Three-dimensional coronary artery automatic segmentation method based on graph attention neural network

Abstract

The invention provides a three-dimensional coronary artery automatic segmentation method based on a graph attention neural network, and belongs to the technical field of medical image segmentation of image segmentation. The technical problem that a traditional method does not fully consider high noise and non-uniform gray scale of a medical image is solved. According to the technical scheme, the method comprises the following steps of: 1, processing window width and window level of an image; 2, reducing the size of the image to be 128 * 128 * 64; 3, performing dilation operation on the image obtained in the step 2; step 4, establishing a graph structure based on the image obtained in the step 2; step 5, segmenting the block images by using a Vnet3D network; and step 6, fusing the coronary artery images obtained by segmentation in the step 4 and the step 5 to obtain a final coronary artery segmentation image. The method has the beneficial effect that the high segmentation accuracy and robustness of the coronary artery are ensured.

Description

Technical Field

[0001] The present invention relates to the fields of computing and deep learning technology, and in particular to a three-dimensional coronary artery automatic segmentation method based on a graph attention neural network. Background Art

[0002] Computed Tomography Angiography (CTA) is a non-invasive vascular imaging technology with the advantages of non-invasiveness, easy operation and clear imaging. It has important application value in the display of vascular lesions and vascular relationships. With the widespread application of CTA technology, vascular segmentation of CTA images has gradually become a hot issue in the field of medical image segmentation. In the analysis of vascular lesions, vascular segmentation algorithms play a particularly important role. However, there are still some challenges in segmenting coronary arteries from CTA images.

[0003] In previous studies, some scholars have tried to propose some methods to solve the problem of segmenting coronary arteries from CTA images. The proposed methods, such as the region growing method, are relatively simple to calculate and have high segmentation efficiency. However, due to the limitation of the growth criterion, the region growing method does not have good segmentation accuracy and universal segmentation criteria, and it is difficult to segment blood vessels with complex structures; the graph cut algorithm can obtain higher segmentation accuracy, but it is less efficient when segmenting complex images; the level set algorithm has simple calculation rules, but it is difficult to effectively converge to the edge of the segmentation target. How to solve the above technical problems is the subject faced by the present invention. Summary of the invention

[0004] The purpose of the present invention is to provide a three-dimensional coronary artery automatic segmentation method based on graph attention neural network.

[0005] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is as follows: it has the characteristics of being fully automated, and no manual intervention is required in the segmentation process. The training process is an end-to-end method.

[0006] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is specifically: a three-dimensional coronary artery automatic segmentation method based on a graph attention neural network, comprising the following steps:

[0007] Step 1: using an anisotropic diffusion filter to remove image noise; in step 1, using an anisotropic diffusion filter to remove image noise includes the following steps:

[0008] Assume that I(x, 0) represents the gray level of the voxel at time t, I0(x) is the original 3D image, div and They represent the divergence operator and the gradient operator respectively, and process the image as follows:

[0009] The conduction coefficient in the above formula is is a monotonically decreasing function as the image gradient increases, defined as:

[0010]

[0011] The coefficient d in the above formula is the threshold used to control the diffusion rate. It is specifically expressed as:

[0012]

[0013] Step 2:

[0014] 1. Image reduction

[0015] Original image size: The original image size is 512 × 512 × 224. On the sagittal plane, with the center point of the image as the center point, the image is reduced to 128 × 128 × 64 using the resize function of the OpenCV image processing library.

[0016] 2: Coarse segmentation of the Vnet3D network. The Vnet3D network consists of an encoder and a decoder, which are composed of multiple stages.

[0017] Resolution change: As data flows over the network, its resolution first decreases and then increases.

[0018] Feature splicing: Similar to U-net, the output feature channel dimension of the same layer on the left is spliced ​​to the right to collect fine-grained details that may be lost and improve the quality of the final contour prediction.

[0019] Stage content: Each stage contains one to three convolutional layers, and adds skip connections (pixel-level summation) to implement the residual function.

[0020] Convolutional layer: Use a 5×5×5 volume kernel (stride=1) to increase the number of channels, and a 2×2×2 volume kernel (stride=2) to reduce the resolution.

[0021] Number of feature channels: The number of feature channels doubles at each stage of the compression path.

[0022] Activation function: PReLu.

