Coronary artery evaluation and prediction method, device, equipment, medium and product

Abstract

The invention provides a coronary artery evaluation and prediction method and device, equipment, a medium and a product, and the method comprises the steps: constructing a plurality of image sub-blocks and a hierarchical topology sensing network according to a preset coronary artery marking region; according to the image sub-blocks and the hierarchical topology sensing network, multilevel topology information extraction of the coronary artery structure is determined; determining at least one target narrow coronary artery from the multi-level topological information; for each target narrow coronary artery, acquiring a plurality of time sequence images of the target narrow coronary artery; and determining the variation trend of the plaque component of each target narrow coronary artery according to the plurality of time sequence images of each target narrow coronary artery. According to the method, multi-level topological information of the coronary artery can be determined at least by constructing image sub-blocks and a hierarchical topological sensing network, and the change trend of plaque components of the narrow coronary artery is determined according to the multi-level topological information of the coronary artery and a plurality of time sequence images.

Description

Technical Field

[0001] The present application relates to the field of biomedical technologies, and in particular, to a method, device, equipment, medium and product for evaluating and predicting coronary arteries. Background Art

[0002] Coronary Heart Disease (CHD) is a major public health problem endangering human health. Approximately 9.6 million people die worldwide each year from acute myocardial infarction and heart failure caused by CHD. The situation is even more severe in China. According to the "China Cardiovascular Disease Report 2023", there are currently 11 million CHD patients in China, and more than 700,000 people die from acute myocardial infarction and sudden cardiac death each year, showing a rapid upward trend. Although the clinical diagnosis and treatment strategies for CHD are constantly advancing, due to the characteristics of the progression of coronary plaque lesions, such as symptom concealment, sudden rupture, large heterogeneity, and unclear prodromal symptoms, it is difficult to accurately predict and timely and effectively intervene in the prognosis evaluation clinically, resulting in a still high incidence and recurrence rate of acute cardiovascular events. Coronary CTA is the most commonly used non-invasive imaging screening method for CHD, which can clarify the branch direction of coronary arteries and three-dimensionally reconstruct coronary blood vessels, avoid underestimating the stenosis degree due to eccentric plaques, and can also analyze the nature and calcification degree of plaques based on CT density values. Especially in combination with artificial intelligence image analysis technology, using the method of computational fluid dynamics based on CT images to simulate the pressure and fractional flow reserve in the coronary arteries can be used for outpatient screening of lesions that require further interventional treatment. Most previous examinations were based on single coronary CTA to analyze the structure and nature of plaques, only observing the plaque size and vulnerability at a certain time point, and it was difficult to predict the future changes and outcomes of plaques. By comparing and analyzing the results of at least two or more coronary CTA at the follow-up and baseline time points, analyzing the dynamic change process of plaque burden, nature and vulnerable imaging features, and predicting the risk of plaques turning into vulnerable types and rupturing to cause acute myocardial infarction within the next 3 to 5 years is crucial for strengthening clinical management and timely intervention to prevent acute cardiovascular adverse events. Summary of the Invention

[0003] In view of this, the purpose of the present application is to provide a method, device, equipment, medium and product for evaluating and predicting coronary arteries, which can at least determine the multi-level topological information of coronary arteries by constructing image sub-blocks and a hierarchical topology perception network, and determine the change trend of plaque components in stenotic coronary arteries according to the multi-level topological information of coronary arteries and multiple temporal images.

[0004] An embodiment of the present application provides a coronary artery evaluation and prediction method, including: constructing a plurality of image sub-blocks and a hierarchical topology-aware network according to a pre-set coronary artery annotation area, where the plurality of image sub-blocks cover the coronary artery structure; determining the extraction of multi-level topology information of the coronary artery structure according to the image sub-blocks and the pre-constructed hierarchical topology-aware network, where the multi-level topology information includes segmentation information of the coronary artery structure, centerline information of the coronary artery structure, and key point information of the coronary artery structure; determining at least one target stenotic coronary artery from the multi-level topology information; for each target stenotic coronary artery, obtaining a plurality of temporal images of the target stenotic coronary artery; and determining the change trend of the plaque components of each target stenotic coronary artery according to the plurality of temporal images of each target stenotic coronary artery.

