model.

[0056] Figure 8 800 according to an embodiment of the present application. It is understood that the device 800 may include the apparatus of the embodiment of the present application, and the device implementing the solution of the present application may be a single device (such as a computing device) or a multi-function device including various peripheral devices.

[0057] like Figure 8 As shown in , the device of the present application may also include a central processing unit or central processing unit ("CPU") 811, which may be a general-purpose CPU, a dedicated CPU, or other information processing and program execution unit. Further, the device 800 may also include a large-capacity memory 812 and a read-only memory ("ROM") 813, wherein the large-capacity memory 812 may be configured to store various types of data, including various multimodal data, vascular surface data, positioning point information and distance features, algorithm data, intermediate results, and various programs required to run the device 800. The ROM 813 may be configured to store the power-on self-test of the device 800, the initialization of each functional module in the system, the basic input / output driver of the system, and the data and instructions required to boot the operating system.

[0058] Optionally, the device 800 may also include other hardware platforms or components, such as the tensor processing unit ("TPU") 814, the graphics processing unit ("GPU") 815, the field programmable gate array ("FPGA") 816, and the machine learning unit ("MLU") 817 shown. It is understood that although a variety of hardware platforms or components are shown in the device 800, they are only exemplary and not restrictive, and those skilled in the art can add or remove corresponding hardware according to actual needs. For example, the device 800 may only include a CPU, related storage devices, and interface devices to implement the method for training a segmentation model for segmenting a blood vessel segment or the method for segmenting a blood vessel segment of the present application.

[0059] In some embodiments, in order to facilitate the transmission and interaction of data with an external network, the device 800 of the present application further includes a communication interface 818, so that it can be connected to a local area network / wireless local area network ("LAN / WLAN") 805 through the communication interface 818, and then connected to a local server 806 or to the Internet ("Internet") 807 through the LAN / WLAN. Alternatively or additionally, the device 800 of the present application can also be directly connected to the Internet or a cellular network through the communication interface 818 based on wireless communication technology, such as wireless communication technology based on the third generation ("3G"), the fourth generation ("4G") or the fifth generation ("5G") of the generation. In some application scenarios, the device 800 of the present application can also access a server 808 and a database 809 of an external network as needed to obtain various known algorithms, data and modules, and can remotely store various data, such as various types of data or instructions for presenting, for example, multimodal data, vascular surface data, positioning point information and distance features.

[0060] The peripheral devices of the device 800 may include a display device 802, an input device 803, and a data transmission interface 804. In one embodiment, the display device 802 may include, for example, one or more speakers and / or one or more visual displays, which are configured to train the segmentation model for segmenting the vascular segment of the present application or to perform voice prompts and / or image video display for segmenting the vascular segment. The input device 803 may include other input buttons or controls such as a keyboard, a mouse, a microphone, a gesture capture camera, etc., which are configured to receive input of audio data and / or user instructions. The data transmission interface 804 may include, for example, a serial interface, a parallel interface or a universal serial bus interface ("USB"), a small computer system interface ("SCSI"), a serial ATA, a FireWire ("FireWire"), a PCI Express, and a high-definition multimedia interface ("HDMI"), etc., which are configured for data transmission and interaction with other devices or systems. According to the solution of the present application, the data transmission interface 804 can receive multimodal data collected from CT and MRI equipment, and transmit data or results including multimodal data or various other types to the device 800.

[0061] The CPU 811, mass storage 812, ROM 813, TPU 814, GPU 815, FPGA 816, MLU 817 and communication interface 818 of the device 800 of the present application can be interconnected through a bus 819, and data interaction with peripheral devices can be achieved through the bus. In one embodiment, through the bus 819, the CPU 811 can control other hardware components in the device 800 and its peripheral devices.

[0062] Combination of the above Figure 8The device that can be used to perform the present application is described. It should be understood that the device structure or architecture here is only exemplary, and the implementation method and implementation entity of the present application are not limited thereto, but can be changed without departing from the spirit of the present application.

[0063] According to the above description in combination with the accompanying drawings, those skilled in the art can also understand that the embodiments of the present application can also be implemented by software programs. Therefore, the present application also provides a computer-readable storage medium, on which computer-readable instructions for training a segmentation model for segmenting a blood vessel segment or for segmenting a blood vessel segment are stored. When the computer-readable instructions are executed by one or more processors, they can be used to implement the present application in combination with the accompanying drawings. Figure 1 The method for training a segmentation model for segmenting a blood vessel segment described in the present invention; or implementing the present invention in combination with the attached Figure 7 A method for segmenting blood vessel segments is described.

[0064] It should be noted that although the operations of the method of the present application are described in a specific order in the accompanying drawings, this does not require or imply that the operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. On the contrary, the steps depicted in the flow chart can be performed in a different order. Additionally or alternatively, some steps can be omitted, multiple steps can be combined into one step, and / or one step can be decomposed into multiple steps.

