Cardiovascular segmentation method and system based on data-driven iterative learning

Abstract

The invention provides a cardiovascular segmentation method and system based on data-driven iterative learning, and relates to the technical field of image segmentation, and the method comprises the steps: inputting a to-be-segmented cardiovascular image into a learned segmentation model, and obtaining a segmentation result; wherein the segmentation model adopts a data-driven iterative learning method, and in each iteration, firstly, based on an image set trained this time, a non-edge set and an edge set are constructed according to the fuzzy degree and the edge pixel proportion of image blocks; secondly, training a segmentation model by using a non-edge set under the constraint of a segmentation loss function, continuing training by using an edge set under the coordination of a distribution alignment loss function on the basis of model parameters after training of the non-edge set, and finally, using the trained model parameters of the current round for parameter initialization of the next round; according to the method, efficient and high-precision cardiovascular image segmentation is trained, so that early diagnosis, treatment planning and curative effect evaluation of cardiovascular diseases are promoted, and meanwhile, the method adapts to a clinical environment where computing resources are limited.

Description

Technical Field

[0001] The present disclosure relates to the technical field of image segmentation, and in particular to a data-driven iterative learning cardiovascular segmentation method and system. Background Art

[0002] With the rapid development of medical imaging technology, a variety of imaging modalities have provided rich data support for the diagnosis and research of cardiovascular diseases, and have become an indispensable tool in clinical diagnosis and disease research; for example, in the diagnosis of coronary heart disease, CT angiography (CTA) can accurately detect the degree of stenosis of the coronary arteries, providing an important basis for the selection of treatment options. However, due to the complexity of the heart structure, such as the different morphologies and functions of different chambers of the heart, and the interweaving of tissues such as the myocardium and blood vessels, the segmentation of cardiovascular structures in CT images has become extremely challenging; in addition, there are differences in the contrast between different tissues in CT images, such as the different contrast between the myocardium and the vascular wall, and the blood in the heart cavity, which may make it difficult to accurately define the boundaries between cardiovascular structures and surrounding tissues during the segmentation process.

[0003] In recent years, cardiovascular segmentation research based on traditional methods has achieved certain results; for example, the method based on region growing (such as application publication number: CN117115186A) locates the cardiovascular system by gradually growing regions in the image according to criteria such as similarity, but this method is sensitive to seed point selection, easily affected by complex image features and noise, and the stability of the segmentation results is poor; for another example, a cardiovascular detection system and method based on image analysis combined with electrocardiogram information (application publication number: CN118212237A) locates the cardiovascular position with the help of electrocardiogram information, but the matching of electrocardiogram and image information may not be accurate enough, affecting the segmentation accuracy; and a cardiovascular CT image intelligent segmentation method (application publication number: CN117952994A) performs segmentation based on relevant feature parameters of the region block, but the computational complexity is high and faces challenges in actual clinical applications.

[0004] With the rapid development of deep learning technology, image segmentation methods based on deep learning have made significant progress in the field of medical imaging, providing a new solution for cardiovascular segmentation. Compared with traditional segmentation methods, deep learning methods can automatically learn complex features in images without relying on manually designed features, thereby significantly improving segmentation accuracy. For example, segmentation models based on convolutional neural networks (such as U-Net, DeepLab, etc.) can effectively capture cardiovascular details through multi-level convolution and pooling operations. In addition, Transformer-based models (such as ViT, Swin Transformer, etc.) further enhance the model's ability to model long-distance dependencies through a global attention mechanism, and show stronger robustness when processing cardiovascular structures. However, deep learning models also face challenges such as long training time, high computing resource requirements, and difficulty in clinical deployment. In this context, developing a CTA image cardiovascular segmentation system with both high segmentation accuracy and high training efficiency is not only of great academic value, but also of far-reaching clinical application significance.