[0005] Optionally, the plurality of image sub-blocks and the hierarchical topology-aware network are constructed through the following steps: extracting the latent space features of the vascular structure from the coronary artery image through an encoder, where the latent space features are used to comprehensively describe the structure, centerline, and key point information of the coronary artery in the form of a low-dimensional vector; where the encoder includes three sub-encoders and a multi-task shared encoder, and the three sub-encoders are respectively used to extract the structure information of the coronary artery, the centerline information of the coronary artery, and the key point information of the coronary artery centerline.

[0006] Optionally, the connectivity branches of the coronary artery are determined through the following steps: determining the multi-channel feature vectors between the central point and a plurality of neighborhood points in the target area based on the image sub-blocks and the hierarchical topology-aware network; determining the multi-channel connectivity probability map specifically quantifying the connectivity according to the multi-channel feature vectors; and fusing the multi-channel probability map and the central points of the corresponding areas according to the connectivity consistency to form a binary map, thereby forming the final segmentation result.

[0007] Optionally, the method further includes: based on the pre-constructed cross-attention interaction module, propagating the dense representation of the neighborhood connectivity graph to the sparse representation of the vascular centerline or vascular key points to strengthen the dense representation information.

[0008] Optionally, for the branches of the vascular key points and the vascular centerline, the weighted Hausdorff distance and the mean square error are used as loss functions; where, for the neighborhood connectivity branches, the cross-entropy and the DICE coefficient are used as loss functions; where, an adversarial training strategy is adopted to evaluate the similarity between the prediction result and the label from the perspective of the data distribution; where, for the segmentation / label result of the blood vessel, the blood vessel skeleton information is extracted based on the iterative refinement algorithm and input into the discriminator, and the weight update loss function is performed through backpropagation, introducing adversarial regularization constraints for the optimization of the hierarchical topology learning network.

[0009] Optionally, the steps of determining the change trend of the plaque components of each target stenotic coronary artery based on multiple temporal images of each target stenotic coronary artery include: respectively taking the necrotic core, calcification, and fibrous components within the plaque causing stenosis as the segmentation targets, and establishing a multi-target segmentation network model with residual modules as the core; based on the multi-target segmentation network model, respectively performing regional division and extraction of the necrotic core, calcification, and fibrous components on the multiple temporal images; calculating the volume curves of the necrotic core, calcification, and fibrous components, analyzing and predicting the change trend of at least one plaque component, and determining the change trend of the plaque components of each target stenotic coronary artery.

[0010] In a second aspect, the present application further provides an evaluation and prediction device for coronary arteries, including: a topology-aware network construction module, configured to construct multiple image sub-blocks and a hierarchical topology-aware network according to a pre-set coronary artery annotation area, and the multiple image sub-blocks cover the coronary artery structure;

[0011] A multi-level topology information extraction module, configured to determine the extraction of multi-level topology information of the coronary artery structure according to the image sub-blocks and the hierarchical topology-aware network, and the multi-level topology information includes segmentation information of the coronary artery structure, centerline information of the coronary artery structure, and key point information of the coronary artery structure;

[0012] A stenotic coronary artery determination module, configured to determine at least one target stenotic coronary artery from the multi-level topology information;

[0013] For each target stenotic coronary artery, obtain multiple temporal images of the target stenotic coronary artery;

[0014] According to the multiple temporal images of each target stenotic coronary artery, determine the change trend of the plaque components of each target stenotic coronary artery.

[0015] In a third aspect, the present application further provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0016] Construct multiple image sub-blocks and a hierarchical topology-aware network according to a pre-set coronary artery annotation area, and the multiple image sub-blocks cover the coronary artery structure;

[0017] Determine the extraction of multi-level topology information of the coronary artery structure according to the image sub-blocks and the hierarchical topology-aware network, and the multi-level topology information includes segmentation information of the coronary artery structure, centerline information of the coronary artery structure, and key point information of the coronary artery structure;

[0018] Determine at least one target stenotic coronary artery from the multi-level topology information;

[0019] For each target stenotic coronary artery, obtain multiple temporal images of the target stenotic coronary artery;

[0020] Determine the changing trend of the plaque components of each target stenotic coronary artery based on multiple temporal images of each target stenotic coronary artery.

