Calcium score integral calculation method and device, electronic equipment and storage medium

Abstract

The application provides a coronary artery calcification score calculation method and device, electronic equipment and a storage medium. The calculation method comprises: acquiring a plurality of 2D coronal plane CT images determined according to a 3D coronary artery plain CT image of a patient; for each 2D coronal plane CT image, inputting the 2D coronal plane CT image into a pre-constructed calcification classification model to determine calcification regions included in the image and coronary artery vessel type labels of each calcification region; performing matrix multiplication on the image corresponding to the determined calcification region and the 2D coronal plane CT image to determine a vessel calcification plaque including only the coronary artery vessel type label; performing calcification score calculation on the vessel calcification plaque including only the coronary artery vessel type label determined for each 2D coronal plane CT image to determine the total coronary artery calcification score of the patient. In this way, the coronary artery calcification score calculation based on the plain CT can be automatically and quickly and accurately completed.

Description

Technical Field

[0001] This application relates to the field of calcium integral calculation technology, and in particular to a method, apparatus, electronic device and storage medium for calculating coronary artery calcium integral. Background Technology

[0002] Cardiovascular disease is the leading cause of death worldwide. The degree of calcification of plaques in the coronary arteries is a crucial indicator for monitoring and predicting cardiovascular disease. Plain CT scans, a commonly used screening tool for coronary artery disease, are frequently used to calculate indicators such as the calcium score. The calcium score, calculated from plain CT images, is an evaluation index reflecting the degree of calcification in the coronary arteries. However, in practice, plain CT images often lack clear visualization of the coronary arteries, and the absence of synchronized electrocardiogram (ECG) results in motion artifacts and scan dose, leading to image noise. This can cause errors or inaccuracies in the marking of manually labeled calcified areas, resulting in inaccurate calcium scores. Furthermore, the method of manually marking calcified areas suffers from low efficiency and significant errors in calculating coronary artery calcium scores. Summary of the Invention

[0003] In view of this, the purpose of this application is to provide a method, device, electronic device and storage medium for calculating coronary artery calcium score, which can automatically, quickly and accurately complete the calculation of coronary artery calcium score based on plain CT scan.

[0004] This application provides a method for calculating coronary artery calcium scores, the method comprising:

[0005] Multiple target 2D coronary CT images were obtained based on the patient's target 3D coronary artery plain CT images.

[0006] For each target 2D coronal CT image, the target 2D coronal CT image is input into a pre-constructed calcification classification model to determine the calcified regions included in the target 2D coronal CT image and the coronary vessel type label of each calcified region. The calcification classification model is composed of a feature extraction sub-model, an instance region molecular model, a calcification region determination sub-model, and a vessel classification sub-model. The feature extraction sub-model is used to extract feature maps, the instance region molecular model is used to determine the bounding boxes of suspected calcified plaques from the feature maps, the calcification region determination sub-model is used to determine the calcified regions from the bounding boxes, and the vessel classification sub-model is used to determine the coronary vessel type label of the calcified regions.

[0007] The image corresponding to the calcified region included in the identified 2D coronal plain CT image of the target is multiplied by the matrix with the 2D coronal plain CT image of the target to identify vascular calcified plaques that only include coronary artery type labels.

[0008] For each target 2D coronary plain CT image, calcification plaques that include only coronary artery type labels are identified, and calcification scores are calculated to determine the patient's total coronary artery calcification score.

[0009] Optionally, the step of inputting the target 2D coronal plain CT image into a pre-constructed calcification classification model to determine the calcified regions included in the target 2D coronal plain CT image and the coronary vessel type label for each calcified region includes:

[0010] The target's 2D coronal CT image is input into the calcification classification model. Feature image extraction is performed through the feature extraction sub-model in the calcification classification model to determine the feature map of the target's 2D coronal CT image.

[0011] The feature map of the target 2D coronal CT image is input into the instance region molecular model in the calcification classification model, and bounding box regression processing is performed to obtain the target bounding box of at least one suspected calcification plaque included in the target 2D coronal CT image; wherein, the target bounding box has multiple pixels; wherein, the target bounding box is a rectangular bounding box;

[0012] The feature map of the target 2D coronal CT image and the target bounding box of the at least one suspected calcified plaque are input one by one into the calcification region determination sub-model in the calcification classification model. It is determined whether the pixel in each target bounding box is a calcified pixel, and the calcification image of the calcification region included in the target bounding box is determined based on the determined calcified pixels.

[0013] The calcification image, the target bounding box, and the feature map of the target 2D coronary CT image are input into the blood vessel classification sub-model in the calcification classification model to determine the coronary blood vessel type and determine the coronary blood vessel type label for each calcification region.

[0014] Optionally, the step of calculating the total coronary artery calcium score for the patient by identifying vascular calcification plaques containing only coronary artery type labels from each target 2D coronary plain CT image includes:

[0015] For each target 2D coronal plain CT image, the calcification integral of each vascular calcification plaque in the target 2D coronal plain CT image is determined based on the vascular calcification plaques included in the target 2D coronal plain CT image.

[0016] The calcium scores of each coronary artery with the same coronary vessel type label in each target 2D coronary plain CT image are summed to determine the calcium score of each coronary vessel.

[0017] The total coronary artery calcium score of the patient is determined by summing the calcium scores of each coronary artery.

[0018] Optionally, the calcification integral includes the Agassone integral, the volume integral, and the mass integral.

[0019] Optionally, the calcification classification model can be constructed through the following steps:

[0020] Acquire multiple 3D coronary artery plain CT images to be trained, and for each 3D coronary artery plain CT image to be trained, acquire multiple 2D coronary plane plain CT images to be trained with calcification information tags.

[0021] For each 2D coronary CT image to be trained in each 3D coronary artery plain CT image to be trained, the training 2D coronary CT image is input into the feature extraction neural network to obtain the training feature map of the training 2D coronary CT image.

