CCTA image-based full coronary artery tree complete automatic segmentation and blood vessel center line extraction system

Abstract

The invention discloses a CCTA image-based full coronary artery tree complete automatic segmentation and blood vessel center line extraction system, and the system comprises a positioning module which is used for carrying out the positioning and extraction of a heart according to a CCTA image; the extraction module is used for performing primary extraction of coronary artery tree segmentation according to the CCTA image after heart positioning to obtain a coronary artery tree; the repairing module is used for extracting the center line of the blood vessel based on the extracted coronary artery tree, and extending the center line at the breakpoint to connect the fractures, or further extending the incomplete center line to the tail end of the blood vessel to obtain a repaired center line; the supplementing module is used for supplementing the coronary artery tree structure and repeating the steps of blood vessel repairing and coronary artery tree supplementing to obtain a coronary artery blood vessel area; and the reconstruction module is used for obtaining a coronary artery tree blood vessel result according to the multiple planes of the reconstructed blood vessel. According to the method, full-automatic accurate extraction of the complete coronary vessel tree can be realized, and technical innovation and clinical application of coronary imaging are promoted.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and in particular relates to a complete automatic segmentation system of the entire coronary artery tree and vascular centerline extraction based on CCTA images. Background Art

[0002] Coronary computed tomography (CCTA), a mature and stable imaging modality, is widely used in the clinical diagnosis of coronary artery disease. This technique uses multi-slice spiral CT to perform volumetric scans of the coronary arteries using rapid injection of contrast agent, revealing the coronary arteries and surrounding anatomical structures and pathological conditions. CCTA can provide geometric and topological relationships of the cardiovascular system, which is crucial for the identification of coronary artery disease.

[0003] With the development of neural networks and deep learning technologies, deep learning-assisted medical image processing has made significant progress in tissue segmentation, lesion detection, and disease identification. Especially in the field of coronary artery segmentation, multiple models have been proposed and have made certain progress in image segmentation and vascular feature extraction.

[0004] Although CCTA technology plays an important role in the diagnosis of coronary artery disease, it still faces challenges in practical application. The original CCTA image usually only shows a certain cross-section of the coronary artery, making it difficult to accurately and systematically evaluate the entire coronary structure. Currently, clinicians need to manually outline the coronary tree area in the CCTA image and perform multi-planar reconstruction based on these outlines to fully display the overall structure of the coronary artery and then determine the abnormal area. This process is not only time-consuming and labor-intensive, but also reduces the efficiency of processing CCTA images, which may affect the timeliness and accuracy of diagnosis. In addition, due to the complexity of the coronary vascular structure, the lack of visibility of small blood vessels, and the interference of other tissues around the heart on vascular segmentation, existing automatic segmentation models often have problems such as vessel loss, breakage, and a large number of false positive segmentations. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes a complete automatic segmentation system of the entire coronary artery tree and vascular centerline extraction based on CCTA images to solve the problems existing in the above-mentioned prior art.

[0006] To achieve the above objectives, in a first aspect, the present invention provides a system for fully automatic segmentation of the entire coronary artery tree and extraction of vascular centerlines based on CCTA images, comprising:

[0007] Positioning module, used to locate and extract the heart based on CCTA images;

[0008] An extraction module is used to perform initial extraction of the coronary artery tree segmentation based on the CCTA image after cardiac positioning to obtain the coronary artery tree;

[0009] The repair module is used to extract the vascular centerline based on the extracted coronary artery tree and extend the centerline at the breakpoint to connect the broken parts to obtain the repair centerline;

[0010] a supplementation module, configured to supplement the coronary tree structure based on the repair centerline, and repeat the steps of vessel repair and coronary tree supplementation to achieve complete segmentation of the entire coronary tree and obtain the coronary vessel region;

[0011] The reconstruction module is used to reconstruct multiple planes of the blood vessels according to the coronary vessel region to obtain a two-dimensional coronary artery tree result.

[0012] Preferably, the positioning module includes:

[0013] The heart annotation unit is used to annotate the heart position and generate a 3D bounding box as the training and test sets of the model;

[0014] The first construction unit is used to build a positioning model based on ResNet-18;

[0015] The model positioning unit is configured to train the positioning model using the training set to obtain a trained positioning model, and input the test set into the trained positioning model to obtain a coronary artery tree positioning image.