[0021] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0022] Construct multiple image sub-blocks and a hierarchical topology-aware network according to a preset coronary artery annotation area, and the multiple image sub-blocks cover the coronary artery structure;

[0023] Determine the extraction of multi-level topology information of the coronary artery structure according to the image sub-blocks and the hierarchical topology-aware network, and the multi-level topology information includes segmentation information of the coronary artery structure, centerline information of the coronary artery structure, and key point information of the coronary artery structure;

[0024] Determine at least one target stenotic coronary artery from the multi-level topology information;

[0025] For each target stenotic coronary artery, obtain multiple temporal images of the target stenotic coronary artery;

[0026] Determine the changing trend of the plaque components of each target stenotic coronary artery based on multiple temporal images of each target stenotic coronary artery.

[0027] In a fifth aspect, the present application also provides a computer program product, including a computer program. When the computer program is executed by a processor, the following steps are implemented:

[0028] Construct multiple image sub-blocks and a hierarchical topology-aware network according to a preset coronary artery annotation area, and the multiple image sub-blocks cover the coronary artery structure;

[0029] Determine the extraction of multi-level topology information of the coronary artery structure according to the image sub-blocks and the hierarchical topology-aware network, and the multi-level topology information includes segmentation information of the coronary artery structure, centerline information of the coronary artery structure, and key point information of the coronary artery structure;

[0030] Determine at least one target stenotic coronary artery from the multi-level topology information;

[0031] For each target stenotic coronary artery, obtain multiple temporal images of the target stenotic coronary artery;

[0032] Determine the changing trend of the plaque components of each target stenotic coronary artery based on multiple temporal images of each target stenotic coronary artery.

[0033] The coronary artery evaluation and prediction method, device, equipment, medium and product provided by the embodiments of the present application can at least determine the multi-level topological information of the coronary artery by constructing image sub-blocks and a hierarchical topological perception network, and determine the change trend of the plaque composition of the stenotic coronary artery according to the multi-level topological information of the coronary artery and multiple temporal images.

[0034] To make the above objects, features, and advantages of the present application more obvious and understandable, the following specifically enumerates preferred embodiments and, in conjunction with the accompanying drawings, provides a detailed description as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] To more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required in the embodiments. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative efforts.

[0036] Figure 1 Schematic flowchart of a coronary artery evaluation and prediction method provided by an embodiment of the present application;

[0037] Figure 2 Schematic comparison diagram of plaque burden and composition based on quantitative analysis of follow-up and baseline coronary CTA images provided by an embodiment of the present application;

[0038] Figure 3 Schematic diagram of dynamic changes in hemodynamics compared between follow-up and baseline coronary CTA provided by an embodiment of the present application;

[0039] Figure 4 Structural block diagram of the recognition of a molecular odor provided by an embodiment of the present application;

[0040] Figure 5 Internal structure diagram of a computer device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some, rather than all, of the embodiments of the present application. Usually, the components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application claimed, but merely represents selected embodiments of the present application. Based on the embodiments of the present application, every other embodiment obtained by those of ordinary skill in the art without creative efforts belongs to the scope of protection of the present application.

[0042] First, the applicable application scenarios of this application are introduced. This application can be applied to the field of biomedical technology.