[0023] Decoder: Uses convolution operations and skip connections similar to the encoder, and uses 2×2 deconvolution to increase the resolution.

[0024] Output: The two-channel features of the output are classified at the pixel level through the softmax function.

[0025] Vnet3D network segmentation process:

[0026] Input: The image after shrinking in step 1.

[0027] Output: coronary artery segmentation image.

[0028] Training parameters: The network training iterations are 30 epochs, the batch size is 1, and the learning rate is set to 0.001.

[0029] 3: Enlarge the segmentation result

[0030] Use the resize function to enlarge the image segmented by the Vnet3D network to 512×512×224, which is the original image size.

[0031] Step 3: Perform dilation operation on the image obtained in step 2, and then use the thinning method to obtain the centerline of the coronary artery, specifically:

[0032] 1. Dilation operation:

[0033] Dilate the image using a 3x3x3 rectangular structure element. Align the center of the structure element to each pixel and take the maximum value of all pixels covered. If the structure element has at least one foreground pixel, the resulting pixel is set as the foreground. Repeat the above steps until all pixels are processed.

[0034] 2. Refinement method of centerline extraction:

[0035] In three-dimensional space, according to the Euclidean distance between point P and its neighboring points, it is divided into 6-adjacency, 18-adjacency and 26-adjacency. N(P) represents the neighborhood of point P, U, D, W, E, S, and N represent the six main directions, and N18(P) and the five-pointed star represent 18-adjacency. It operates on the binary image, 1 is the target point, and 0 is the background point. If N26(P) / Pi has exactly one target point, P is the endpoint of the curve; if N(P) contains at least one pair of relative background points, P is the endpoint of the surface. The boundary point is 6-adjacent to at least one background point. Simplified points can be deleted without changing the graph topology, and are composed of certain types of boundary points. The endpoints of curves or surfaces are related to the graph topology and cannot be deleted.

[0036] 3. Three-dimensional blood vessel thinning processing:

[0037] The present invention uses 12 directions to refine the blood vessels after three-dimensional reconstruction. Each iterative step is refined in parallel by 12 directions. The deletion direction is divided into 4 groups in sequence, each group has 3 directions, and each group contains 6 main directions. In each sub-iteration process, the boundary points in a specific direction can be deleted. The coronary artery shrinks evenly in each direction, corroding from the outer layer to the inside. When there are no more target points that can be deleted in the 12 sub-iteration processes, the centerline extraction is completed.

[0038] Step 4: As a three-dimensional coronary artery automatic segmentation method based on a graph attention neural network provided by the present invention, in the step 4, a graph structure is established based on the image obtained in step 2, and based on the coronary artery centerline obtained in step 3, a three-dimensional coronary artery image is segmented using a graph attention neural network, specifically comprising the following steps:

[0039] 1. Mesh construction: The vascular wall surface mesh is represented by a graph G(V,E), where V and edges E represent the vertices and connecting edges of the mesh, respectively.

[0040] 2. Vascular wall and centerline: The vascular wall is regarded as a deformable tube in 3D space, and the position of each vertex is determined by the parameter rv, which is the distance from the vertex to the centerline C.

[0041] 3. Cross-section and vertex (node) positioning: Each centerline point ci corresponds to an orthogonal 2D cross-section, and the vertices are distributed at equal angular intervals in the cross-section. The position is expressed by polar coordinates (φv, rv).

[0042] 4. Mesh formation and edge connection: All cross-section vertices are combined to form a surface mesh G. Edges are added between adjacent vertices and between vertices with the same angle φv to form a quadrilateral mesh surface, which is further divided into triangular surfaces.

[0043] 5. Graph Attention Network (GAT): GAT assigns weights to nodes and neighbors through a self-attention mechanism, focusing on important neighbors and ignoring irrelevant nodes.

[0044] 6. Attention mechanism details: The attention mechanism is a single-layer feedforward neural network that uses the LeakyReLU nonlinear activation function to calculate the attention coefficient between nodes.

[0045] 7. Feature output and model training: The normalized attention coefficient is used to calculate the linear combination of node features as the output feature. GAT predicts the rv value of each vertex based on image information.