[0022] The feature map to be trained is input into the first convolutional neural network to predict the bounding box of the suspected calcified plaque, determine the predicted bounding box of the suspected calcified plaque in the 2D coronal CT image to be trained, and determine the value of the first loss function based on the predicted bounding box and the calcification information label.

[0023] The training feature map and predicted bounding box of the 2D coronal CT image to be trained are input into the second convolutional neural network to predict whether all pixels within the predicted bounding box are calcified pixels. Based on the prediction results, the calcified region included in the predicted bounding box is determined. Based on the determined calcified region included in the predicted bounding box and the calcification information label, the value of the second loss function is determined.

[0024] The training feature map, predicted bounding box, and calcified region included in the predicted bounding box of the 2D coronary CT image to be trained are input into the third convolutional neural network to predict the coronary artery type label corresponding to the calcified region. The value of the third loss function is determined based on the prediction result of the coronary artery type label and the calcification information label.

[0025] Based on the first loss function value, the second loss function value, and the third loss function value, the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are iteratively updated simultaneously.

[0026] Once the network parameters of each neural network converge, the training ends, and the sub-models corresponding to each neural network are obtained. The calcification classification model is then constructed based on the obtained sub-models corresponding to each neural network.

[0027] Optionally, the step of iteratively updating the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network based on the first loss function value, the second loss function value, and the third loss function value includes:

[0028] The first loss function value, the second loss function value, and the third loss function value are summed to determine the total loss function value.

[0029] Based on the total loss function value, the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are iteratively updated simultaneously.

[0030] Optionally, the calcification information label includes a true calcification region annotation box, a labeling result indicating whether each pixel in the calcification region annotation box is a calcified pixel, and a true coronary artery type label corresponding to the calcification region formed by all calcified pixels in the same calcification region annotation box.

[0031] This application embodiment also provides a device for calculating coronary artery calcium scores, the device comprising:

[0032] The acquisition module is used to acquire multiple target 2D coronary CT images based on the patient's target 3D coronary artery plain CT images.

[0033] The first determination module is used to input each target 2D coronal plain CT image into a pre-constructed calcification classification model to determine the calcified regions included in the target 2D coronal plain CT image and the coronary vessel type label of each calcified region. The calcification classification model is composed of a feature extraction sub-model, an instance region molecular model, a calcification region determination sub-model, and a vessel classification sub-model. The feature extraction sub-model is used to extract feature maps, the instance region molecular model is used to determine the bounding boxes of suspected calcified plaques from the feature maps, the calcification region determination sub-model is used to determine the calcified regions from the bounding boxes, and the vessel classification sub-model is used to determine the coronary vessel type label of the calcified regions.

[0034] The second determining module is used to perform matrix multiplication between the image corresponding to the calcified region included in the determined target 2D coronal plain CT image and the target 2D coronal plain CT image to determine the vascular calcified plaque that only includes the coronary artery type label.

[0035] The calculation module is used to calculate the calcification score of the vascular calcification plaques identified in each target 2D coronary plain CT image, which only includes the coronary artery type label, and to determine the total coronary artery calcification score of the patient.

[0036] Optionally, when the first determining module inputs the target 2D coronal CT image into a pre-built calcification classification model to determine the calcification regions included in the target 2D coronal CT image and the coronary vessel type label of each calcification region, the first determining module is used to:

[0037] The target's 2D coronal CT image is input into the calcification classification model. Feature image extraction is performed through the feature extraction sub-model in the calcification classification model to determine the feature map of the target's 2D coronal CT image.

[0038] The feature map of the target 2D coronal CT image is input into the instance region molecular model in the calcification classification model, and bounding box regression processing is performed to obtain the target bounding box of at least one suspected calcification plaque included in the target 2D coronal CT image; wherein, the target bounding box has multiple pixels; wherein, the target bounding box is a rectangular bounding box;

[0039] The feature map of the target 2D coronal CT image and the target bounding box of the at least one suspected calcified plaque are input one by one into the calcification region determination sub-model in the calcification classification model. It is determined whether the pixel in each target bounding box is a calcified pixel, and the calcification image of the calcification region included in the target bounding box is determined based on the determined calcified pixels.

[0040] The calcification image, the target bounding box, and the feature map of the target 2D coronary CT image are input into the blood vessel classification sub-model in the calcification classification model to determine the coronary blood vessel type and determine the coronary blood vessel type label for each calcification region.

[0041] Optionally, when the calculation module is used to calculate the total coronary artery calcium score of the patient by analyzing the calcification score of vascular calcification plaques identified in each target 2D coronary plain CT image that include only coronary artery type labels, the calculation module is used to:

[0042] For each target 2D coronal plain CT image, the calcification integral of each vascular calcification plaque in the target 2D coronal plain CT image is determined based on the vascular calcification plaques included in the target 2D coronal plain CT image.

[0043] The calcium scores of each coronary artery with the same coronary vessel type label in each target 2D coronary plain CT image are summed to determine the calcium score of each coronary vessel.

[0044] The total coronary artery calcium score of the patient is determined by summing the calcium scores of each coronary artery.

[0045] Optionally, the calcification integral includes the Agassone integral, the volume integral, and the mass integral.

[0046] Optionally, the computing device further includes a model building module, the model building module being used for:

[0047] Acquire multiple 3D coronary artery plain CT images to be trained, and for each 3D coronary artery plain CT image to be trained, acquire multiple 2D coronary plane plain CT images to be trained with calcification information tags.

[0048] For each 2D coronary CT image to be trained in each 3D coronary artery plain CT image to be trained, the training 2D coronary CT image is input into the feature extraction neural network to obtain the training feature map of the training 2D coronary CT image.