[0016] Preferably, the extraction module includes:

[0017] a data set acquisition unit, configured to divide the coronary tree localization image into a training set and a test set;

[0018] The second construction unit is used to build a deep learning model based on 3D-Unet;

[0019] The model extraction unit is used to train the deep learning model using the training set to obtain a trained deep learning model, and input the test set into the trained deep learning model to obtain a preliminary extraction result of the coronary artery tree.

[0020] Preferably, the repair module includes:

[0021] An extraction unit, configured to extract the centerline of the coronary artery using a skeletonization function;

[0022] An image processing unit, which resamples the image and crops the image blocks into a specified size;

[0023] The model repair unit is used to construct a neural network model, and repair and improve the blood vessel centerline through the neural network model to obtain a repaired centerline.

[0024] Preferably, the supplementary module includes:

[0025] an information combining unit, configured to combine the repair centerline with the boundary information of the blood vessel to reconstruct the coronary artery tree structure;

[0026] The structure optimization unit is used to analyze and optimize the reconstructed coronary tree structure to obtain a coherent and consistent entire coronary tree vascular structure.

[0027] Preferably, the reconstruction module includes:

[0028] a two-dimensional mapping unit, configured to map the extracted coronary tree vessel structure onto a two-dimensional plane using a multi-planar reconstruction algorithm;

[0029] an image visualization unit, configured to generate a coronary vessel visualization image from the coronary tree vessel structure;

[0030] The coronary vascular display unit is used to clearly display the complete coronary vascular system.

[0031] In a second aspect, the present invention further discloses a computer device comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the system described in the first aspect.

[0032] In a third aspect, the present invention further discloses a computer-readable storage medium having a computer program stored thereon, which implements the steps of the system described in the first aspect when executed by a processor.

[0033] In a fourth aspect, the present invention further discloses a computer program product, comprising a computer program, which implements the steps of the system described in the first aspect when executed by a processor.

[0034] Compared with the prior art, the present invention has the following advantages and technical effects:

[0035] The present invention provides a complete automatic segmentation system for the entire coronary tree and vascular centerline extraction based on CCTA images, comprising: a positioning module for positioning and extracting the heart based on CCTA images; an extraction module for performing initial extraction for coronary tree segmentation based on CCTA images after heart positioning to obtain a coronary tree; a repair module for extracting vascular centerlines based on the extracted coronary tree, and extending the centerlines at breakpoints to connect the breaks to obtain a repaired centerline; a supplementation module for supplementing the coronary tree structure based on the repaired centerline, and repeating the vascular repair and coronary tree supplementation steps to achieve complete segmentation of the entire coronary tree and obtain a coronary vascular region; and a reconstruction module for reconstructing multiple planes of the vascular structure based on the complete coronary vascular region to obtain a coronary tree vascular result.

[0036] The present invention solves the current problems of incomplete coronary artery tree extraction, vascular breakage and loss, and a large number of false-positive segmentations. It can achieve fully automatic and accurate extraction of the complete coronary artery tree, reduce manual intervention, improve work efficiency, and generate high-quality visual images, assisting clinical decision-making and promoting technological innovation and clinical application of coronary imaging. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0038] Figure 1 Schematic diagram of a system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0039] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0040] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0041] Example 1

[0042] like Figure 1 As shown, this embodiment provides a complete automatic segmentation system for the entire coronary artery tree and vascular centerline extraction based on CCTA images, including:

[0043] Positioning module, used to locate and extract the heart based on CCTA images;

[0044] Specifically, based on the distribution of the heart in the CCTA image, the position of the heart bounding box is manually determined and accurately labeled, and a residual neural network model composed of three dimensions of x, y, and z is used to perform positioning analysis and cropping extraction of the heart.