[0005] At present, the development of such systems faces the following key problems: the density difference between the blood vessel wall and the surrounding tissue is small, which is manifested in medical images as unclear and irregular edge areas, and the rest are non-edge areas. The edge area has unclear boundaries, which makes feature extraction and segmentation difficult, while the non-edge area has clear features. Existing methods usually train the two together, resulting in the model having to process two completely different feature distributions at the same time, which increases the difficulty of training and reduces efficiency; therefore, the lack of differentiated training strategies based on the characteristics of edge areas and non-edge areas has become one of the key factors affecting system performance. Summary of the invention

[0006] In order to solve the above problems, the present disclosure proposes a data-driven iterative learning cardiovascular segmentation method and system to achieve efficient training and high-precision cardiovascular image segmentation, so as to promote the early diagnosis, treatment planning and efficacy evaluation of cardiovascular diseases, while adapting to the clinical environment with limited computing resources.

[0007] According to some embodiments, the present disclosure adopts the following technical solutions: A data-driven iterative learning cardiovascular segmentation method, comprising: Acquire a cardiovascular image to be segmented; Inputting the cardiovascular image into the learned segmentation model to obtain the segmentation result; Among them, the segmentation model adopts a data-driven iterative learning method. In each iteration, first, based on the image set of this training, the non-edge set and the edge set are constructed according to the blur degree and the edge pixel ratio of the image block; then, the segmentation model is first trained with the non-edge set under the constraint of the segmentation loss function, and based on the model parameters after the non-edge set training, the edge set is used to continue training under the coordination of the distribution alignment loss function. Finally, the model parameters trained in this round are used for parameter initialization of the next round.

[0008] According to some embodiments, the present disclosure adopts the following technical solutions: A data-driven iterative learning cardiovascular segmentation system, comprising: The image acquisition module is configured to: acquire a cardiovascular image to be segmented; The image segmentation module is configured to: input the cardiovascular image into the learned segmentation model to obtain a segmentation result; Among them, the segmentation model adopts a data-driven iterative learning method. In each iteration, first, based on the image set of this training, the non-edge set and the edge set are constructed according to the blur degree and the edge pixel ratio of the image block; then, the segmentation model is first trained with the non-edge set under the constraint of the segmentation loss function, and based on the model parameters after the non-edge set training, the edge set is used to continue training under the coordination of the distribution alignment loss function. Finally, the model parameters trained in this round are used for parameter initialization of the next round.

[0009] According to some embodiments, the present disclosure adopts the following technical solutions: A computer program product comprises a computer program, wherein when the computer program is executed by a processor, the cardiovascular segmentation method based on data-driven iterative learning is implemented.

[0010] According to some embodiments, the present disclosure adopts the following technical solutions: A non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the data-driven iterative learning cardiovascular segmentation method is implemented.

[0011] According to some embodiments, the present disclosure adopts the following technical solutions: An electronic device comprises: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes the cardiovascular segmentation method of implementing the data-driven iterative learning.

[0012] Compared with the prior art, the present invention has the following beneficial effects: The present invention proposes a data-driven iterative learning cardiovascular segmentation method and system, which is specially designed for CTA cardiovascular image segmentation tasks. Compared with the prior art, the present invention has the following advantages and beneficial effects: (1) Highly targeted, optimizing edge area segmentation accuracy: The present invention significantly improves the segmentation accuracy of edge areas (where the density difference between the blood vessel wall and surrounding tissue is small and the boundary is unclear) by dividing CTA images into non-edge images and edge set images and adopting different training strategies respectively; the edge set loss function (including segmentation loss and distribution alignment loss) designed for the edge area further enhances the ability to capture fuzzy boundaries and solves the problem that traditional methods have poor segmentation effect in edge areas.

[0013] (2) High training efficiency and fast model convergence speed: By giving priority to training non-edge areas and further training edge areas based on their model parameters, the present invention significantly reduces the learning difficulty of the model and improves the training efficiency.

[0014] (3) Feature distribution alignment to improve model generalization ability: The present invention introduces a distribution alignment module, which realizes the alignment of feature distributions of two types of regions by calculating and minimizing the Wasserstein distance between the feature distributions of non-marginal sets and marginal sets. This design effectively integrates the feature information of non-marginal regions and marginal regions, enhances the model's ability to model the overall cardiovascular structure, and improves the generalization ability of the model, enabling it to adapt to CTA images under different patients and imaging conditions.