[0046] Step 5: As a three-dimensional coronary artery automatic segmentation method based on a graph attention neural network provided by the present invention, in step 5, based on the center line of step 3 and the original image, block images of sizes 32×32×32 and 64×64×64 are obtained, and the block images are segmented using the Vnet3D network. Specifically:

[0047] 1. Select center voxel: Select a voxel from the center line as the center point.

[0048] 2. Extract block images: With the central voxel as the center, extract 32×32×32 and 64×64×64 cubic block images, and perform AND operations with the original image to obtain block images. If the centerline voxel is located at the end of the CTA image slice sequence and it is impossible to intercept a cubic block image with it as the center, then start from the slice where the centerline voxel is located and extend in the opposite direction of the slice sequence to intercept a complete cubic block image.

[0052] 3. Input Vnet3D network: Send the block image as input to the Vnet3D network, and the network parameters are the same as the previous step.

[0054] 4. Segmentation and fusion: The Vnet3D network segments the block image and then fuses the segmentation results to generate a segmented image with the same size as the original image.

[0055] Step 6: As a three-dimensional coronary artery automatic segmentation method based on graph attention neural network provided by the present invention, in step 6, the coronary artery images obtained by step 4 and step 5 are fused to obtain the final coronary artery segmentation image.

[0056] The result image is obtained by segmenting the Vnet3D block, denoted as A, and the result image is obtained by segmenting the graph attention, denoted as B.

[0057] 1. Define pixel value: Let A(x,y,z) and B(x,y,z) represent the pixel values ​​of images A and B at the coordinates (x,y,z) in three-dimensional space respectively.

[0058] 2. Weighted average of identical pixels: For identical pixels in images A and B, i.e. A(x,y,z) = B(x,y,z), we calculate the weighted average:

[0059] F(x,y,z)=W A A(x,y,z)+W B B(x,y,z)

[0060] Where W A ,W B are the weights of images A and B respectively, and W A +W B = 1. Here we set W A =0.5,W B =0.5.

[0061] 3. Union of different pixels: For different pixels in images A and B, that is, A(x,y,z)≠B(x,y,z), take the union, that is, keep the two pixel values:

[0062] F(x,y,z)={A(x,y,z),B(x,y,z)}

[0063] 4. Fusion logic formula: Combining the above two situations, the fused image F(x,y,z) is expressed as follows:

[0064]

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] 1. The present invention combines the feature extraction characteristics and advantages of different image segmentation methods to design a new fully automatic coronary artery segmentation method. It has the characteristics of full automation and does not require manual intervention during the segmentation process. The training process is an end-to-end method.

[0067] 2. The present invention uses a cascade network and two methods to segment the coronary artery: the first segmentation branch uses Vnet3D to segment the block image, and the second segmentation branch uses GAT to segment the block image.

[0068] 3. In the present invention, the coarse segmentation result of Vnet3D is refined to obtain the centerline of the coronary artery, and the coarse segmentation result is used to generate a graph structure as the input of the graph convolutional network.

[0069] 4. The present invention uses a graph convolutional network to achieve the final segmentation of the coronary artery and obtain the segmentation result. During the segmentation process, GCN predicts the radius distance of the coronary artery boundary point from the center based on the center point to determine the position of the boundary point. All boundary points on the cross section constitute the edge contour of the coronary artery. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used together with Example 1 and Example 2 of the present invention to explain the present invention, but do not constitute a limitation of the present invention.

[0071] Figure 1 This is an overall flow chart of the three-dimensional coronary artery automatic segmentation method based on graph attention neural network provided by the present invention.

[0072] Figure 2 Schematic diagrams of 6-adjacency, 18-adjacency and 26-adjacency proposed in step 3 of the three-dimensional coronary artery automatic segmentation method based on graph attention neural network provided by the present invention.

[0073] Figure 3 This is a schematic diagram of 12 directions in step three of the three-dimensional coronary artery automatic segmentation method based on graph attention neural network provided by the present invention.

[0074] Figure 4 Schematic diagram of graph attention neural network prediction nodes in step 4 of the three-dimensional coronary artery automatic segmentation method based on graph attention neural network provided by the present invention.

[0075] Figure 5 This is an example diagram of the block image obtained in step five of the three-dimensional coronary artery automatic segmentation method based on graph attention neural network provided by the present invention, the upper left and upper right images are the original block image images, and the lower left and lower right images are the block image segmentation result images.