[0045] Furthermore, the positioning module includes:

[0046] The heart annotation unit is used to annotate the heart position and generate a 3D bounding box as the training and test sets of the model;

[0047] Specifically, the heart is annotated in detail according to its position in each image. At the same time, a corresponding 3D bounding box is generated for each image. This framework must not only ensure that all parts of the coronary tree are completely included, but also try to exclude other tissues around the heart that may interfere with the segmentation. This step usually involves manual or semi-automatic methods for precise calibration to ensure the accuracy of the bounding box. Ultimately, the generated 3D bounding box will serve as the basic data for subsequent model training and inference, providing accurate geometric information for the extraction and analysis of the coronary tree structure.

[0048] This embodiment innovatively uses positioning and cutting to avoid misjudging tubular structures of other tissues (such as the bronchial part of the lung) as coronary arteries, which can greatly avoid false positives.

[0049] The first construction unit is used to build a positioning model based on ResNet-18;

[0050] Specifically, based on the ResNet-18 network, its architecture was modified so that the network output is 4 boundary values, which represent the position coordinates of the heart in 3D space. A total of three independent ResNe-18 network positioning models were established, respectively for the positioning tasks of CCTA images in the three axes of x, y, and z. Each model focuses on positioning in a specific direction, which can make full use of the spatial information of the image and improve the accuracy of positioning. Finally, reliable heart positioning results are obtained by statistically analyzing the frequency data output by the model in multiple training sessions; finally, the positioning results are obtained by statistically analyzing the frequency;

[0051] The model positioning unit is configured to train the positioning model using the training set to obtain a trained positioning model, and input the test set into the trained positioning model to obtain a coronary artery tree positioning image.

[0052] Specifically, based on the annotated 3D bounding boxes, the image is sliced ​​along the x, y, and z axes to obtain a localization network dataset for each direction. This dataset is then divided into training, validation, and test sets to facilitate subsequent model training and performance evaluation. During the training phase, the model is trained and adjusted using the training and validation sets, and model performance is improved by continuously optimizing network parameters. To measure the model's localization capabilities, the mean square error (MSE) is used as a loss function to calculate the model's error during the localization process. This effectively monitors and optimizes the model's performance, ultimately resulting in more accurate coronary artery tree localization results.

[0053] An extraction module is used to perform initial extraction of the coronary artery tree segmentation based on the CCTA image after cardiac positioning to obtain the coronary artery tree;

[0054] Specifically, a dataset of CCTA images was obtained, including some publicly available and private CCTA datasets. The public datasets contained fully annotated coronary artery trees, while the private datasets were annotated by professional physicians. After data preparation, the CCTA images were processed using an improved residual network model to obtain preliminary coronary artery tree localization results.

[0055] Furthermore, the extraction module includes:

[0056] a data set acquisition unit, configured to divide the coronary tree localization image into a training set and a test set;

[0057] Specifically, based on existing CCTA datasets with complete coronary artery tree annotation (public and private CCTA datasets), the training set, validation set, and training set were divided, and the images were uniformly adjusted in terms of window width and window position to ensure consistency in contrast and brightness across all images, facilitating subsequent analysis and processing. Based on the positioning results from the first phase, the images and corresponding labels were cropped to ensure that the annotations corresponded to the images, excluding non-relevant areas that interfered with model training, retaining more information related to the coronary artery tree, and providing clearer basic data for subsequent segmentation.

[0058] The second construction unit is used to build a deep learning model based on 3D-Unet;

[0059] Specifically, based on the 3D-Unet network, residual blocks are introduced in the convolution blocks of the encoder and decoder to solve the problems of gradient disappearance and explosion. Each residual block consists of a 3D convolution, a batch normalization (BN) layer, and a linear correction unit with a jump connection. In the jump connection part, in order to further improve the sensitivity of the model to important features, an attention gate module (AG) is added to effectively suppress the interference of irrelevant areas in the image on the segmentation. During the processing, the attention gate module inputs the two matrices to be spliced ​​in the jump connection into the attention module for processing. The module also contains 3D convolution and BN layers. The calculation results are processed by the LeakyRelu activation function and then spliced. Finally, the integrated output result is obtained through convolution and sampling operations. This mechanism significantly enhances the network's ability to extract meaningful features and improves the accuracy of segmentation.