[0015] (4) Low computing resource requirements, suitable for clinical deployment: By prioritizing the training of non-edge areas and optimizing the edge area training strategy, the present invention significantly reduces the computational complexity of the model; at the same time, the lightweight design of the segmentation model and the efficient training process enable the system to operate efficiently in a clinical environment with limited computing resources, meeting the needs of actual applications.

[0016] (5) Support iterative learning and gradually optimize segmentation performance: The present invention adopts an iterative learning mechanism to gradually optimize the model performance through multiple rounds of training; this iterative learning method can continuously correct the segmentation error and further improve the segmentation accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings constituting a part of the present disclosure are used to provide a further understanding of the present disclosure. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation on the present disclosure.

[0018] Figure 1This is a diagram of the internal structure of the segmentation model of Example 1. Figure 2 This is a diagram of the internal structure of the encoder of Example 1.

[0019] Figure 3 This is a diagram of the internal structure of the decoder of Example 1.

[0020] Figure 4 This is a learning flow chart of Example 1.

[0021] Figure 5 This is the data partitioning flow chart of Example 1.

[0022] Figure 6 This is a flow chart of non-edge set segmentation in Example 1.

[0023] Figure 7 This is the edge set segmentation flow chart of Example 1.

[0024] Figure 8 This is a distribution alignment flow chart of Example 1.

[0025] Fig. 9 is an example diagram of the cardiovascular image and label image of Example 1, wherein: Fig. 9 (a) and (b) in the figure are cardiovascular images and corresponding label images, respectively. DETAILED DESCRIPTION

[0026] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0027] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present disclosure belongs.

[0028] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0029] Example 1 In one embodiment of the present disclosure, a data-driven iterative learning cardiovascular segmentation method is provided to achieve efficient training and high-precision cardiovascular image segmentation, so as to promote early diagnosis, treatment planning and efficacy evaluation of cardiovascular diseases, while adapting to clinical environments with limited computing resources, including: Acquire a cardiovascular image to be segmented; Inputting the cardiovascular image into the learned segmentation model to obtain the segmentation result; Among them, the segmentation model adopts a data-driven iterative learning method. In each iteration, first, based on the image set of this training, the non-edge set and the edge set are constructed according to the blur degree and the edge pixel ratio of the image block; then, the segmentation model is first trained with the non-edge set under the constraint of the segmentation loss function, and based on the model parameters after the non-edge set training, the edge set is used to continue training under the coordination of the distribution alignment loss function. Finally, the model parameters trained in this round are used for parameter initialization of the next round.

[0030] As an example, the segmentation model is used as follows: Figure 1 The structure shown in FIG. 1 is an N-layer convolutional encoder, an N-layer convolutional decoder, and a segmentation prediction head connected in sequence. There is a residual connection between each layer of encoder and decoder. The encoder and decoder are used for feature extraction. The features output by the last layer of decoder are input into the segmentation prediction head to obtain the segmented cardiovascular image.

[0031] The encoder and decoder use conventional structures, such as Figure 2 , Figure 3 As shown in the figure, the encoder consists of a convolutional layer, a normalization layer, an activation layer, an activation layer, a Dropout layer and a downsampling layer in sequence, while the decoder consists of a 1*1 convolutional layer, an upsampling layer, a feature concatenation layer, a convolutional layer, a normalization layer, an activation layer and a Dropout layer in sequence, and the segmentation prediction head consists of a fully connected layer.

[0032] Based on the above model structure, a data-driven iterative learning method is used to batch train the segmentation model. Figure 4 The four aspects of data partitioning, non-edge segmentation, edge segmentation and distribution alignment are shown to illustrate the specific learning process of each iteration.