[0049] The feature map to be trained is input into the first convolutional neural network to predict the bounding box of the suspected calcified plaque, determine the predicted bounding box of the suspected calcified plaque in the 2D coronal CT image to be trained, and determine the value of the first loss function based on the predicted bounding box and the calcification information label.

[0050] The training feature map and predicted bounding box of the 2D coronal CT image to be trained are input into the second convolutional neural network to predict whether all pixels within the predicted bounding box are calcified pixels. Based on the prediction results, the calcified region included in the predicted bounding box is determined. Based on the determined calcified region included in the predicted bounding box and the calcification information label, the value of the second loss function is determined.

[0051] The training feature map, predicted bounding box, and calcified region included in the predicted bounding box of the 2D coronary CT image to be trained are input into the third convolutional neural network to predict the coronary artery type label corresponding to the calcified region. The value of the third loss function is determined based on the prediction result of the coronary artery type label and the calcification information label.

[0052] Based on the first loss function value, the second loss function value, and the third loss function value, the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are iteratively updated simultaneously.

[0053] Once the network parameters of each neural network converge, the training ends, and the sub-models corresponding to each neural network are obtained. The calcification classification model is then constructed based on the obtained sub-models corresponding to each neural network.

[0054] Optionally, when the model building module iteratively updates the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network based on the first loss function value, the second loss function value, and the third loss function value, the model building module is used to:

[0055] The first loss function value, the second loss function value, and the third loss function value are summed to determine the total loss function value.

[0056] Based on the total loss function value, the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are iteratively updated simultaneously.

[0057] Optionally, the calcification information label includes a true calcification region annotation box, a labeling result indicating whether each pixel in the calcification region annotation box is a calcified pixel, and a true coronary artery type label corresponding to the calcification region formed by all calcified pixels in the same calcification region annotation box.

[0058] This application also provides an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the calculation method described above are performed.

[0059] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the calculation method described above.

[0060] This application provides a method, apparatus, electronic device, and storage medium for calculating coronary artery calcium scores. The calculation method includes: acquiring multiple target 2D coronary CT images determined based on a patient's target 3D coronary CT image; for each target 2D coronary CT image, inputting the target 2D coronary CT image into a pre-constructed calcium classification model to determine the calcified regions included in the target 2D coronary CT image and the coronary vessel type label for each calcified region; wherein, the calcium classification model is composed of a feature extraction sub-model, an instance region molecular model, a calcified region determination sub-model, and a vessel classification sub-model connected together, the feature extraction sub-model... The feature map is extracted using the instance region molecular model, which identifies the bounding box of the suspected calcified plaque from the feature map. The calcified region identification sub-model identifies the calcified region from the bounding box, and the vessel classification sub-model identifies the coronary vessel type label of the calcified region. The image corresponding to the calcified region in the identified target 2D coronary plain CT image is matrix-multiplied with the target 2D coronary plain CT image to identify vascular calcified plaques that only include the coronary vessel type label. For each vascular calcified plaque identified in the target 2D coronary plain CT image that only includes the coronary vessel type label, a calcification integral is calculated to determine the patient's total coronary calcification integral.

[0061] In this way, through the collaborative work of multiple sub-models in the calcification classification model disclosed in this application, the coronary artery type label of the calcified region can be determined quickly in one step, thereby improving the calculation efficiency of the calcification integral. Furthermore, by solving the complex 3D plain CT calcification integral calculation at the 2D coronal plane level, the complexity of the network can be reduced while maintaining performance. Thus, fast and accurate coronary artery calcification integral calculation is achieved.

[0062] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0063] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0064] Figure 1 A flowchart illustrating a method for calculating coronary artery calcium scores provided in an embodiment of this application;

[0065] Figure 2A schematic diagram of the structure of the calcification classification model provided in this application;

[0066] Figure 3 This is a schematic diagram of the plain CT image and the calcified area image provided in this application;

[0067] Figure 4 This is one of the structural schematic diagrams of a coronary artery calcification integral calculation device provided in an embodiment of this application;

[0068] Figure 5 This is a second schematic diagram of a coronary artery calcification integral calculation device provided in an embodiment of this application;

[0069] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0071] Cardiovascular disease is the leading cause of death worldwide. The degree of calcification of plaques in the coronary arteries is a crucial indicator for monitoring and predicting cardiovascular disease. Plain CT scans, a commonly used screening tool for coronary artery disease, are frequently used to calculate indicators such as the calcium score. The calcium score, calculated from plain CT images, is an evaluation index reflecting the degree of calcification in the coronary arteries. However, in practice, plain CT images often lack clear visualization of the coronary arteries, and the absence of synchronized electrocardiogram (ECG) results in motion artifacts and scan dose, leading to image noise. This can cause errors or inaccuracies in the marking of manually labeled calcified areas, resulting in inaccurate calcium scores. Furthermore, the method of manually marking calcified areas suffers from low efficiency and significant errors in calculating coronary artery calcium scores.

[0072] Based on this, embodiments of this application provide a method, apparatus, electronic device, and storage medium for calculating coronary artery calcium scores, which can automatically, quickly, and accurately complete the calculation of coronary artery calcium scores based on plain CT scans.

[0073] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for calculating coronary artery calcium scores, provided as an embodiment of this application. Figure 1 As shown in the embodiments of this application, the calculation method includes:

[0074] S101. Obtain multiple target 2D coronary CT images based on the target 3D coronary artery plain CT images of the patient.

[0075] Here, a 2D coronary CT image is determined based on a 3D coronary artery plain CT image. Specifically, this can be achieved by: determining the depth value of the 3D coronary artery plain CT image; and then segmenting the 3D coronary artery plain CT image into equal-thickness segments along the depth direction based on the determined depth value, thereby determining multiple target 2D coronary CT images of the target 3D coronary artery. The coronary artery refers to the coronary artery itself.