[0060] The model extraction unit is used to train the deep learning model using the training set to obtain a trained deep learning model, and input the test set into the trained deep learning model to obtain a preliminary extraction result of the coronary artery tree.

[0061] Specifically, the CCTA and its annotations are divided into training, validation, and test sets. The training and validation sets are used for model training and hyperparameter tuning, while the test set is used to evaluate the model's performance on unseen data. The Dice correlation coefficient (DICE) and binary cross entropy (BCE) are used as loss functions, calculated as follows:

[0062]

[0063] Among them, y n is the true label corresponding to the nth category, is the prediction result for the nth category, where N is the total number of categories. The DICE loss function effectively increases the model's sensitivity to the completeness of the overall coronary tree extraction, allowing it to focus more on the integrity of the overall structure. The BCE loss function, on the other hand, focuses more on the extraction accuracy of each local region, emphasizing the model's accuracy in processing details. Combining these two loss functions helps enhance model performance, making the initial coronary tree extraction results more accurate and reliable.

[0064] The repair module is used to extract the vascular centerline based on the extracted coronary artery tree and extend the centerline at the breakpoint to connect the broken parts to obtain the repair centerline;

[0065] This embodiment innovatively uses a method for extracting and repairing the centerline of the coronary tree to ensure the continuity and integrity of the extraction results.

[0066] Specifically, based on the preliminary segmentation results of the coronary tree, the vascular centerline structure of the coronary tree is further extracted and optimized. These centerlines provide the basic direction and branching information of the coronary tree; a neural network model that can track the direction of the vascular centerlines is constructed to repair and improve the vascular centerlines. This model can automatically identify missing parts of the centerlines and generate coherent vascular paths; based on relevant machine learning algorithms, the false positive areas generated during the segmentation process are filtered and eliminated to obtain more accurate and complete coronary tree extraction results.

[0067] Furthermore, the repair module includes:

[0068] An extraction unit, configured to extract the centerline of the coronary artery using a skeletonization function;

[0069] Specifically, for existing CCTA images with annotated coronary arteries, a skeletonization function is first used to extract the coronary artery centerlines. This step refines the vascular structure, making the main trunk of the vessels more prominent in the image. The image is then resampled to a uniform resolution. A cropping center is randomly selected on the centerline, and image blocks of a specified size are cropped. Based on the distribution of the vessel centerlines, two vectors of the vessel extension directions are calculated for each image block. To expand the dataset and enhance the robustness of the model, the center of the cropped image block is translated in any direction by a certain length, and the direction vectors are recalculated for data augmentation.

[0070] An image processing unit, which resamples the image and crops the image blocks into a specified size;

[0071] Specifically, during label generation, a sphere with a radius of 1 is constructed, and 50 direction points are evenly distributed on its surface. These direction points represent the possible directions of blood vessels. Based on the direction vectors obtained during the data processing phase, the direction points closest to these direction vectors are found and assigned corresponding classification labels.

[0072] In this embodiment, the image processing unit performs the following functions: "For one endpoint of a centerline, determine the direction in which it should be extended and extend it by a certain length, repeating this process until it is determined that no extension is necessary." Specifically, the unit: 1. For a broken centerline, after finding one end, use the unit to extend it to the other end of the broken line, completing the "connection." 2. For an incomplete centerline extracted from a blood vessel end, use the unit to extend it to extract the complete vessel end.

[0073] The model repair unit is used to construct a neural network model, and repair and improve the blood vessel centerline through the neural network model to obtain a repaired centerline.

[0074] Specifically, the network is constructed, taking the cropped image blocks as input, and the model is expected to output two direction categories. To achieve this goal, a variety of deep learning architectures can be selected, such as convolutional neural networks (CNN), residual networks (Res-Net), or vision transformers (VIT). These models each have their own advantages and can effectively capture the feature information in the image. When selecting a specific model, considering the characteristics of the data and the requirements of the task, the model can be fine-tuned and optimized to ensure the best performance in the coronary artery centerline classification task;

[0075] Specifically, centerline repair, based on clustering methods in machine learning, clusters the coronary artery centerlines initially segmented in the second stage according to vascular branches, filtering out shorter branches to focus on the longer main vessels. For each identified breakpoint, the centerline repair model is used to extend it, connecting it to other vascular branches or extending it to the end. During this process, the repair model will gradually analyze the surrounding structure of each breakpoint and determine through reasoning whether it needs to be extended or whether it has already been connected to adjacent vascular branches, thereby obtaining a more complete coronary artery centerline structure.

