Coronary artery segmentation method, device, electronic device and storage medium

Abstract

The present application provides a coronary artery segmentation method, apparatus, electronic device, and storage medium, comprising: extracting the centerline of an acquired unilateral coronary angiography image to be segmented to obtain the centerline of the coronary artery trunk; determining the target coronary artery side to which the coronary angiography image belongs based on the centerline of the coronary artery trunk; performing channel splicing on the coronary angiography image and its two preceding and following frames to determine a first spliced ​​image; performing gradient domain processing on the coronary angiography image to obtain its gradient domain mapping, and performing channel splicing to determine a second image; inputting the first spliced ​​image into a unilateral coronary artery segmentation model corresponding to the target coronary artery side to obtain a first coronary artery image; inputting the second spliced ​​image into a hybrid coronary artery segmentation model to obtain a second coronary artery image; and fusing the first coronary artery image and the second coronary artery image to obtain a target coronary artery vessel image. Thus, the technical solution of the present application improves the accuracy of the coronary artery vessel segmentation results.

Description

Technical Field

[0001] The present application relates to the technical field of medical image processing, and in particular to a coronary artery segmentation method, device, electronic device, and storage medium. Background Art

[0002] Coronary artery disease has become a major threat to human health. Cardiovascular angiography (CAA) images have become the gold standard for diagnosis and treatment of coronary heart disease. However, due to the extremely similar shapes of blood vessels in cardiac angiography images, existing technologies make it difficult to determine the specific type of each vessel. During surgery, doctors use the naked eye to view angiography videos to locate, assess, and diagnose vascular stenosis and plaques. Based on experience, they quickly make qualitative judgments about the patient's coronary artery condition and formulate treatment plans. This direct method is significantly affected by human factors and lacks accuracy, objectivity, and consistency. Computer-assisted diagnosis and treatment systems based on angiography images can effectively assist doctors in diagnosing the condition and formulating treatment plans, and are therefore of great research significance. In CAD systems based on angiography images, vessel segmentation is a particularly critical technology, forming the foundation for techniques such as radius measurement and three-dimensional reconstruction. Therefore, accurately segmenting coronary artery images is a pressing technical challenge. Summary of the Invention

[0003] In view of this, the purpose of the present application is to provide a coronary artery segmentation method, device, electronic device and storage medium, which obtains the target coronary artery vascular image by fusing the output results of the unilateral coronary artery segmentation model and the hybrid coronary artery segmentation model, thereby improving the accuracy of the coronary artery segmentation results.

[0004] An embodiment of the present application provides a coronary artery segmentation method, the segmentation method comprising:

[0005] Acquire a unilateral coronary angiography image to be segmented and two frames of coronary angiography images before and after the unilateral coronary angiography image to be segmented;

[0006] Performing centerline extraction on the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented;

[0007] Determining the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary artery trunk; wherein the target coronary artery side is the left coronary artery or the right coronary artery;

[0008] performing channel splicing processing on the unilateral coronary angiography image to be segmented and the two frames of coronary angiography images preceding and following the unilateral coronary angiography image to be segmented to determine a first spliced ​​coronary angiography image;

[0009] performing gradient domain processing on the unilateral coronary angiography image to be segmented to obtain a gradient domain mapping image of the unilateral coronary angiography image to be segmented, and performing channel stitching processing on the unilateral coronary angiography image to be segmented using the determined gradient domain mapping image and the unilateral coronary angiography image to determine a second stitched coronary angiography image;

[0010] Inputting the first spliced ​​coronary angiography image into a pre-trained unilateral coronary artery segmentation model corresponding to the target coronary artery side to obtain a first coronary artery image;

[0011] inputting the second spliced ​​coronary angiography image into a pre-trained hybrid coronary segmentation model to obtain a second coronary artery image;

[0012] The first coronary artery image and the second coronary artery image are fused to obtain a segmented target coronary artery image.

[0013] Optionally, construct a unilateral coronary artery segmentation model by following the steps below:

[0014] Acquire multiple unilateral coronary angiography images to be trained on the same side, two frames of coronary angiography images before and after each unilateral coronary angiography image to be trained, and a coronary vessel segmentation image corresponding to each unilateral coronary angiography image to be trained;

[0015] For each unilateral coronary angiography image to be trained, performing channel splicing processing on the unilateral coronary angiography image to be trained and the two frames of coronary angiography images before and after the unilateral coronary angiography image to be trained to determine a third spliced ​​coronary angiography image;

[0016] The third spliced ​​coronary angiography image is used as input data of the first image segmentation neural network, and the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained is used as output data of the first image segmentation neural network. Model iterative training is performed. When the model converges, the training is stopped to obtain the unilateral coronary artery segmentation model.