[0076] It should be noted that plain CT images are three-dimensional medical images acquired without contrast agents during the imaging process. Plain CT images are often used for initial screening of coronary artery disease and for calculating calcification scores. While high-density structures such as calcifications and stents can be observed on plain CT scans, blood vessels are difficult to visualize.

[0077] The coronal plane is a cross-section that longitudinally divides the human body into anterior and posterior parts along the left and right directions. All planes passing through the vertical and horizontal axes, and all planes parallel to them, are called coronal planes. These planes divide the human body into anterior and posterior parts.

[0078] For example, assuming the size of the target 3D coronary artery plain CT image is 512*512*100, 100 2D coronary plain CT images can be obtained, each with a size of 512*512.

[0079] S102. For each target 2D coronal plain CT image, input the target 2D coronal plain CT image into a pre-built calcification classification model to determine the calcification regions included in the target 2D coronal plain CT image and the coronary vessel type label of each calcification region.

[0080] Here, the calcification classification model is composed of a feature extraction sub-model, an instance region molecular model, a calcification region determination sub-model, and a vessel classification sub-model. The feature extraction sub-model is used to extract feature maps, the instance region molecular model is used to determine the bounding boxes of suspected calcified plaques from the feature maps, the calcification region determination sub-model is used to determine the calcified region from the bounding boxes, and the vessel classification sub-model is used to determine the coronary vessel type label of the calcified region.

[0081] For an example, please refer to Figure 2 , Figure 2 A schematic diagram of the structure of the calcification classification model provided in this application is shown below. Figure 2 As shown, the calcification classification model 200 includes a feature extraction sub-model 201, an instance region molecular model 202, a calcification region determination sub-model 203, and a blood vessel classification sub-model 204. The 2D coronal CT image input into the calcification classification model first passes through the feature extraction sub-model 201. The output of the feature extraction sub-model 201 is used as the input of the instance region molecular model 202, the calcification region determination sub-model 203, and the blood vessel classification sub-model 204, respectively. The output of the instance region molecular model 202 is also used as the input of the calcification region determination sub-model 203 and the blood vessel classification sub-model 204. The output of the calcification region determination sub-model 203 is also used as the input of the blood vessel classification sub-model 204. The output of the blood vessel classification sub-model 204 is the output of the calcification classification model 200.

[0082] Here, when inputting multiple target 2D coronal plain CT images into a pre-built calcification classification model, the method of inputting and identifying images one by one can be adopted to determine the calcification regions included in the target 2D coronal plain CT images and the coronary vessel type label of each calcification region.

[0083] The target 2D coronary CT image may or may not show at least one calcified region. The identified calcified region may be an irregularly shaped area. The coronary vessel type label may specifically include: left main coronary artery (LM), left anterior descending artery (LAD), left circular artery (LCX), right coronary artery (RCA), and branch vessels.

[0084] In one embodiment of this application, the step of inputting the target 2D coronal CT image into a pre-constructed calcification classification model to determine the calcification regions included in the target 2D coronal CT image and the coronary vessel type label of each calcification region includes: inputting the target 2D coronal CT image into the calcification classification model; extracting feature images through the feature extraction sub-model in the calcification classification model to determine the feature map of the target 2D coronal CT image; inputting the feature map of the target 2D coronal CT image into the instance region molecular model in the calcification classification model; performing bounding box regression processing to obtain at least one suspected calcification included in the target 2D coronal CT image. The target bounding box of the plaque; wherein the target bounding box has multiple pixels; wherein the target bounding box is a rectangular bounding box; the feature map of the target 2D coronal plain CT image and the target bounding box of the at least one suspected calcified plaque are input one by one into the calcification region determination sub-model of the calcification classification model to determine whether the pixels in each target bounding box are calcified pixels, and to determine the calcification image of the calcified region included in the target bounding box based on the determined calcified pixels; the calcification image, the target bounding box and the feature map of the target 2D coronal plain CT image are input into the blood vessel classification sub-model of the calcification classification model to determine the coronary blood vessel type and determine the coronary blood vessel type label of each calcified region.

[0085] Here, the obtained feature map is the image extracted after passing through the convolutional layer in the feature extraction submodule.

[0086] When multiple target bounding boxes are obtained, the size of each target bounding box can be different, specifically including nine different sizes of bounding boxes.

[0087] The calcification region determination sub-model identifies calcified regions within the target bounding boxes of suspected calcified plaques one by one. For each suspected calcified plaque's target bounding box, it confirms whether each pixel within that bounding box is a calcified pixel. The region formed by the identified calcified pixels is then defined as the calcified region, and a calcified image of the calcified region is obtained. The number of calcified images of the calcified regions identified by the calcification region determination sub-model is generally the same as the number of target bounding boxes identified by the instance region molecular model.

[0088] The blood vessel classification sub-model determines the coronary vessel type label of the calcified region based on the output results of the feature extraction sub-model (feature map), the output results of the instance region molecular model (target bounding box), and the output results of the calcification classification model (calcification image of the calcified region).