[0076] Figure 6 This is a segmentation effect diagram of the three-dimensional coronary artery automatic segmentation method based on graph attention neural network provided by the present invention. The left figure is the true value map provided by the data set, and the right figure is the result map obtained by the segmentation of the present invention. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and Example 1 and Example 2. Of course, the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention.

[0078] Example 1

[0079] See also Figures 1 to 6 ,This Example 1 proposes a three-dimensional coronary artery automatic segmentation method based on graph attention neural network.

[0080] In this embodiment, the experiment of this embodiment is implemented on the Windows 10 platform using the Python language, and the computer platform implementing this solution has an Intel Core i5 1.6GHz CPU, 16GB of running memory, and 8GB of GPU memory. During the experiment, the model was tested and evaluated on the ImageCAS public dataset (the ImageCAS dataset contains 1000 CTA image samples), and compared with other methods that also use this dataset.

[0081] This embodiment includes the following steps:

[0082] Step 1: Use anisotropic diffusion filter to remove image noise;

[0083] Step 2: Reduce the image to 128×128×64, use the Vnet3D network to roughly segment the reduced image, and then enlarge the segmentation result to 512×512×224;

[0084] Step 3: Perform dilation operation on the image obtained in step 2, and then use the thinning method to obtain the centerline of the coronary artery;

[0085] Step 4: Build a graph structure based on the image obtained in step 2, and segment the 3D coronary artery image using a graph attention neural network based on the coronary artery centerline obtained in step 3;

[0086] Step 5: Based on the center line of step 3 and the original image, block images of sizes 32×32×32 and 64×64×64 are obtained, and the block images are segmented using the Vnet3D network;

[0087] Step 6: Fusing the coronary artery images segmented in step 4 and step 5 to obtain the final coronary artery segmentation image.

[0088] Step 7: Use FCN and Unet3D as baseline methods for experimental control. To achieve the best segmentation results for FCN and Unet3D, train for 50 epochs. The learning rate is 0.001, the same as Vnet3D. The training batch size is 1, the same as Vnet3D.

[0089] The effects of the three models are shown in Table 1 below:

[0090]

[0091] As can be seen from the above table, the method proposed in this embodiment has better results than the other two methods in terms of HD (Hausdorff Distance), AHD (Average Hausdorff Distance) and DSC (Dice Similarity Coefficient) indicators, indicating that this embodiment can achieve good coronary artery segmentation effect.

[0092] Example 2

[0093] In order to further demonstrate the effect of using parallel Vnet3D block segmentation and GAT network to segment coronary arteries in this embodiment, Example 2 is given on the basis of Example 1 to illustrate the segmentation results in the following two different cases: A. The proposed method does not use GAT for segmentation; B. The proposed method uses GAT for segmentation after using Vnet3D for segmentation, and uses the Vnet block segmentation result and the GAT segmentation result to fuse as the final segmentation result.

[0094] The model effect is shown in Table 2 below:

[0095]

[0096] It can be seen from Table 2 above that the method proposed in this embodiment has better results than the other two methods in terms of HD (Hausdorff Distance), AHD (Average Hausdorff Distance) and DSC (Dice Similarity Coefficient) indicators, indicating that the innovation in this embodiment can improve the effect of coronary artery segmentation.

[0097] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A three-dimensional coronary artery automatic segmentation method based on graph attention neural network, characterized in that: The following steps are involved: Step 1: Use anisotropic diffusion filter to remove image noise; Step 2: Reduce the image to 128×128×64, use the Vnet3D network to roughly segment the reduced image, and then enlarge the segmentation result to 512×512×224; Step 3: Perform dilation operation on the image obtained in step 2, and then use the thinning method to obtain the centerline of the coronary artery; Step 4: Build a graph structure based on the image obtained in step 2, and segment the 3D coronary artery image using a graph attention neural network based on the coronary artery centerline obtained in step 3; Step 5: Based on the center line of step 3 and the original image, block images of sizes 32×32×32 and 64×64×64 are obtained, and the block images are segmented using the Vnet3D network; Step 6: Fusing the coronary artery images segmented in step 4 and step 5 to obtain the final coronary artery segmentation image.