[0017] Optionally, construct a hybrid coronary segmentation model by following the steps below:

[0018] Acquire multiple unilateral coronary angiography images to be trained and coronary vessel segmentation images corresponding to each unilateral coronary angiography image to be trained; wherein the unilateral coronary angiography images to be trained include a left coronary angiography image and a right coronary angiography image;

[0019] For each unilateral coronary angiography image to be trained, performing gradient domain processing on the image to determine a gradient domain mapping image of the unilateral coronary angiography image to be trained;

[0020] performing channel stitching processing using the to-be-trained unilateral coronary angiography image and the gradient domain mapping image of the to-be-trained unilateral coronary angiography image to determine a fourth stitched coronary angiography image;

[0021] The fourth spliced ​​coronary angiography image is used as input data of the second image segmentation neural network, and the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained is used as output data of the second image segmentation neural network. Model iterative training is performed. When the model converges, the training is stopped to obtain the hybrid coronary segmentation model.

[0022] Optionally, the extracting the centerline of the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented includes:

[0023] extracting a blood vessel skeleton from the unilateral coronary angiography image to be segmented;

[0024] The center line of the blood vessel segment included in the unilateral coronary angiography image to be segmented is extracted along the blood vessel skeleton to obtain the center line of the coronary artery trunk.

[0025] Optionally, determining the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary artery trunk includes:

[0026] According to the centerline of the coronary trunk, the starting position of the coronary blood vessel in the unilateral coronary angiography image to be segmented is determined; wherein the starting position of the coronary blood vessel is used to determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs.

[0027] Optionally, the fusing the first coronary image and the second coronary image to obtain a segmented target coronary vessel image includes:

[0028] The first coronary artery image and the second coronary artery image are used to perform pixel multiplication processing, and the image obtained after the pixel multiplication is determined as the target coronary artery image.

[0029] Optionally, the unilateral coronary artery segmentation model includes a left coronary artery segmentation model and a right coronary artery segmentation model.

[0030] The present application also provides a coronary artery segmentation device, the segmentation device comprising:

[0031] An acquisition module, configured to acquire a unilateral coronary angiography image to be segmented and two frames of coronary angiography images preceding and following the unilateral coronary angiography image to be segmented;

[0032] an extraction module, configured to extract the centerline of the unilateral coronary angiography image to be segmented, and obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented;

[0033] a determination module, configured to determine, based on the centerline of the coronary artery trunk, the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs; wherein the target coronary artery side is the left coronary artery or the right coronary artery;

[0034] a first splicing module, configured to perform channel splicing processing on the unilateral coronary angiography image to be segmented and the two preceding and succeeding frames of coronary angiography images of the unilateral coronary angiography image to be segmented, to determine a first spliced ​​coronary angiography image;

[0035] a second stitching module, configured to perform gradient domain processing on the unilateral coronary angiography image to be segmented to obtain a gradient domain mapping image of the unilateral coronary angiography image to be segmented, and perform channel stitching processing on the unilateral coronary angiography image to be segmented using the determined gradient domain mapping image to determine a second stitched coronary angiography image;

[0036] a first input module, configured to input the first spliced ​​coronary angiography image into a pre-trained unilateral coronary artery segmentation model corresponding to the target coronary artery side to obtain a first coronary artery image;

[0037] a second input module, configured to input the second spliced ​​coronary angiography image into a pre-trained hybrid coronary segmentation model to obtain a second coronary image;

[0038] A fusion module is used to fuse the first coronary artery image and the second coronary artery image to obtain a segmented target coronary artery image.

[0039] Optionally, the segmentation device further includes a first model building module, wherein the first model building module is configured to:

[0040] Acquire multiple unilateral coronary angiography images to be trained on the same side, two frames of coronary angiography images before and after each unilateral coronary angiography image to be trained, and a coronary vessel segmentation image corresponding to each unilateral coronary angiography image to be trained;

[0041] For each unilateral coronary angiography image to be trained, performing channel splicing processing on the unilateral coronary angiography image to be trained and the two frames of coronary angiography images before and after the unilateral coronary angiography image to be trained to determine a third spliced ​​coronary angiography image;

[0042] The third spliced ​​coronary angiography image is used as input data of the first image segmentation neural network, and the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained is used as output data of the first image segmentation neural network. Model iterative training is performed. When the model converges, the training is stopped to obtain the unilateral coronary artery segmentation model.