[0089] In another embodiment provided in this application, the calcification classification model is constructed through the following steps: acquiring multiple 3D coronary artery plain CT images to be trained, and for each 3D coronary artery plain CT image to be trained, acquiring multiple 2D coronary plane plain CT images to be trained with added calcification information labels; for each 2D coronary plane plain CT image to be trained in each 3D coronary artery plain CT image to be trained, inputting the 2D coronary plane plain CT image to be trained into a feature extraction neural network to obtain the training feature map of the 2D coronary plane plain CT image to be trained; inputting the training feature map into a first convolutional neural network to predict the bounding boxes of suspected calcified plaques, determining the predicted bounding boxes of suspected calcified plaques in the 2D coronary plane plain CT image to be trained, and determining the first loss function value based on the predicted bounding boxes and the calcification information labels; inputting the training feature map and the predicted bounding boxes of the 2D coronary plane plain CT image to be trained into a second convolutional neural network to perform pre- The process involves predicting whether all pixels within the bounding box are calcified pixels, determining the calcified regions included within the predicted bounding box based on the prediction results, and determining a second loss function value based on the determined calcified regions within the predicted bounding box and the calcification information label. The training feature map, predicted bounding box, and calcified regions within the predicted bounding box of the 2D coronary CT image to be trained are input into a third convolutional neural network to predict the coronary artery type label corresponding to the calcified regions. A third loss function value is determined based on the prediction results of the coronary artery type label and the calcification information label. The network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are iteratively updated based on the first, second, and third loss function values. Training ends when the network parameters of each neural network converge, resulting in a sub-model corresponding to each neural network. The calcification classification model is then constructed based on the obtained sub-models.

[0090] Here, the calcification information label includes a true calcification region annotation box, a labeling result indicating whether each pixel in the calcification region annotation box is a calcified pixel, and a true coronary artery type label corresponding to the calcification region formed by all calcified pixels in the same calcification region annotation box.

[0091] The feature extraction neural network can be a VGG-16 convolutional neural network. The feature maps extracted by the feature extraction neural network are used as shared features and input into the first, second, and third convolutional neural networks respectively. When using the feature extraction network for shared feature extraction, features shared with the other three neural networks can be obtained by extracting features from 2D coronal CT scans through convolutional layers, and the dimensionality of the shared features can be reduced by using 3×3 convolutional layers.

[0092] The first convolutional neural network structure can be built based on a Feature Pyramid Network (FPN) + Region Proposal Networks (RPN) structure. The RPN predicts bounding box locations and confidence levels using fully convolutional methods. The FPN is placed within the RPN network to generate regression boxes, with the aim of generating features of different scales and fusing them as input to the RPN network. This is because calcified plaques in 2D coronal CT scans and even 3D CT scans are usually not particularly large; a 16*16 bounding box may be sufficient to cover the plaque's location. Therefore, the introduction of the pyramidal FPN structure provides more features of different sizes for the RPN to generate bounding boxes of various sizes, aiming to encompass the location of the calcified plaque with the most suitable bounding box possible.

[0093] Specifically, the RPN structure predicts the location and confidence of bounding boxes containing calcified plaques. First, a feature map with a fixed number of channels and the same size as the shared features is generated by convolution with a 3×3 sliding window on the feature map to be trained. Then, a point is randomly selected from the feature map; this point has a one-to-one mapping with the original 2D coronal CT scan to be trained. Since the original 2D coronal CT scan and the feature map are different sizes, a point on the feature map will correspond to a bounding box on the original 2D coronal CT scan, but this is not the desired bounding box. The top-left corner or center of the bounding box is used as an anchor point, and a set of bounding boxes with different sizes are constructed. The RPN result is to determine whether these bounding boxes contain plaques and their offsets. The size and width of the bounding boxes are pre-defined, with nine possible combinations. The final predicted bounding boxes containing calcified plaques are derived from these nine combinations.

[0094] The step of determining the first loss function value based on the predicted bounding box and the calcification information label may specifically include: determining the first loss function value based on the actual calcification region annotation box in the predicted bounding box and the calcification information label.

[0095] When the second convolutional neural network predicts the calcified region within the bounding box, it determines the calcified region within the predicted bounding box one by one based on the feature map to be trained. When determining the calcified region within the predicted bounding box, it first predicts whether all pixels within the predicted bounding box are calcified pixels, and determines the region where the predicted calcified pixels are located as the calcified region included in the predicted bounding box.

[0096] The output of the second convolutional neural network is a pixel-level classification mask for the predicted bounding box of each suspected calcification patch. The size of the output mask is the same as the size of the feature map extracted by the region of interest pooling layer (Rol pooling) for that box.

[0097] The step of determining the second loss function value based on the calcified region included in the determined predicted bounding box and the calcified information label includes: determining the second loss function value based on the annotation result of whether each pixel in the calcified region annotation box in the calcified information label is a calcified pixel.

[0098] Here, the calcified region included within the predicted bounding box is determined based on the output of the first convolutional neural network, therefore the determined value of the second loss function depends on the output of the previous stage.

[0099] The step of determining the third loss function value based on the prediction result of the coronary artery type label and the calcification information label includes: determining the third loss function value based on the prediction result of the coronary artery type label and the actual coronary artery type label corresponding to the calcification region formed by all calcified pixels within the same calcification region annotation box in the calcification information label.

[0100] Here, the coronary artery type label prediction corresponding to the calcified region is determined based on the output of the first convolutional neural network and the output of the second convolutional neural network. Therefore, the determined value of the third loss function depends on the output of the previous two stages.

[0101] In another embodiment provided in this application, the step of iteratively updating the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network based on the first loss function value, the second loss function value, and the third loss function value includes: summing the first loss function value, the second loss function value, and the third loss function value to determine a total loss function value; and iteratively updating the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network based on the total loss function value.

[0102] The process of ending training after the network parameters of each neural network converge, and obtaining the sub-models corresponding to each neural network, includes: ending training after the network parameters of each neural network converge, and obtaining the feature extraction sub-model corresponding to the feature extraction neural network, the instance region molecular model corresponding to the first convolutional neural network, the calcification region determination sub-model corresponding to the second convolutional neural network, and the blood vessel classification sub-model corresponding to the third convolutional neural network.

[0103] S103. Multiply the image corresponding to the calcified region included in the determined 2D coronal plain CT image of the target by matrix multiplication with the target 2D coronal plain CT image to determine the vascular calcified plaques that only include the coronary artery type label.