2. The three-dimensional coronary artery automatic segmentation method based on graph attention neural network according to claim 1, characterized in that: In the step 1, using an anisotropic diffusion filter to remove image noise includes the following steps: Assume that I(x, 0) represents the gray level of the voxel at time t, I0(x) is the original 3D image, div and They represent the divergence operator and the gradient operator respectively, and process the image as follows: The conduction coefficient in the above formula is It is a monotonically decreasing function as the image gradient increases, defined as follows: The coefficient d in the above formula is the threshold for controlling the diffusion rate.

3. The three-dimensional coronary artery automatic segmentation method based on graph attention neural network according to claim 1, characterized in that: In the step 2, the image is reduced to a size of 128×128×64, the reduced image is roughly segmented using the Vnet network, and then the segmentation result is enlarged to 512×512×224; specifically, the following steps are included: (2-1) Image preprocessing: Image resizing: Use the resize function of the OpenCV image processing library to resize the original image from 512×512×224 to 128×128×64; (2-2) Vnet3D network structure Encoder and decoder composition: The Vnet3D network consists of an encoder and a decoder, which are composed of multiple stages. The data is first reduced in resolution and then increased in resolution in the network. In the compression path, the output feature channel dimensions of the same level are spliced ​​to the right to collect the lost fine-grained details and improve the quality of contour prediction; (2-3) Convolutional layers and skip connections Convolutional layer configuration: Each stage contains one to three convolutional layers, using a 5×5×5 convolution kernel, without changing the size, only increasing the number of channels, followed by a 2×2×2 convolution kernel with stride=2 to halve the resolution. The number of feature channels is doubled at each stage of the compression path, which is in line with the design of the residual network. The activation function of the network is PReLu; Skip connection: Implement the residual function through pixel-level summation; (2-4) Decoder Operation Decoder convolution operations: Each stage of the decoder uses convolution operations and skip connections similar to those of the encoder; Deconvolution operation: After the convolution, a 2×2 deconvolution operation is used to double the resolution in each dimension. (2-5) Output and classification Feature channel classification: The output two-channel features are classified at the pixel level through the softmax function; (2-6) Vnet segmentation process Input and output: The input is the reduced image obtained in step 3, and the output is the coronary artery segmentation image; Training parameters: The network training iterations were 30 epochs, the batch size was 1, and the learning rate was set to 0.001; (2-7) Image post-processing Image size restoration: After segmentation using Vnet, the original image is resized back to 512×512×224 using the resize function, which is the original size of the original image.

4. The three-dimensional coronary artery automatic segmentation method based on graph attention neural network according to claim 1, characterized in that: In the step 3, the image obtained in the step 2 is dilated, and then the coronary artery centerline is obtained using a thinning method, which specifically includes the following steps: (3-1) Expansion operation: Select structural element: Select a rectangular structural element of size 3×3×3; Align pixels: Align the center of the structure element to each pixel in the image; Take the maximum value: For all pixels covered by the structural element, take the maximum value; Foreground pixel setting: If there is at least one foreground pixel in the structure element, the result pixel is set to foreground; Repeat processing: Repeat the above steps until all pixels of the image are processed and the dilation operation is completed; (3-2) Extracting centerline by thinning method Three-dimensional space point P: determine a point P in three-dimensional space; Adjacency relationship classification: According to the Euclidean distance from point P to its adjacent points, the relationship is divided into 6-adjacency, 18-adjacency and 26-adjacency; Neighborhood definition: N(P), j=6, 18, 26 represents the set of points P and i's neighboring points, i.e., the neighborhood; Direction representation: U, D, W, E, S, N are used to represent the six directions in the six adjacencies, and N18 (P) and six five-pointed star symbols are used to represent N2 (P); Endpoint and surface endpoint determination: When N26(P) / Ni contains exactly one target; Binarization image operation: Perform operations on the binary image, define points with a value of 1 as target points, and points with a value of 0 as background points.