[0043] Optionally, the segmentation device further includes a second model building module, wherein the second model building module is configured to:

[0044] Acquire multiple unilateral coronary angiography images to be trained and coronary vessel segmentation images corresponding to each unilateral coronary angiography image to be trained; wherein the unilateral coronary angiography images to be trained include a left coronary angiography image and a right coronary angiography image;

[0045] For each unilateral coronary angiography image to be trained, performing gradient domain processing on the image to determine a gradient domain mapping image of the unilateral coronary angiography image to be trained;

[0046] performing channel stitching processing using the to-be-trained unilateral coronary angiography image and the gradient domain mapping image of the to-be-trained unilateral coronary angiography image to determine a fourth stitched coronary angiography image;

[0047] The fourth spliced ​​coronary angiography image is used as input data of the second image segmentation neural network, and the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained is used as output data of the second image segmentation neural network. Model iterative training is performed. When the model converges, the training is stopped to obtain the hybrid coronary segmentation model.

[0048] Optionally, when the extraction module is used to extract the centerline of the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented, the extraction module is used to:

[0049] extracting a blood vessel skeleton from the unilateral coronary angiography image to be segmented;

[0050] The center line of the blood vessel segment included in the unilateral coronary angiography image to be segmented is extracted along the blood vessel skeleton to obtain the center line of the coronary artery trunk.

[0051] Optionally, when the determination module is used to determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary artery trunk, the determination module is used to:

[0052] According to the centerline of the coronary trunk, the starting position of the coronary blood vessel in the unilateral coronary angiography image to be segmented is determined; wherein the starting position of the coronary blood vessel is used to determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs.

[0053] Optionally, when the fusion module is used to fuse the first coronary artery image and the second coronary artery image to obtain a segmented target coronary artery image, the fusion module is used to:

[0054] The first coronary artery image and the second coronary artery image are used to perform pixel multiplication processing, and the image obtained after the pixel multiplication is determined as the target coronary artery image.

[0055] Optionally, the unilateral coronary artery segmentation model includes a left coronary artery segmentation model and a right coronary artery segmentation model.

[0056] An embodiment of the present application also provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the memory communicate through the bus, and when the machine-readable instructions are executed by the processor, the steps of the above-mentioned segmentation method are performed.

[0057] An embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned segmentation method are executed.

[0058] The embodiments of the present application provide a method, device, electronic device and storage medium for segmenting coronary arteries. The segmentation method includes: obtaining a unilateral coronary angiography image to be segmented and two frames of coronary angiography images before and after the unilateral coronary angiography image to be segmented; performing centerline extraction on the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary trunk of the unilateral coronary angiography image to be segmented; determining the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary trunk; wherein the target coronary artery side is the left coronary artery or the right coronary artery; performing channel splicing processing on the unilateral coronary angiography image to be segmented and the two frames of coronary angiography images before and after the unilateral coronary angiography image to be segmented to determine the first coronary artery. a spliced ​​coronary angiography image; performing gradient domain processing on the unilateral coronary angiography image to be segmented to obtain a gradient domain mapping image of the unilateral coronary angiography image to be segmented, and using the determined gradient domain mapping image and the unilateral coronary angiography image to be segmented to perform channel splicing processing to determine a second spliced ​​coronary angiography image; inputting the first spliced ​​coronary angiography image into a pre-trained unilateral coronary artery segmentation model corresponding to the target coronary artery side to obtain a first coronary artery image; inputting the second spliced ​​coronary angiography image into a pre-trained hybrid coronary artery segmentation model to obtain a second coronary artery image; and fusing the first coronary artery image and the second coronary artery image to obtain a segmented target coronary vessel image.