[0104] Here, the image corresponding to the calcified region in the 2D coronal CT image of the target, as determined by the calcification classification model, is a black and white image containing only the calcified region, where the pixel value of the calcified region is 1. The pixel value of the other parts is 0.

[0105] Here, multiplying two images involves multiplying the pixel matrices of the two images to determine the actual pixel value (0-255) of each pixel in the calcified region. Based on the actual pixel value, the CT value of each point is determined, and then the calcification integral is calculated.

[0106] For an example, please refer to Figure 3 , Figure 3 The images provided in this application are schematic diagrams of plain CT images and calcified region images. The left image is a 2D coronal plain CT image of the target, and the right image is an image of the calcified region determined by the coronary artery classification model.

[0107] S104. For each target 2D coronary plain CT image, the calcification plaques that are identified and include only the coronary artery type label are calculated to determine the total coronary artery calcification score of the patient.

[0108] In one embodiment of this application, the step of calculating the total coronary artery calcium score for the patient by identifying vascular calcification plaques in each target 2D coronary CT image that include only coronary artery type labels includes a calcium score. This includes: for each target 2D coronary CT image, determining the calcium score of each vascular calcification plaque in the target 2D coronary CT image based on the identified vascular calcification plaques; summing the calcium scores of the same coronary artery type labels in each target 2D coronary CT image to determine the calcium score of each coronary artery; and summing the calcium scores of each coronary artery to determine the patient's total coronary artery calcium score. The calcium score includes the Agaston score, volume score, and mass score.

[0109] Here, firstly, the Agaston integral, volume integral, and mass integral of the coronary arteries LM, LAD, LCX, RCA, and branch vessels on each target 2D coronary CT image are calculated. Then, the Agaston integral, volume integral, and mass integral of the coronary arteries LM, LAD, LCX, RCA, and branch vessels on each target 2D coronary CT image are added together to obtain the Agaston integral, volume integral, and mass integral of the coronary arteries LM, LAD, LCX, RCA, and branch vessels. Finally, the Agaston integral, volume integral, and mass integral of the coronary arteries LM, LAD, LCX, RCA, and branch vessels are added together to obtain the total coronary artery calcification score of the patient.

[0110] In this way, through the collaborative work of multiple sub-models in the calcification classification model disclosed in this application, the coronary artery type label of the calcified region can be determined quickly in one step, thereby improving the calculation efficiency of the calcification integral. Furthermore, by solving the complex 3D plain CT calcification integral calculation at the 2D coronal plane level, the complexity of the network can be reduced while maintaining performance. Thus, fast and accurate coronary artery calcification integral calculation is achieved.

[0111] Please see Figure 4 , Figure 5 , Figure 4 This is one of the structural schematic diagrams of a coronary artery calcium integration calculation device provided in an embodiment of this application. Figure 5 This is a second schematic diagram of a coronary artery calcium integration calculation device provided in an embodiment of this application. Figure 4 As shown, the computing device 400 includes:

[0112] The acquisition module 410 is used to acquire multiple target 2D coronary CT images determined based on the patient's target 3D coronary artery plain CT images.

[0113] The first determining module 420 is used to input each target 2D coronal plain CT image into a pre-constructed calcification classification model to determine the calcified regions included in the target 2D coronal plain CT image and the coronary vessel type label of each calcified region. The calcification classification model is composed of a feature extraction sub-model, an instance region molecular model, a calcification region determination sub-model, and a vessel classification sub-model. The feature extraction sub-model is used to extract feature maps, the instance region molecular model is used to determine the bounding boxes of suspected calcified plaques from the feature maps, the calcification region determination sub-model is used to determine the calcified regions from the bounding boxes, and the vessel classification sub-model is used to determine the coronary vessel type label of the calcified regions.

[0114] The second determining module 430 is used to perform matrix multiplication between the image corresponding to the calcified region included in the determined target 2D coronal plain CT image and the target 2D coronal plain CT image to determine vascular calcified plaques that only include coronary artery type labels.

[0115] The calculation module 440 is used to calculate the calcification score of the vascular calcification plaques identified in each target 2D coronary plain CT image, which only includes the coronary artery type label, and to determine the total coronary artery calcification score of the patient.

[0116] Optionally, when the first determining module 420 inputs the target 2D coronal plain CT image into a pre-built calcification classification model to determine the calcification regions included in the target 2D coronal plain CT image and the coronary vessel type label of each calcification region, the first determining module 420 is used to:

[0117] The target's 2D coronal CT image is input into the calcification classification model. Feature image extraction is performed through the feature extraction sub-model in the calcification classification model to determine the feature map of the target's 2D coronal CT image.

[0118] The feature map of the target 2D coronal CT image is input into the instance region molecular model in the calcification classification model, and bounding box regression processing is performed to obtain the target bounding box of at least one suspected calcification plaque included in the target 2D coronal CT image; wherein, the target bounding box has multiple pixels; wherein, the target bounding box is a rectangular bounding box;

[0119] The feature map of the target 2D coronal CT image and the target bounding box of the at least one suspected calcified plaque are input one by one into the calcification region determination sub-model in the calcification classification model. It is determined whether the pixel in each target bounding box is a calcified pixel, and the calcification image of the calcification region included in the target bounding box is determined based on the determined calcified pixels.

[0120] The calcification image, the target bounding box, and the feature map of the target 2D coronary CT image are input into the blood vessel classification sub-model in the calcification classification model to determine the coronary blood vessel type and determine the coronary blood vessel type label for each calcification region.

[0121] Optionally, when the calculation module 440 is used to calculate the total coronary artery calcium score of the patient by performing calcium score calculation on vascular calcification plaques that include only coronary artery type labels identified in each target 2D coronary plain CT image, the calculation module 440 is used to:

[0122] For each target 2D coronal plain CT image, the calcification integral of each vascular calcification plaque in the target 2D coronal plain CT image is determined based on the vascular calcification plaques included in the target 2D coronal plain CT image.