[0076] a supplementation module, configured to supplement the coronary tree structure based on the repair centerline, and repeat the steps of vessel repair and coronary tree supplementation to achieve complete segmentation of the entire coronary tree and obtain the coronary vessel region;

[0077] This embodiment innovatively repairs the initially extracted blood vessels through a supplementary module to improve the integrity of the segmentation results.

[0078] Specifically, the coronary tree structure is supplemented based on the patched centerline. This process involves combining the centerline with vessel boundary information to reconstruct the complete coronary tree structure. Analysis and optimization of the supplemented structure ensures the authenticity and consistency of the individual branches, curvatures, and connections of the coronary tree. Parts of the third stage are further repeated to ensure high-quality extraction results. Through continuous iterative optimization, we are able to effectively improve the segmentation accuracy of the coronary tree and gradually achieve complete segmentation of the entire coronary tree.

[0079] Furthermore, the supplementary module includes:

[0080] an information combining unit, configured to combine the repair centerline with the boundary information of the blood vessel to reconstruct the coronary artery tree structure;

[0081] Specifically, the repaired coronary artery centerline is combined with the vessel boundary information. The boundary information provides the outer contour of the vessel, while the centerline represents the main path of the vessel. By combining the two, a complete coronary artery tree structure can be reconstructed. Specifically, image segmentation and edge detection algorithms can be used to extract boundary information, and geometric models can be used to align the centerline with the boundary information.

[0082] The structure optimization unit is used to analyze and optimize the reconstructed coronary tree structure to obtain a coherent and consistent entire coronary tree vascular structure.

[0083] Specifically, after reconstructing the coronary tree structure, it is necessary to analyze and optimize it. The analysis process mainly involves evaluating the authenticity, curvature and connection relationship of each branch. It is necessary to ensure that the shape of each vascular branch is consistent with physiological characteristics and that the connection relationship between branches conforms to anatomical principles. To this end, quantitative analysis can be performed using metrics such as curvature, length and relative position. Optimization ensures the coherence and consistency of the entire coronary tree by adjusting and correcting unreasonable parts of the structure;

[0084] To effectively improve the segmentation accuracy of the coronary tree, this embodiment further repeats some of the steps in the previous stage. Specifically, the image processing model can be reapplied to perform vessel repair and coronary tree supplementation on the supplemented coronary tree structure, and continuous iterative optimization can be used to gradually achieve complete segmentation of the entire coronary tree.

[0085] The reconstruction module is used to reconstruct multiple planes of the blood vessels according to the coronary vessel region to obtain a two-dimensional coronary artery tree result.

[0086] Specifically, a multi-planar reconstruction algorithm is used to map the extracted coronary artery tree structure onto a two-dimensional plane. This technology generates high-quality coronary vascular visualization images that clearly display the complete coronary vascular system.

[0087] Furthermore, the reconstruction module includes:

[0088] a two-dimensional mapping unit, configured to map the extracted coronary tree vessel structure onto a two-dimensional plane using a multi-planar reconstruction algorithm;

[0089] Specifically, the multi-planar reconstruction algorithm is applied. It maps the extracted coronary artery tree vascular structure onto a two-dimensional plane. By analyzing the three-dimensional data and selecting multiple slice planes, vascular information is obtained from different perspectives and depths. These slices are then synthesized to effectively construct the overall morphology and spatial relationships of the vessels, ensuring that every detail is fully captured.

[0090] an image visualization unit, configured to generate a coronary vessel visualization image from the coronary tree vessel structure;

[0091] Specifically, high-quality visualization images are generated using the reconstructed data. This process involves image processing techniques, including anti-aliasing, color enhancement, and contrast adjustment, to improve visualization.

[0092] The coronary vascular display unit is used to clearly display the complete coronary vascular system.