[0033] 1. Data Division like Figure 5 As shown, the cardiovascular image set used for this iteration is obtained ,in, N is the sample size, For the samples, is the original cardiovascular image, is the label image corresponding to the cardiovascular image, the sample is as follows Fig. 9 As shown, Fig. 9(a) and (b) are the cardiovascular images and the corresponding label images, respectively. The red mark is the cardiovascular (coronary artery) to be identified. According to the blur degree of the image block and the edge pixel ratio, each image is divided into non-edge images and edge images. The non-edge image corresponds to the area with clear features and clear boundaries, while the edge set image corresponds to the area with small density difference between the blood vessel wall and the surrounding tissue and unclear boundaries. The specific steps are as follows: (1) Divide each cardiovascular image and the corresponding label image in the image set into blocks to obtain a number of image blocks; Specifically, for each sample Divide the cardiovascular images into blocks and label image Divided into The size is image blocks.

[0034] (2) Computational cardiovascular images The blur level of each image block in ; The average entropy is used to characterize the blur level, and the following formula is used to calculate the cardiovascular image The average entropy of each image block in is:

[0035] in, Representing images Middle In the image block The probability of pixels, the average entropy value The higher it is, the more chaotic the pixel distribution of the image block is, that is, the higher the degree of blur.

[0036] (3) Calculate label image The ratio of edge pixels of each image block in :

[0037] in, is the indicator function, Represents a label image Middle In the image block pixels are edge pixels.

[0038] (4) Using soft threshold masking, based on the blur level and edge pixel ratio, construct the The non-marginal set mask of samples and edge set mask .

[0039] Specifically, For the The non-marginal set mask of samples is given byL indivual composition, For the In the sample The non-edge set mask of the image patch; For the The edge set mask of samples is given by L indivual composition, For the In the sample The edge set mask of the image patch, and The calculation method is expressed as follows:

[0040]

[0041] in, is a hyperparameter, in this embodiment, it is set to 0.8; Defined as the average entropy of the image The ratio of edge pixels to labels The maximum value of .

[0042] (5) Based on the mask, generate non-edge sets and edge sets.

[0043] Use the The mask of samples and , for cardiovascular images and label image Generate non-edge images and edge images respectively, expressed as follows: , , ,

[0044] in, , Cardiovascular images The non-edge image and edge image of , Label images respectively Non-edge images and edge images.

[0045] Based on each sample Generated non-edge images and edge images, construct edge sets and non-marginal sets , used for the following training.

[0046] 2. Non-marginal set segmentation like Figure 6 As shown, the non-edge set image is input into Figure 3 In the segmentation model shown, the segmentation loss function is used for parameter training to obtain the trained model parameters and non-edge set features, where the non-edge set features are the features output by the last layer of decoder.

[0047] The segmentation loss function consists of cross entropy loss and Dice loss to enhance the model's ability to capture fuzzy boundaries and segmentation accuracy. By prioritizing training of non-edge areas, the model's ability to learn clear features can be quickly optimized, laying the foundation for subsequent training of edge areas.

[0048] Specifically, the segmentation loss function of the non-marginal set is: , , , in, is the cross entropy loss, is the Dice loss, is the label image, is the probability value predicted by the model, N is the number of samples, is a small constant set to .

[0049] The parameter learning process of the model is: , in, For non-marginal sets The model parameters of the training iteration, express Figure 3 The segmentation model shown, Representing cardiovascular images Middle image blocks, Represents a label image Middle image blocks, For the edge set The model parameters of the iterative training are hour, , which is the model initialization parameter.

[0050] 3. Edge Set Segmentation like Figure 7As shown in the figure, based on the model parameters after non-edge set training, the edge set image is input into the segmentation model, and the edge set loss function is used for parameter training to obtain the trained model parameters and edge set features. The edge set features here are also the features output by the last layer of decoder. The edge set loss function includes a segmentation loss function and a distribution alignment loss function. Specifically, the edge set loss function is:

[0051] in, is the segmentation loss, is the distribution alignment loss obtained by distribution alignment, It is a hyperparameter and is set to 0.1 in this embodiment.

[0052] The parameter learning process of the model is:

[0053] in, For the edge set The model parameters of the training iteration, express Figure 3 The segmentation model shown, For non-marginal sets The model parameters trained at this iteration.