[0059] In this way, the present application takes into account the differences in characteristic information of the left and right coronary arteries, and adopts a scheme of separate training and mixed training of the left and right coronary arteries. The model obtained by separate training of the left and right coronary arteries can better perform unilateral coronary artery segmentation; by introducing a gradient mapping constraint item at the channel position during the mixed training process to eliminate the accuracy problem caused by the large difference in the ratio of background pixels and blood vessel pixels, the model obtained by the mixed training can better retain the edge information of the coronary artery image, further improve the segmentation precision and accuracy of the coronary artery, and thus make the final coronary artery segmentation image obtained by fusing the segmentation results of the two models more accurate.

[0060] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0062] Figure 1 A flow chart of a coronary artery segmentation method provided in an embodiment of the present application;

[0063] Figure 2 Coronary angiography images and gradient domain mapping images provided in this application;

[0064] Figure 3 A schematic diagram of another coronary angiography image provided in this application;

[0065] Figure 4 This is one of the structural schematic diagrams of a coronary artery segmentation device provided in an embodiment of the present application;

[0066] Figure 5 This is a second structural diagram of a coronary artery segmentation device provided in an embodiment of the present application;

[0067] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for which protection is claimed, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, each other embodiment obtained by those skilled in the art without making creative work falls within the scope of protection of the present application.

[0069] Coronary artery disease has become a major threat to human health. Cardiovascular angiography (CAA) images have become the gold standard for diagnosis and treatment of coronary heart disease. However, due to the extremely similar shapes of blood vessels in cardiac angiography images, existing technologies make it difficult to determine the specific type of each vessel. During surgery, doctors use the naked eye to view angiography videos to locate, assess, and diagnose vascular stenosis and plaques. Based on experience, they quickly make qualitative judgments about the patient's coronary artery condition and formulate treatment plans. This direct method is significantly affected by human factors and lacks accuracy, objectivity, and consistency. Computer-assisted diagnosis and treatment systems based on angiography images can effectively assist doctors in diagnosing the condition and formulating treatment plans, and are therefore of great research significance. In CAD systems based on angiography images, vessel segmentation is a particularly critical technology, forming the foundation for techniques such as radius measurement and three-dimensional reconstruction. Therefore, accurately segmenting coronary artery images is a pressing technical challenge.

[0070] Existing techniques generally design filters based on the characteristics of coronary arteries to enhance vascular features and suppress background noise. These methods can be broadly categorized into threshold-based segmentation, machine learning-based vascular segmentation techniques, and model-based segmentation methods.

[0071] Among them, threshold-based segmentation methods first enhance the vascular structure and then use different threshold strategies for segmentation. The most common approach is to manually design a specific filter so that after filtering the vascular image, tubular structures are enhanced and non-tubular structures are suppressed. Machine learning-based vascular segmentation technology mainly treats vascular segmentation as a binary classification problem. It achieves segmentation by manually selecting features or acquiring features through deep learning, and combining them with corresponding classifiers to classify foreground and background.

[0072] However, due to the low contrast, uneven distribution of contrast agents, and high noise levels of angiographic images, threshold-based segmentation methods cannot effectively distinguish coronary artery structures from background areas in angiographic images. Because the shapes of blood vessels in cardiac angiographic images are extremely similar, the vascular background area contains many pseudo-vascular structures that resemble coronary artery structures. Machine learning-based methods often struggle to effectively distinguish these structures from coronary artery structures.

[0073] Model-based segmentation and recognition methods fall into two main categories: single-frame segmentation and multi-frame segmentation. CN108830155B selects any frame from a segmented cardiac angiography DICOM video as a training sample and then uses pyramid fusion to learn cardiovascular feature information at different scales to improve segmentation accuracy. Wang et al. extract temporal information from video sequences using a 3D convolutional layer and then use a 2D CE-Net to segment the image sequences, achieving good segmentation results in poor-quality coronary angiography video sequences.

[0074] However, because angiographic images are a temporally continuous series of images, the single-frame solution in CN108830155B cannot effectively eliminate the low signal-to-noise ratio problem caused by low illumination, and does not make good use of the temporal dimension of the video (for example, blood vessels that block each other in one image may be separated in another). Wang et al.'s 3D convolution + 2D convolution operation is redundant, and the computational effort is disproportionate to the performance.

[0075] Based on this, the embodiments of the present application provide a coronary artery segmentation method, device, electronic device and storage medium, which obtain the target coronary artery vascular image by fusing the output results of the unilateral coronary artery segmentation model and the hybrid coronary artery segmentation model, thereby improving the accuracy of the coronary artery segmentation results.