[0065] It should be understood that when the terms "first", "second", "third" and "fourth" are used in the claims, the specification and the drawings of the present application, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the specification and claims of the present application indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their collections.

[0066] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in this application specification and claims, unless the context clearly indicates otherwise, the singular forms of "a", "an" and "the" are intended to include plural forms. It should also be further understood that the term "and / or" used in this application specification and claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0067] Although the implementation methods of the present application are as above, the contents described are only examples adopted to facilitate the understanding of the present application, and are not intended to limit the scope and application scenarios of the present application. Any technician in the technical field described in the present application can make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed in the present application, but the scope of patent protection of the present application shall still be subject to the scope defined in the attached claims.

[0068] In addition, the collection and acquisition of various data in this application complies with relevant laws and regulations and is authorized by the data provider. Any organization or individual that needs to obtain external data must obtain authorization in accordance with the law and ensure data security. It is not allowed to illegally collect, use, process, or transmit unauthorized or unprotected data, or to illegally buy, sell, provide, or disclose unauthorized or unprotected data.

Claims

1. A method for training a segmentation model for segmenting a blood vessel segment, wherein the segmentation model comprises a localization model and a segmentation model, and the method comprises: Acquiring multimodal imaging data, wherein the multimodal imaging data includes digital subtraction angiography data, CT angiography data, and magnetic resonance angiography data; Performing blood vessel segmentation and extracting blood vessel surface based on the multimodal image data to obtain blood vessel surface data; marking positioning points and marking blood vessel segments on the blood vessel surface data, and inputting the marked blood vessel surface data into the positioning model for training to output positioning point information; Calculating the distance feature on the blood vessel surface data according to the positioning point information; and The blood vessel surface data, the distance features and the blood vessel segmentation annotations are input into the segmentation model to perform blood vessel segmentation to obtain blood vessel segments, so as to train the segmentation model used for segmenting blood vessel segments.

2. The method according to claim 1, wherein the positioning points at least include connection points at key blood vessels in the blood vessel surface data; and the positioning point information includes coordinate information of the positioning points or index information of the positioning points.

3. The method according to claim 2, wherein calculating the distance feature on the blood vessel surface data according to the positioning point information comprises: The projection values ​​of the vector formed between the target point and the target positioning point on the blood vessel surface data and the vectors formed between the target positioning point and the remaining positioning points are calculated respectively according to the positioning point information to calculate the distance feature on the blood vessel surface data.

4. The method according to claim 1 or 3, wherein calculating the distance feature on the blood vessel surface data according to the positioning point information further comprises: The projection value of the vector formed between the target point and the target positioning point on the blood vessel surface data on the target vector is calculated according to the positioning point information to calculate the distance feature on the blood vessel surface data; wherein the target vector is perpendicular to the vectors formed between the target positioning point and the remaining positioning points.

5. The method according to claim 4, further comprising: Calculating the geodesic distance on the blood vessel surface data according to the positioning point information; Adding the geodesic distance to the distance feature to obtain a new distance feature; as well as The blood vessel surface data, the new distance feature and the blood vessel segmentation annotation are input into the segmentation model to perform blood vessel segmentation to obtain blood vessel segments, so as to train the segmentation model used for segmenting blood vessel segments.

6. The method according to claim 5, wherein calculating the geodesic distance on the blood vessel surface data according to the positioning point information comprises: Calculating the original geodetic distance from the target point on the blood vessel surface data to each positioning point; and / or The shortest geodesic distance from the target point on the blood vessel surface data to each positioning point is calculated. 7 . The method according to claim 1 , wherein the positioning model and the segmentation model both comprise a point cloud processing model or an image processing model.

8. A method for segmenting a blood vessel segment, comprising: Acquiring image data to be segmented and extracting blood vessel surface data based on the image data to be segmented; as well as The blood vessel surface data is input into a segmentation model trained according to the method of any one of claims 1 to 7 to perform blood vessel segmentation to obtain a blood vessel segmentation result.

9. A device comprising: processor; as well as A memory having computer instructions for training a segmentation model for segmenting a blood vessel segment stored thereon, which, when executed by a processor, implement the method according to any one of claims 1 to 7; or having computer instructions for segmenting a blood vessel segment stored thereon, which, when executed by a processor, implement the method according to claim 8.

10. A computer-readable storage medium storing computer program instructions for training a segmentation model for segmenting blood vessel segments, wherein when the computer program instructions are executed by one or more processors, the method according to any one of claims 1 to 7 is implemented; or storing computer program instructions for segmenting blood vessel segments, wherein when the computer program instructions are executed by one or more processors, the method according to claim 8 is implemented.

Images