[0123] The calcium scores of each coronary artery with the same coronary vessel type label in each target 2D coronary plain CT image are summed to determine the calcium score of each coronary vessel.

[0124] The total coronary artery calcium score of the patient is determined by summing the calcium scores of each coronary artery.

[0125] Optionally, the calcification integral includes the Agassone integral, the volume integral, and the mass integral.

[0126] Optional, such as Figure 5 As shown, the computing device 400 further includes a model building module 450, which is used for:

[0127] Acquire multiple 3D coronary artery plain CT images to be trained, and for each 3D coronary artery plain CT image to be trained, acquire multiple 2D coronary plane plain CT images to be trained with calcification information tags.

[0128] For each 2D coronary CT image to be trained in each 3D coronary artery plain CT image to be trained, the training 2D coronary CT image is input into the feature extraction neural network to obtain the training feature map of the training 2D coronary CT image.

[0129] The feature map to be trained is input into the first convolutional neural network to predict the bounding box of the suspected calcified plaque, determine the predicted bounding box of the suspected calcified plaque in the 2D coronal CT image to be trained, and determine the value of the first loss function based on the predicted bounding box and the calcification information label.

[0130] The training feature map and predicted bounding box of the 2D coronal CT image to be trained are input into the second convolutional neural network to predict whether all pixels within the predicted bounding box are calcified pixels. Based on the prediction results, the calcified region included in the predicted bounding box is determined. Based on the determined calcified region included in the predicted bounding box and the calcification information label, the value of the second loss function is determined.

[0131] The training feature map, predicted bounding box, and calcified region included in the predicted bounding box of the 2D coronary CT image to be trained are input into the third convolutional neural network to predict the coronary artery type label corresponding to the calcified region. The value of the third loss function is determined based on the prediction result of the coronary artery type label and the calcification information label.

[0132] Based on the first loss function value, the second loss function value, and the third loss function value, the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are iteratively updated simultaneously.

[0133] Once the network parameters of each neural network converge, the training ends, and the sub-models corresponding to each neural network are obtained. The calcification classification model is then constructed based on the obtained sub-models corresponding to each neural network.

[0134] Optionally, when the model building module 450 iteratively updates the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network based on the first loss function value, the second loss function value, and the third loss function value, the model building module 450 is used to:

[0135] The first loss function value, the second loss function value, and the third loss function value are summed to determine the total loss function value.

[0136] Based on the total loss function value, the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are iteratively updated simultaneously.

[0137] Optionally, the calcification information label includes a true calcification region annotation box, a labeling result indicating whether each pixel in the calcification region annotation box is a calcified pixel, and a true coronary artery type label corresponding to the calcification region formed by all calcified pixels in the same calcification region annotation box.

[0138] Please see Figure 6 , Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 6 As shown, the electronic device 600 includes a processor 610, a memory 620, and a bus 630.

[0139] The memory 620 stores machine-readable instructions executable by the processor 610. When the electronic device 600 is running, the processor 610 and the memory 620 communicate via the bus 630. When the machine-readable instructions are executed by the processor 610, they can perform the operations described above. Figure 1 The steps of the calculation method in the illustrated method embodiment can be found in the method embodiment for specific implementation, and will not be repeated here.

[0140] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can perform the above-described actions. Figure 1 The steps of the calculation method in the illustrated method embodiment can be found in the method embodiment for specific implementation, and will not be repeated here.

[0141] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0142] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0143] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0144] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0145] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0146] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for calculating coronary artery calcium integral, characterized in that, The calculation method includes: Multiple target 2D coronary CT images were obtained based on the patient's target 3D coronary artery plain CT images. For each target 2D coronal CT image, the image is input into a pre-constructed calcification classification model to determine the calcified regions and coronary vessel type labels for each calcified region. The calcification classification model is composed of a feature extraction sub-model, an instance region molecular model, a calcification region determination sub-model, and a vessel classification sub-model. The feature extraction sub-model extracts feature maps, which are then used as the instance region molecular model, the calcification region determination sub-model, and the vessel classification sub-model, respectively. The input to the classification sub-model; the instance region molecular model is used to determine the bounding box of suspected calcified plaques from the feature map, and the bounding box output by the instance region molecular model is also used as the input to the calcification region determination sub-model and the vessel classification sub-model; the calcification region determination sub-model is used to determine the calcified region from the bounding box based on the feature map and the bounding box, and the calcified region output by the calcification region determination sub-model is also used as the input to the vessel classification sub-model; the vessel classification sub-model is used to determine the coronary vessel type label of the calcified region based on the feature map, the bounding box, and the calcified region; The image corresponding to the calcified region in the identified 2D coronal CT image of the target is multiplied by a matrix to identify vascular calcified plaques that only include coronary artery type labels. The image corresponding to the calcified region in the identified 2D coronal CT image of the target, determined by the calcification classification model, is a black and white image that only includes the calcified region, where the pixel value of the calcified region is 1 and the pixel value of other parts is 0. For each target 2D coronary plain CT image, calcification plaques that include only coronary artery type labels are identified, and calcification scores are calculated to determine the patient's total coronary artery calcification score.