[0093] Specifically, the images generated by this technology clearly display the entire coronary artery system, including not only the main vessels but also all branches and smaller vessels, making the entire system's anatomy clear at a glance. This visualization provides important reference for subsequent diagnosis, treatment plan formulation, and surgical planning, improving the accuracy and efficiency of clinical decision-making.

[0094] Example 2

[0095] This embodiment further discloses a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the system described in the first embodiment.

[0096] Example 3

[0097] This embodiment further discloses a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the system described in the first embodiment are implemented.

[0098] Example 4

[0099] This embodiment also discloses a computer program product, including a computer program, which implements the steps of the system described in the first embodiment when executed by a processor.

[0100] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A complete automatic segmentation system for the entire coronary artery tree and vascular centerline extraction based on CCTA images, characterized by: include: Positioning module, used to locate and extract the heart based on CCTA images; An extraction module is used to perform initial extraction of the coronary artery tree segmentation based on the CCTA image after cardiac positioning to obtain the coronary artery tree; The repair module is used to extract the vascular centerline based on the extracted coronary artery tree and extend the centerline at the breakpoint to connect the broken parts, or further extend the incomplete centerline to the end of the vessel to obtain the repaired centerline; a supplementation module, configured to supplement the coronary tree structure based on the repair centerline, and repeat the steps of vessel repair and coronary tree supplementation to achieve complete segmentation of the entire coronary tree and obtain the coronary vessel region; A reconstruction module, configured to reconstruct multiple planes of the coronary vessels according to the coronary vessel region to obtain a two-dimensional coronary artery tree result; Wherein, the positioning module, the extraction module, the repair module, the supplement module, and the reconstruction module are connected in sequence.

2. The system according to claim 1, wherein: The positioning module includes: The heart annotation unit is used to annotate the heart position and generate a 3D bounding box as the training and test sets of the model; The first construction unit is used to build a positioning model based on ResNet-18; a model positioning unit, configured to train the positioning model using the training set to obtain a trained positioning model, and input the test set into the trained positioning model to obtain a coronary artery tree positioning image; Wherein, the heart labeling unit, the first construction unit, and the model positioning unit are respectively connected to the extraction module.

3. The system according to claim 1, wherein: The extraction module includes: a data set acquisition unit, configured to divide the coronary tree localization image into a training set and a test set; The second construction unit is used to build a deep learning model based on 3D-Unet; a model extraction unit, configured to train the deep learning model using the training set to obtain a trained deep learning model, and input the test set into the trained deep learning model to obtain a preliminary extraction result of the coronary artery tree; Wherein, the data set acquisition unit, the second construction unit, and the model extraction unit are respectively connected to the repair module.

4. The system according to claim 1, wherein: The repair module includes: An extraction unit, configured to extract the centerline of the coronary artery using a skeletonization function; An image processing unit, which resamples the image and crops the image blocks into a specified size; a model repair unit, configured to construct a neural network model, and repair and improve the blood vessel centerline through the neural network model to obtain a repaired centerline; Wherein, the extraction unit, the image processing unit, and the model repairing unit are respectively connected to the supplementing module.

5. The system according to claim 1, wherein: The supplementary module includes: an information combining unit, configured to combine the repair centerline with the boundary information of the blood vessel to reconstruct the coronary artery tree structure; Structural optimization unit, used to analyze and optimize the reconstructed coronary tree structure to obtain a coherent and consistent vascular structure of the entire coronary tree; Wherein, the information combining unit and the structure optimizing unit are respectively connected to the reconstruction module.

6. The system according to claim 1, wherein: The reconstruction module includes: a two-dimensional mapping unit, configured to map the extracted coronary tree vessel structure onto a two-dimensional plane using a multi-planar reconstruction algorithm; an image visualization unit, configured to generate a coronary vessel visualization image from the coronary tree vessel structure; The coronary vascular display unit is used to clearly display the complete coronary vascular system.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory, wherein: The processor executes the computer program to implement the steps of the system according to any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the system according to any one of claims 1 to 6 are implemented.

9. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the system according to any one of claims 1 to 6 are implemented.

Images