[0076] See also Figure 1 , Figure 1 This is a flow chart of a coronary artery segmentation method provided in an embodiment of the present application. Figure 1 As shown in , the segmentation method provided in the embodiment of the present application includes:

[0077] S101 , obtaining a unilateral coronary angiography image to be segmented and two frames of coronary angiography images preceding and following the unilateral coronary angiography image to be segmented.

[0078] It should be noted that angiographic images are medical images taken after a patient is injected with a vascular contrast agent. Generally, the medical angiographic images taken are 3D images, that is, images composed of multiple frames of 2D images.

[0079] Here, the acquired unilateral coronary angiography image to be segmented is a frame of left coronary angiography image or a frame of right coronary angiography image.

[0080] Typically, a single frame of angiography image with the highest contrast agent content (i.e., the clearest) is selected from a set of coronary angiography images on a particular side of the patient as the unilateral coronary angiography image to be segmented. The angiography image selection can be performed using a pre-trained image selection model or by a professional. The image selection model can determine the image with the highest contrast agent content from multiple angiography frames.

[0081] It should also be noted that the two preceding and following frames of the unilateral coronary angiography image to be segmented are selected for the following reasons: First, blood vessels in angiography images have complex shapes and are easily deformed. Blood vessels are tubular and curved, and some may obstruct, overlap, or entangle with each other, obscuring the semantic information in the image. Second, angiography images contain not only blood vessels but also other organs and tissues. Even worse, some tissues have shapes and grayscale values ​​similar to those of blood vessels, making accurate object extraction even more difficult. Furthermore, to minimize X-ray damage, illumination is reduced by lowering the image's signal-to-noise ratio. To address these challenges, considering that coronary angiography videos are a temporally continuous sequence of images rather than a single image, combining and processing several consecutive image frames may provide insights and solutions. For example, blood vessels that obstruct each other in one image may be separated in another. Overlaying multiple images can eliminate the low signal-to-noise ratio issue caused by low illumination. Therefore, the two preceding and following coronary angiography frames are also acquired.

[0082] S102 , performing centerline extraction on the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented.

[0083] In one embodiment provided in the present application, the centerline extraction of the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary trunk of the unilateral coronary angiography image to be segmented includes: extracting a vascular skeleton from the unilateral coronary angiography image to be segmented; and extracting the centerline of the vascular segment included in the unilateral coronary angiography image to be segmented along the vascular skeleton to obtain the centerline of the coronary trunk.

[0084] Here, when extracting the centerline of the coronary trunk from the unilateral coronary angiography image to be segmented, the coronary vascular skeleton is first extracted from the unilateral coronary angiography image to be segmented, and then the skeleton is refined based on the extracted coronary vascular skeleton to determine the centerline of the coronary trunk, thereby completing the extraction of the centerline.

[0085] S103 . Determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary artery trunk.

[0086] Here, the target coronary artery side is the left coronary artery or the right coronary artery.

[0087] In one embodiment provided in the present application, determining the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary trunk includes: determining the starting position of the coronary artery in the unilateral coronary angiography image to be segmented based on the centerline of the coronary trunk; wherein the starting position of the coronary artery is used to determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs.

[0088] It should be noted that after determining the centerline of the coronary artery trunk, the starting point of the coronary artery can be automatically identified, thereby determining the target coronary artery side of the unilateral coronary angiography image to be segmented. When it is determined that the starting point of the coronary artery is located on the left side of the image, the target coronary artery side of the unilateral coronary angiography image to be segmented is determined to be the left coronary artery; when it is determined that the starting point of the coronary artery is located on the right side of the image, the target coronary artery side of the unilateral coronary angiography image to be segmented is determined to be the right coronary artery.

[0089] Here, the reason why it is necessary to determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs is to subsequently determine the required unilateral coronary artery segmentation model.

[0090] S104 , performing channel stitching processing on the unilateral coronary angiography image to be segmented and the two frames of coronary angiography images preceding and following the unilateral coronary angiography image to be segmented, to determine a first stitched coronary angiography image.

[0091] Here, by performing channel stitching on the images, the correlation between images can be better learned. This is because the stitching on the channel position can convert the correlation between frames into the correlation between channels. A good explanation is that natural images are composed of three RGB images. R, G and B are all different representations of the same image. Similarly, the difference between frames in the coronary angiography sequence is the same object at different time points. In this way, channel stitching can reduce the amount of calculation while helping to improve segmentation accuracy.