[0054] The advantages of first splitting the non-edge set and then the edge set are: Edge areas usually contain fuzzy and complex details, and direct segmentation can easily introduce noise and errors; the features of non-edge areas are relatively uniform and the structure is clear. Segmenting the non-edge areas first can obtain relatively stable and accurate preliminary segmentation results, laying a good foundation for the overall segmentation task.

[0055] 4. Distribution Alignment like Figure 8 As shown in Figure 1, the output features of the last layer decoder of the segmentation model in the non-edge set segmentation and the edge set segmentation are input, the Gaussian distribution of the two is calculated, and the Wasserstein distance between the two Gaussian distributions is minimized to obtain the distribution alignment loss function, thereby realizing the alignment of the non-edge set feature distribution and the edge set feature distribution. The specific steps are: (1) Using the segmentation model, features are extracted from the non-edge set and the edge set respectively; Get the features of the last decoder output of the segmentation model that are not concentrated on the edge The features output by the decoder at the last layer of the segmentation model in the edge set segmentation .

[0056] (2) Based on the extracted features, estimate the Gaussian distribution to obtain the multivariate Gaussian distribution of non-marginal set features and the multivariate Gaussian distribution of marginal set features; Specifically, we estimate the non-marginal set features Each category Multivariate Gaussian distribution of , and edge set features Each category Multivariate Gaussian distribution of ,in, In this embodiment, the category set Includes two categories: cardiovascular and background.

[0057] (3) Add perturbations to the covariance matrix of the multivariate Gaussian distribution; Considering that the multivariate Gaussian distribution of each category obtained from the non-marginal set may not fully cover all the data of the category, the obtained distribution has a shift, that is, the covariate shift phenomenon. Therefore, this embodiment adds appropriate disturbances to the estimated empirical covariance matrix to mitigate the impact of covariance shift. , and its covariance is updated as:

[0058] in, represents a matrix of all 1s, is a random number that controls the degree of disturbance, and the updated distribution is obtained .

[0059] (4) After adding the perturbation, calculate the Wasserstein distance between the two multivariate Gaussian distributions; The Wasserstein distance between two multivariate Gaussian distributions is defined to measure the difference in the representation distribution of fuzzy and non-fuzzy region sets; for each category , the Wasserstein distance is:

[0060] Where tr is the trace of the matrix.

[0061] (5) Based on the Wasserstein distance, a distribution alignment loss function is constructed.

[0062] Finally, the overall distribution alignment loss is defined as the sum of the Wasserstein distances of each category: .

[0063] By minimizing the Wasserstein distance between the feature distributions of the non-edge set and the edge set, the feature distributions of the two parts can be effectively aligned, reducing the model bias caused by data distribution differences, thereby improving the recognition consistency of the model in different regions; by achieving consistency in the feature distributions of the fuzzy area and the non-fuzzy area, the model can maintain stable and excellent segmentation performance when processing cardiovascular edge fuzzy areas, thereby enhancing the robustness and generalization ability of the model.

[0064] Example 2 In one embodiment of the present disclosure, a data-driven iterative learning cardiovascular segmentation system is provided, comprising: The image acquisition module is configured to: acquire a cardiovascular image to be segmented; The image segmentation module is configured to: input the cardiovascular image into the learned segmentation model to obtain a segmentation result; Among them, the segmentation model adopts a data-driven iterative learning method. In each iteration, first, based on the image set of this training, the non-edge set and the edge set are constructed according to the blur degree and the edge pixel ratio of the image block; then, the segmentation model is first trained with the non-edge set under the constraint of the segmentation loss function, and based on the model parameters after the non-edge set training, the edge set is used to continue training under the coordination of the distribution alignment loss function. Finally, the model parameters trained in this round are used for parameter initialization of the next round.

[0065] Example 3 In one embodiment of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the data-driven iterative learning cardiovascular segmentation method is implemented.

[0066] Example 4 In one embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the data-driven iterative learning cardiovascular segmentation method is implemented.