[0043] Coronary Heart Disease (CHD) is a major public health problem threatening human health. Approximately 9.6 million people worldwide die each year from acute myocardial infarction and heart failure caused by coronary heart disease. The situation in our country is even more severe. According to the "China Cardiovascular Disease Report 2023", there are currently 11 million coronary heart disease patients in our country, and more than 700,000 people die from acute myocardial infarction and sudden cardiac death every year, showing a rapid upward trend. Although the clinical diagnosis and treatment strategies for coronary heart disease have been continuously improving, due to the characteristics of the progression of coronary plaque lesions, such as symptom concealment, sudden rupture, large heterogeneity, and unclear prodromal symptoms, it is difficult to accurately predict and timely and effectively intervene in the prognosis assessment clinically, resulting in a still high incidence and recurrence rate of acute cardiovascular events. Coronary CTA is the most commonly used non-invasive imaging screening method for coronary heart disease, which can clarify the branch direction of the coronary arteries and three-dimensionally reconstruct the coronary blood vessels, avoid underestimating the stenosis degree due to eccentric plaques, and can also analyze the nature and calcification degree of plaques based on CT density values. Especially in combination with artificial intelligence image analysis technology, using the method of computational fluid dynamics based on CT images to simulate the pressure and fractional flow reserve in the coronary arteries can be used for outpatient screening of lesions that require further interventional treatment. Most previous examinations were based on single coronary CTA to analyze the structure and nature of plaques, only observing the plaque size and vulnerability at a certain time point, and it was difficult to predict the future changes and outcomes of plaques. By comparing and analyzing the results of at least two or more coronary CTAs at the follow-up and baseline time points, analyzing the dynamic change process of plaque burden, nature, and vulnerable imaging features, and predicting the risk of plaques turning into vulnerable types and rupturing to cause acute myocardial infarction within the next 3-5 years is crucial for clinically strengthening management and timely intervention to prevent acute cardiovascular adverse events.

[0044] Based on this, the embodiments of this application provide a method, device, equipment, medium, and product for evaluating and predicting coronary arteries, which can at least determine the multi-level topological information of coronary arteries by constructing image sub-blocks and a hierarchical topological perception network, and determine the change trend of the plaque components of stenotic coronary arteries according to the multi-level topological information of coronary arteries and multiple temporal images.

[0045] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of a method for evaluating and predicting coronary arteries provided by the embodiments of this application. As shown in Figure 1 , the method for evaluating and predicting coronary arteries provided by the embodiments of this application includes:

[0046] S101. Construct multiple image sub - blocks and a hierarchical topology - aware network according to the pre - set coronary artery annotation area.

[0047] Among them, the multiple image sub - blocks cover the coronary artery structure.

[0048] Specifically, the multiple image sub - blocks and the hierarchical topology - aware network can be constructed through the following steps: Extract the latent space features of the vascular structure from the coronary artery image through an encoder. The latent space features are used to comprehensively describe the structure, centerline, and key point information of the structure containing the coronary artery in the form of a low - dimensional vector. Among them, the encoder includes three sub - encoders and a multi - task shared encoder. The three sub - encoders are respectively used to extract the structural information of the coronary artery, the centerline information of the coronary artery, and the key point information of the coronary artery centerline.

[0049] S102. Determine the multi - level topology information extraction of the coronary artery structure according to the image sub - blocks and the hierarchical topology - aware network. The multi - level topology information includes the segmentation information of the coronary artery structure, the centerline information of the coronary artery structure, and the key point information of the coronary artery structure.

[0050] Among them, the connectivity branches of the coronary artery can be determined through the following steps: Based on the image sub - blocks and the hierarchical topology - aware network, determine the multi - channel feature vectors between the center point of the target area and multiple neighborhood points; According to the multi - channel feature vectors, determine the multi - channel connectivity probability map for specifically quantifying connectivity; According to the multi - channel probability map and the connectivity consistency of the center points of the corresponding areas, fuse them into a binary map to form the final segmentation result.

[0051] Optionally, the method further includes: Based on the pre - constructed cross - attention interaction module, propagate the dense representation of the neighborhood connectivity graph to the sparse representation of the vascular centerline or vascular key points to strengthen the dense representation information.

[0052] Among them, for the branches of the vascular key points and the vascular centerline, the weighted Hausdorff distance and the mean square error are used as the loss functions.

[0053] Among them, for the neighborhood connectivity branch, the cross - entropy and the DICE coefficient are used as the loss functions.