5. The three-dimensional coronary artery automatic segmentation method based on graph attention neural network according to claim 1, characterized in that: In the step 4, a graph structure is established based on the image obtained in the step 2, and a graph attention neural network is used to segment the three-dimensional coronary artery image based on the coronary artery centerline obtained in the step 3, which specifically includes the following steps: (4-1) Mesh representation and vascular wall model Mesh model definition: The blood vessel wall surface mesh is represented by a graph G(V,E), where V represents vertices and E represents edge connections between vertices; Blood vessel wall assumption: Assume that the blood vessel wall is a deformable tube in 3D Euclidean space with a known centerline C; Vertex position parameterization: The position of each vertex depends only on the parameter rv, which is the distance to the center line point; (4-2) Determination of center line and cross section Centerline points and cross sections: For each centerline point ci, determine a 2D cross section orthogonal to the centerline direction; Cross-section vertex definition: In the cross-section plane, equiangularly spaced vertices define the cross-section of the surface mesh; (4-3) Polar coordinate definition and grid formation Polar coordinate definition: In the 2D plane, the position of each vertex is defined in polar coordinates (φv, rv), where φv is fixed; Polygonal inner cavity surface mesh formation: The combination of all vertices in all cross sections forms a polygonal inner cavity surface mesh G; Adding edges: Add edges between adjacent vertices in a cross section and between vertices of the same angle φv in adjacent cross sections; (4-4) Mesh surface definition and subdivision Quadrilateral mesh surface definition: Four edges define a quadrilateral mesh surface; Triangular face subdivision: Each quadrilateral face is further divided into two triangular faces; (4-5) Graph Attention Mechanism GAT Self-attention mechanism: GAT uses a self-attention mechanism to assign different weights to each node and its neighbors; Focus: The model is able to focus on more important neighbors when aggregating neighbor information and ignore less relevant nodes; Attention mechanism implementation: The attention mechanism a is a single-layer feed-forward neural network parameterized by a weight vector and applies a LeakyReLU nonlinearity with a negative input slope α=0.2; (4-6) Attention coefficient calculation and feature output Attention coefficient calculation: After full expansion, the coefficient calculated by the attention mechanism is expressed as a specific mathematical expression, involving transposition and concatenation operations. The calculation formula for the importance of node j to node a is expressed as follows: LeakyReLU is a nonlinear activation function with a negative input slope of α = 0.2, where T represents transposition and || represents concatenation. The attention mechanism a is a single-layer feedforward neural network parameterized by the weight vector is the input vector, where F is the feature dimension; Normalization and linear combination: The normalized attention coefficient is used to calculate the linear combination of the corresponding features as the final output feature of each node; (4-7) GAT training and prediction GAT training objective: GAT is trained to predict the rv value of each vertex based on the information of image X encoded in the input vector xv; Predicting rv value: Through training, GAT can predict the distance rv from each vertex to the centerline point, thereby assisting in the construction and analysis of the vascular wall surface mesh.

6. The three-dimensional coronary artery automatic segmentation method based on graph attention neural network according to claim 1, characterized in that: In the step 5, based on the center line of step 3 and the original image, block images of sizes 32×32×32 and 64×64×64 are obtained respectively, and the block images are segmented using the Vnet3D network, specifically including the following steps: (5-1) Block image extraction preparation Select center voxel: select a voxel from the center line as the center point; Define cube size: Determine the sizes of the two cubes, 32×32×32 and 64×64×64 respectively; (5-2) Block image extraction Obtain block image by AND operation: Use the above two cubes of different sizes to perform AND operation with the original image to obtain a partial image of the original image, i.e., the block image; Block image generation: Through AND operation, two block images of different sizes are obtained; (5-3) Vnet3D network input Input block image: The extracted block image is used as the input of the Vnet3D network; Parameter setting: The parameter setting of the Vnet3D network is the same as in step 2 to ensure that the network can correctly process the input block image; (5-4) Block Image Segmentation Vnet3D network segmentation: Use the Vnet3D network to segment the input block image; Segmentation result acquisition: After network processing, the segmentation result of each block image is obtained; (5-5) Segmentation result fusion Result image fusion: fuse the block image results obtained by Vnet3D network segmentation; Restore the original image size: Through fusion, obtain a segmented image with the same size as the original image; (5-6) Final segmentation image output Output segmented image: The final segmented image is the desired result. It has the same size as the original image but contains detailed information after segmentation.

7. The method for automatic segmentation of three-dimensional coronary arteries based on graph attention neural network according to claim 6, characterized in that: In the step six, the coronary artery images obtained by the segmentation of step four and step five are fused to obtain a final coronary artery segmentation image, specifically: the segmentation of the coronary artery is a parallel segmentation network, and the final segmentation result image is obtained by fusing the coronary artery images obtained by the segmentation of step four and step five.

Images