[0067] Example 5 In one embodiment of the present disclosure, an electronic device is provided, comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the cardiovascular segmentation method of implementing the data-driven iterative learning.

[0068] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0069] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0070] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. A data-driven iterative learning cardiovascular segmentation method, characterized in that: include: Acquire a cardiovascular image to be segmented; Inputting the cardiovascular image into the learned segmentation model to obtain the segmentation result; Among them, the segmentation model adopts a data-driven iterative learning method. In each iteration, first, based on the image set of this training, the non-edge set and the edge set are constructed according to the blur degree and the edge pixel ratio of the image block; then, the segmentation model is first trained with the non-edge set under the constraint of the segmentation loss function, and based on the model parameters after the non-edge set training, the edge set is used to continue training under the coordination of the distribution alignment loss function. Finally, the model parameters trained in this round are used for parameter initialization of the next round.

2. A data-driven iterative learning cardiovascular segmentation method according to claim 1, characterized in that: The image set consists of a plurality of cardiovascular images and corresponding label images; The specific steps of constructing the non-edge set and the edge set are as follows: Each cardiovascular image and the corresponding label image in the image set are divided into blocks to obtain a number of image blocks; Calculate the blur level of each image block in the cardiovascular image; Calculate the edge pixel ratio of each image block in the label image; Using soft threshold masking, masks of non-edge set and edge set are constructed based on blur degree and edge pixel ratio; Based on the mask, non-edge sets and edge sets are generated.

3. A data-driven iterative learning cardiovascular segmentation method as claimed in claim 2, characterized in that: The masks of the non-edge set and the edge set are respectively composed of the non-edge set masks and the edge set masks of a plurality of image blocks. The non-edge set masks and the edge set masks of the image blocks are expressed by the formula: in, For the In the sample The non-edge set mask of the image patch, For the In the sample The edge set mask of the image patch, is a hyperparameter, Defined as the blurriness of the image patch and the edge pixel ratio of the label image patch The maximum value of .

4. The data-driven iterative learning cardiovascular segmentation method according to claim 1, characterized in that: The loss function of the non-edge set is a segmentation loss function composed of a cross entropy loss and a Dice loss; The loss function of the edge set is an edge set loss function composed of a segmentation loss function and a distribution alignment loss function.

5. The data-driven iterative learning cardiovascular segmentation method according to claim 1, characterized in that: The distribution alignment loss function is used to minimize the Wasserstein distance between the feature distributions of the non-marginal set and the marginal set.

6. A data-driven iterative learning cardiovascular segmentation method according to claim 5, characterized in that: The construction process of the distribution alignment loss function is: Using the segmentation model, features are extracted from the non-edge set and the edge set respectively; Based on the extracted features, the Gaussian distribution is estimated to obtain the multivariate Gaussian distribution of the non-marginal set features and the multivariate Gaussian distribution of the marginal set features; Add perturbations to the covariance matrix of a multivariate Gaussian distribution; After adding the perturbation, the Wasserstein distance between two multivariate Gaussian distributions is calculated; Based on the Wasserstein distance, a distribution alignment loss function is constructed.

7. A data-driven iterative learning cardiovascular segmentation system, characterized in that: include: The image acquisition module is configured to: acquire a cardiovascular image to be segmented; The image segmentation module is configured to: input the cardiovascular image into the learned segmentation model to obtain a segmentation result; Among them, the segmentation model adopts a data-driven iterative learning method. In each iteration, first, based on the image set of this training, the non-edge set and the edge set are constructed according to the blur degree and the edge pixel ratio of the image block; then, the segmentation model is first trained with the non-edge set under the constraint of the segmentation loss function, and based on the model parameters after the non-edge set training, the edge set is used to continue training under the coordination of the distribution alignment loss function. Finally, the model parameters trained in this round are used for parameter initialization of the next round.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the data-driven iterative learning cardiovascular segmentation method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by the processor, the cardiovascular segmentation method of data-driven iterative learning is implemented as described in any one of claims 1-6.

10. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes a cardiovascular segmentation method of data-driven iterative learning as described in any one of claims 1-6.

Images