[0054] Among them, an adversarial training strategy is adopted to evaluate the similarity between the prediction result and the label from the perspective of the data distribution.

[0055] Among them, for the segmentation / label result of the blood vessel, based on the iterative refinement algorithm, extract the blood vessel skeleton information, input it into the discriminator, and update the weights through backpropagation for the loss function, introducing adversarial regularization constraints for the optimization of the hierarchical topology learning network.

[0056] As an example, multiple image sub - blocks can be constructed based on the pre - established coronary artery annotation regions, and a hierarchical topology - aware network is constructed, which consists of a multi - task shared encoder and three decoders.

[0057] Among them, the encoder extracts the latent space features of the vascular structure from the coronary artery image, and the three encoders respectively extract vascular key points (end points and bifurcation points), centerlines, and vascular structures. In the key point extraction branch, considering the uncertainty of the number of key points in different vascular structures, an improved weighted Hausdorff distance function is constructed to describe the similarity between the predicted and labeled key points.

[0058] In the centerline extraction branch, considering that the centerline perpendicular to the vascular cross - section should contain a single pixel, a small positional deviation may have a greater impact on the centerline prediction accuracy. A three - dimensional heatmap regression method is proposed. An adaptive Gaussian kernel in three dimensions is used to construct the centerline heatmap unit, and the kernel width is approximately equal to the vascular radius, thereby adaptively and dynamically adjusting the shape of the heatmap.

[0059] This centerline prediction module ensures the consistency of the centerline and the continuity of the vascular segments.

[0060] In the neighborhood connectivity branch, it is proposed to describe the vascular context information according to the neighborhood connectivity relationship.

[0061] Exemplarily, the vascular continuity assessment can be transformed into the connectivity relationship of a cubic neighborhood. For a 3×3×3 cubic region, the set of point pairs composed of the central point and its surrounding 26 points, if there is a connection relationship label, it is 1, otherwise it is 0. Thus, for each voxel, a 27 - channel feature vector can be obtained, thereby generating a 27 - channel probability map for quantifying connectivity.

[0062] Among them, in the inference stage, the 27 - channel probability map is fused into a binary map according to the connectivity consistency of the voxels with each other, forming the final segmentation result. Considering the lack of semantic information fusion between the above - mentioned topological branches, the overall constraint and representation learning of blood vessels cannot be carried out. It is proposed to construct a bottom - up cross - attention interaction module to propagate the dense representation (neighborhood connectivity graph) to the sparse representation (centerline or key points), strengthen the dense representation information, and improve the accuracy of the sparse representation.

[0063] In the corresponding network design, this module is added to the intermediate network layer of the sparse feature branch, and the multi - head attention mechanism provides supplementary information.

[0064] In the construction of the loss function, for the key point and centerline branches, the weighted Hausdorff distance and mean squared error are used as the loss functions; for the neighborhood connectivity branch, the cross - entropy and DICE coefficient are used as the loss functions.

[0065] To further strengthen the topological information of the constrained segmentation results, an adversarial training strategy is additionally adopted to evaluate the similarity between the prediction results and the labels from the perspective of data distribution.

[0066] For the segmentation / label results of blood vessels, based on the iterative refinement algorithm, the blood vessel skeleton information is extracted and input into the discriminator. Through backpropagation, the weight update loss function is used to introduce adversarial regularization constraints for the optimization of the hierarchical topology learning network, further improving the perception of neighborhood connectivity for blood vessel topological information.

[0067] In this way, the multi-level topological information extraction of the coronary artery structure can be realized, including information such as the segmentation of the coronary artery structure, the centerline of the coronary artery structure, and key points.

[0068] S103. Determine at least one target stenotic coronary artery from the multi-level topological information.

[0069] Specifically, through the segmentation of the coronary artery structure, the coronary artery stenosis information can be calculated, and the branches with stenosis can be selected as the analysis target stenotic coronary arteries.

[0070] S104. For each target stenotic coronary artery, obtain multiple temporal images of the target stenotic coronary artery.