2. The calculation method according to claim 1, characterized in that, The step of inputting the target 2D coronal CT image into a pre-built calcification classification model to determine the calcification regions included in the target 2D coronal CT image and the coronary vessel type label for each calcification region includes: The target's 2D coronal CT image is input into the calcification classification model. Feature image extraction is performed through the feature extraction sub-model in the calcification classification model to determine the feature map of the target's 2D coronal CT image. The feature map of the target 2D coronal CT image is input into the instance region molecular model in the calcification classification model, and bounding box regression processing is performed to obtain the target bounding box of at least one suspected calcification plaque included in the target 2D coronal CT image; wherein, the target bounding box has multiple pixels; wherein, the target bounding box is a rectangular bounding box; The feature map of the target 2D coronal CT image and the target bounding box of the at least one suspected calcified plaque are input one by one into the calcification region determination sub-model in the calcification classification model. It is determined whether the pixel in each target bounding box is a calcified pixel, and the calcification image of the calcification region included in the target bounding box is determined based on the determined calcified pixels. The calcification image, the target bounding box, and the feature map of the target 2D coronary CT image are input into the blood vessel classification sub-model in the calcification classification model to determine the coronary blood vessel type and determine the coronary blood vessel type label for each calcification region.

3. The calculation method according to claim 1, characterized in that, The calcification score of the vascular calcification plaques identified in each target 2D coronary plain CT image, which includes only coronary artery type labels, is calculated to determine the patient's total coronary artery calcification score, including: For each target 2D coronal plain CT image, the calcification integral of each vascular calcification plaque in the target 2D coronal plain CT image is determined based on the vascular calcification plaques included in the target 2D coronal plain CT image. The calcium scores of each coronary artery with the same coronary vessel type label in each target 2D coronary plain CT image are summed to determine the calcium score of each coronary vessel. The total coronary artery calcium score of the patient is determined by summing the calcium scores of each coronary artery.

4. The calculation method according to claim 3, characterized in that, The calcification integral includes the Agassone integral, the volume integral, and the mass integral.

5. The calculation method according to claim 1, characterized in that, The calcification classification model is constructed using the following steps: Acquire multiple 3D coronary artery plain CT images to be trained, and for each 3D coronary artery plain CT image to be trained, acquire multiple 2D coronary plane plain CT images to be trained with calcification information tags. For each 2D coronary CT image to be trained in each 3D coronary artery plain CT image to be trained, the training 2D coronary CT image is input into the feature extraction neural network to obtain the training feature map of the training 2D coronary CT image. The feature map to be trained is input into the first convolutional neural network to predict the bounding box of the suspected calcified plaque, determine the predicted bounding box of the suspected calcified plaque in the 2D coronal CT image to be trained, and determine the value of the first loss function based on the predicted bounding box and the calcification information label. The training feature map and predicted bounding box of the 2D coronal CT image to be trained are input into the second convolutional neural network to predict whether all pixels within the predicted bounding box are calcified pixels. Based on the prediction results, the calcified region included in the predicted bounding box is determined. Based on the determined calcified region included in the predicted bounding box and the calcification information label, the value of the second loss function is determined. The training feature map, predicted bounding box, and calcified region included in the predicted bounding box of the 2D coronary CT image to be trained are input into the third convolutional neural network to predict the coronary artery type label corresponding to the calcified region. The value of the third loss function is determined based on the prediction result of the coronary artery type label and the calcification information label. Based on the first loss function value, the second loss function value, and the third loss function value, the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are iteratively updated simultaneously. Once the network parameters of each neural network converge, the training ends, and the sub-models corresponding to each neural network are obtained. The calcification classification model is then constructed based on the obtained sub-models corresponding to each neural network.

6. The calculation method according to claim 5, characterized in that, The step of iteratively updating the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network based on the first loss function value, the second loss function value, and the third loss function value includes: The first loss function value, the second loss function value, and the third loss function value are summed to determine the total loss function value. Based on the total loss function value, the network parameters of the feature extraction neural network, the first convolutional neural network, the second convolutional neural network, and the third convolutional neural network are iteratively updated simultaneously.

7. The calculation method according to claim 5, characterized in that, The calcification information label includes a true calcification region annotation box, a labeling result indicating whether each pixel in the calcification region annotation box is a calcified pixel, and a true coronary artery type label corresponding to the calcification region formed by all calcified pixels in the same calcification region annotation box.

8. A device for calculating coronary artery calcium integral, characterized in that, The computing device includes: The acquisition module is used to acquire multiple target 2D coronary CT images based on the patient's target 3D coronary artery plain CT images. The first determination module is used to input each target 2D coronal CT image into a pre-constructed calcification classification model to determine the calcification regions included in the target 2D coronal CT image and the coronary vessel type label of each calcification region. The calcification classification model is composed of a feature extraction sub-model, an instance region molecular model, a calcification region determination sub-model, and a vessel classification sub-model. The feature extraction sub-model extracts feature maps, which are then used as the instance region molecular model and the calcification region determination sub-model, respectively. The input to the feature map and the vessel classification sub-model; the instance region molecular model is used to determine the bounding box of the suspected calcified plaque from the feature map, and the bounding box output by the instance region molecular model is also used as the input to the calcified region determination sub-model and the vessel classification sub-model; the calcified region determination sub-model is used to determine the calcified region from the bounding box based on the feature map and the bounding box, and the calcified region output by the calcified region determination sub-model is also used as the input to the vessel classification sub-model; the vessel classification sub-model is used to determine the coronary vessel type label of the calcified region based on the feature map, the bounding box and the calcified region; The second determining module is used to perform matrix multiplication between the image corresponding to the calcified region included in the determined 2D coronal plain CT image of the target and the target 2D coronal plain CT image to determine the vascular calcified plaques that only include the coronary artery type label; the image corresponding to the calcified region included in the target 2D coronal plain CT image determined by the calcification classification model is a black and white image that only includes the calcified region, where the pixel value of the calcified region is 1 and the pixel value of other parts is 0; The calculation module is used to calculate the calcification score of the vascular calcification plaques identified in each target 2D coronary plain CT image, which only includes the coronary artery type label, and to determine the total coronary artery calcification score of the patient.

9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and the machine-readable instructions are executed by the processor to perform the steps of the computation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the computation method as described in any one of claims 1 to 7.

Images