[0071] Specifically, for the coronary CTA images of multiple temporal images, a temporal image sub-block centered on the stenotic area can be constructed.

[0072] S105. According to the multiple temporal images of each target stenotic coronary artery, determine the change trend of the plaque components of each target stenotic coronary artery.

[0073] Specifically, the steps of determining the change trend of the plaque components of each target stenotic coronary artery according to the multiple temporal images of each target stenotic coronary artery include: respectively taking the necrotic core, calcification, and fibrous components in the plaque causing stenosis as the segmentation targets, and establishing a multi-target segmentation network model with a residual module as the core; based on the multi-target segmentation network model, respectively perform regional division and extraction of the necrotic core, calcification, and fibrous components on the multiple temporal images; calculate the volume curves of the necrotic core, calcification, and fibrous components, analyze and predict the change trend of at least one plaque component, and determine the change trend of the plaque components of each target stenotic coronary artery.

[0074] As an example, the necrotic core, calcification, and fibrous components in the plaque causing stenosis can be respectively taken as the segmentation targets, and a multi-target segmentation network model with a residual module as the core can be established. The temporal image sub-blocks are respectively subjected to regional division and extraction of the necrotic core, calcification, and fibrous components. Then, by calculating the volume curves of the three regions, the change trend of the plaque components is analyzed and predicted.

[0075] Exemplarily, the proportion changes of the necrotic core, calcification, and fibrous components within the plaque; necrotic core (CT HU value: -30 to 30 HU), fiber (CT HU value: 31 to 350 HU), calcification (CT HU value > 351 HU)( Figure 1 ).

[0076] The quantitative change in the number of newly emerging vulnerable plaque features (including napkin-ring sign, low-attenuation plaque, positive remodeling, and punctate calcification) is defined as follows: Positive remodeling is defined as the external vessel diameter (long dashed line) being 10% larger than the average diameter of the segments proximal (short dashed line) and distal to the plaque.

[0077] Lipid-rich plaque is defined as the focal central area of the plaque with an attenuation density < 30 HU; punctate calcification is defined as the intracoronary wall calcification on CT with a density > 130 HU and a diameter < 3 mm; the "napkin-ring" sign is defined as the central area of a low-attenuation plaque with a high-attenuation rim. An increase in the above plaque features is observed.

[0078] Please refer to Figure 2 and Figure 3 , Figure 2 The first baseline coronary CTA image analysis results (including plaque burden, necrotic core, calcification, fibrous components, etc.) are compared with the coronary CTA image analysis results at follow-up to obtain the dynamic changes in the results as shown in Figure 2 . Figure 3 The fractional flow reserve (FFR) value of the left anterior descending (LAD) artery before the first baseline coronary CTA image analysis is 0.83, and is compared with the FFR value of the LAD at follow-up coronary CTA image analysis, which is 0.82. The fractional flow reserve has decreased, demonstrating that this application uses deep learning (large dense neural network and lightweight neural network) to achieve the registration and fusion of quantitative parameters such as coronary stenosis degree, structural features, biomechanical changes, and CT fractional flow reserve, and establishes a multi-dimensional evaluation system reflecting coronary structure and myocardial function.

[0079] The coronary evaluation and prediction method, device, equipment, medium, and product provided by the embodiments of this application can at least determine the multi-level topological information of the coronary artery by constructing image sub-blocks and a hierarchical topological perception network, and determine the change trend of the plaque components of the stenotic coronary artery based on the multi-level topological information of the coronary artery and multiple temporal images.

[0080] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0081] Based on the same inventive concept, an embodiment of the present application further provides an identification and control device for sub-odors for implementing the coronary artery evaluation and prediction method described above. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the coronary artery evaluation and prediction device provided below can refer to the limitations of the typhoon problem answering method in the above text, and will not be repeated here.

[0082] Please refer to Figure 4 , in an exemplary embodiment, a coronary artery evaluation and prediction device is provided, including:

[0083] A topology-aware network construction module for constructing a plurality of image sub-blocks and a hierarchical topology-aware network according to a pre-set coronary artery annotation area, and the plurality of image sub-blocks cover the coronary artery structure;

[0084] A multi-level topology information extraction module for determining the multi-level topology information extraction of the coronary artery structure according to the image sub-blocks and the hierarchical topology-aware network, and the multi-level topology information includes segmentation information of the coronary artery structure, centerline information of the coronary artery structure, and key point information of the coronary artery structure;

[0085] A narrow coronary artery determination module for determining at least one target narrow coronary artery from the multi-level topology information;

[0086] For each target narrow coronary artery, obtain a plurality of temporal images of the target narrow coronary artery;

[0087] According to the plurality of temporal images of each target narrow coronary artery, determine the change trend of the plaque components of each target narrow coronary artery.

[0088] Each module in the above-mentioned molecular odor-based recognition device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above modules can be embedded in the processor of a computer device in hardware form or be independent of it, or be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.

[0089] In an exemplary embodiment, a computer device is provided. The computer device can be a terminal. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface, the display unit, and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used for the processor to exchange information with external devices. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a method for coronary artery evaluation and prediction. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0090] Those skilled in the art can understand that Figure 5 the structure shown in

[0091] In an exemplary embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the following steps are implemented:

[0092] Construct a plurality of image sub-blocks and a hierarchical topology perception network according to a pre-set coronary artery annotation area, and the plurality of image sub-blocks cover the coronary artery structure;

[0093] Determine the multi-level topological information extraction of the coronary artery structure based on the image sub-blocks and the hierarchical topology-aware network, where the multi-level topological information includes the segmentation information of the coronary artery structure, the centerline information of the coronary artery structure, and the key point information of the coronary artery structure;

[0094] Determine at least one target stenotic coronary artery from the multi-level topological information;

[0095] For each target stenotic coronary artery, obtain multiple temporal images of the target stenotic coronary artery;

[0096] Determine the change trend of the plaque composition of each target stenotic coronary artery according to the multiple temporal images of each target stenotic coronary artery.

[0097] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0098] Construct multiple image sub-blocks and a hierarchical topology-aware network according to the pre-set coronary artery annotation area, where the multiple image sub-blocks cover the coronary artery structure;

[0099] Determine the multi-level topological information extraction of the coronary artery structure based on the image sub-blocks and the hierarchical topology-aware network, where the multi-level topological information includes the segmentation information of the coronary artery structure, the centerline information of the coronary artery structure, and the key point information of the coronary artery structure;

[0100] Determine at least one target stenotic coronary artery from the multi-level topological information;

[0101] For each target stenotic coronary artery, obtain multiple temporal images of the target stenotic coronary artery;

[0102] Determine the change trend of the plaque composition of each target stenotic coronary artery according to the multiple temporal images of each target stenotic coronary artery.

[0103] In one embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the following steps are implemented:

[0104] Construct multiple image sub-blocks and a hierarchical topology-aware network according to the pre-set coronary artery annotation area, where the multiple image sub-blocks cover the coronary artery structure;

[0105] Determine the multi-level topological information extraction of the coronary artery structure based on the image sub-blocks and the hierarchical topology-aware network, where the multi-level topological information includes the segmentation information of the coronary artery structure, the centerline information of the coronary artery structure, and the key point information of the coronary artery structure;

[0106] Determine at least one target stenotic coronary artery from the multi-level topological information;

[0107] For each target stenotic coronary artery, a plurality of temporal images of the target stenotic coronary artery are acquired;

[0108] Based on the plurality of temporal images of each target stenotic coronary artery, the change trend of the plaque components of each target stenotic coronary artery is determined.

[0109] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0110] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this application.

[0111] The above-described embodiments merely represent several implementation manners of this application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of this application. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of this application shall be subject to the appended claims.

Claims

1. A method for coronary artery evaluation and prediction, characterized in that The method includes: Constructing a plurality of image sub - blocks and a hierarchical topology - aware network according to a pre - set coronary artery annotation area, where the plurality of image sub - blocks cover the coronary artery structure; Determining the extraction of multi - level topology information of the coronary artery structure according to the image sub - blocks and the pre - constructed hierarchical topology - aware network, where the multi - level topology information includes segmentation information of the coronary artery structure, centerline information of the coronary artery structure, and key - point information of the coronary artery structure; Determining at least one target stenotic coronary artery from the multi - level topology information; For each target stenotic coronary artery, obtaining a plurality of temporal images of the target stenotic coronary artery; Determining the change trend of the plaque components of each target stenotic coronary artery according to the plurality of temporal images of each target stenotic coronary artery.

2. The method according to claim 1, characterized in that, Constructing a plurality of image sub - blocks and a hierarchical topology - aware network through the following steps: Extracting the latent space features of the vascular structure from the coronary artery image through an encoder, where the latent space features are used to comprehensively describe the structure, centerline, and key - point information of the coronary artery in the form of a low - dimensional vector; Among them, the encoder includes three sub - encoders and a multi - task shared encoder, and the three sub - encoders are respectively used to extract the structure information of the coronary artery, the centerline information of the coronary artery, and the key - point information of the coronary artery centerline.

3. The method according to claim 1, characterized in that, Determining the connected branches of the coronary artery through the following steps: Based on the image sub - blocks and the hierarchical topology - aware network, determining the multi - channel feature vectors between the central point and multiple neighborhood points of the target area; Determining the multi - channel connectivity probability map for specifically quantifying connectivity according to the multi - channel feature vectors; Fusing the multi - channel probability map and the connectivity consistency of the central points of the corresponding areas into a binary map to form the final segmentation result.

4. The method according to claim 1, wherein The method further includes: Based on the pre - constructed cross - attention interaction module, propagating the dense representation of the neighborhood connectivity graph to the sparse representation of the vascular centerline or vascular key - points to strengthen the dense representation information.

5. The method according to claim 1, wherein For the branches of the vascular key - points and vascular centerline, using the weighted Hausdorff distance and mean square error as the loss function; Among them, for the neighborhood connectivity branches, using cross - entropy and DICE coefficient as the loss function; Among them, adopting an adversarial training strategy to evaluate the similarity between the prediction result and the label from the perspective of data distribution; Among them, for the segmentation / label result of the blood vessel, based on the iterative refinement algorithm, extracting the blood vessel skeleton information and inputting it into the discriminator, and updating the weights of the loss function through back - propagation, introducing adversarial regularization constraints for the optimization of the hierarchical topology learning network.

6. The method according to claim 1, characterized in that, The steps of determining the change trend of the plaque components of each target stenotic coronary artery according to the plurality of temporal images of each target stenotic coronary artery include: Respectively taking the necrotic core, calcification, and fibrotic components in the plaque causing stenosis as the segmentation targets, and establishing a multi - target segmentation network model with a residual module as the core; Based on the multi - target segmentation network model, respectively performing regional division and extraction of the necrotic core, calcification, and fibrotic components on the plurality of temporal images; Calculating the volume curves of the necrotic core, calcification, and fibrotic components, analyzing and predicting the change trend of at least one plaque component, and determining the change trend of the plaque components of each target stenotic coronary artery.

7. An evaluation and prediction device for coronary arteries, characterized in that The device includes: A topology-aware network construction module, configured to construct a plurality of image sub-blocks and a hierarchical topology-aware network according to a preset coronary artery annotation region, wherein the plurality of image sub-blocks cover the coronary artery structure; A multi-level topology information extraction module, configured to determine the extraction of multi-level topology information of the coronary artery structure according to the image sub-blocks and the hierarchical topology-aware network, where the multi-level topology information includes segmentation information of the coronary artery structure, centerline information of the coronary artery structure, and key point information of the coronary artery structure; A stenosis coronary artery determination module, configured to determine at least one target stenosis coronary artery from the multi-level topology information; A temporal image acquisition module, configured to acquire a plurality of temporal images of each target stenosis coronary artery; A change trend prediction module, configured to determine the change trend of the plaque components of each target stenosis coronary artery according to the plurality of temporal images of each target stenosis coronary artery.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

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

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

Images