[0092] S105. Perform gradient domain processing on the unilateral coronary angiography image to be segmented to obtain a gradient domain mapping image of the unilateral coronary angiography image to be segmented, and perform channel stitching processing on the unilateral coronary angiography image to be segmented using the determined gradient domain mapping image to determine a second stitched coronary angiography image.

[0093] Here, the gradient domain processing is performed on the segmented unilateral coronary angiography image in order to better preserve the edge features of the coronary vessels in the image.

[0094] For examples, see Figure 2 , Figure 2 The coronary angiography image and its gradient domain mapping provided by this application. Figure 2 As shown, the left side is the unilateral coronary angiography image to be segmented, and the right side is the gradient domain mapping image of the unilateral coronary angiography image to be segmented.

[0095] S106: Input the first spliced ​​coronary angiography image into a pre-trained unilateral coronary artery segmentation model corresponding to the target coronary artery side to obtain a first coronary artery image.

[0096] Here, the unilateral coronary artery segmentation model includes a left coronary artery segmentation model and a right coronary artery segmentation model, and the inputting of the first spliced ​​coronary angiography image into the pre-trained unilateral coronary artery segmentation model corresponding to the target coronary artery side includes: when it is determined that the target coronary artery side is the left coronary artery, the first spliced ​​coronary angiography image is input into the pre-trained left coronary artery segmentation model; when it is determined that the target coronary artery side is the right coronary artery, the first spliced ​​coronary angiography image is input into the pre-trained right coronary artery segmentation model.

[0097] Among them, the first coronary artery image determined by the left coronary artery segmentation model is an image that only includes the left coronary artery. The three main blood vessels, namely the left anterior descending artery, the left circumflex artery, and the left main trunk, can be specifically marked in the first coronary artery image, and other left branch vessels can also be specifically marked.

[0098] The first coronary artery image determined by the right coronary artery segmentation model is an image including only the right coronary artery. In the first coronary artery image, a main right coronary artery can be specifically marked, and other right-side branch vessels can also be specifically marked.

[0099] In one embodiment provided in the present application, a unilateral coronary artery segmentation model is constructed by the following steps: obtaining multiple unilateral coronary angiography images to be trained on the same side, the two frames of coronary angiography images before and after each unilateral coronary angiography image to be trained, and a coronary vessel segmentation image corresponding to each unilateral coronary angiography image to be trained; for each unilateral coronary angiography image to be trained, performing channel splicing processing using the unilateral coronary angiography image to be trained and the two frames of coronary angiography images before and after the unilateral coronary angiography image to be trained to determine a third spliced ​​coronary angiography image; using the third spliced ​​coronary angiography image as input data of a first image segmentation neural network, and using the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained as output data of the first image segmentation neural network, performing model iterative training, and stopping training when the model converges to obtain the unilateral coronary artery segmentation model.

[0100] Here, multiple unilateral coronary angiography images to be trained on the same side are obtained, specifically including: when it is necessary to build a left coronary artery segmentation model, multiple left coronary angiography images to be trained are obtained; when it is necessary to build a right coronary artery segmentation model, multiple right coronary angiography images to be trained are obtained.

[0101] Wherein, each frame of the unilateral coronary angiography image to be trained is an angiography image filled with contrast agent. The first image segmentation neural network can adopt a Unet neural network.

[0102] For examples, see Figure 3 , Figure 3 This is a schematic diagram of another coronary angiography image provided in this application. Figure 3 As shown, the image on the left is a coronary angiography image corresponding to a state where the contrast agent is not filled, and the image on the right is a coronary angiography image corresponding to a state where the contrast agent is filled.

[0103] In this way, by training the left and right coronary artery segmentation models separately, the right coronary artery segmentation model only processes the right coronary angiography image, better fitting the characteristic information of the right coronary artery; the left coronary artery segmentation model only processes the left coronary angiography image, better fitting the characteristic information of the right coronary artery, thereby improving the accuracy of the coronary artery segmentation results.

[0104] It should be noted that the general processing schemes of the left coronary artery segmentation model and the right coronary artery segmentation model are the same, and the initial first image segmentation neural networks corresponding to the two models are the same. However, since they are two identical networks, their segmentation capabilities of the left and right coronary arteries for the given data sets are different.

[0105] S107 . Input the second spliced ​​coronary angiography image into a pre-trained hybrid coronary segmentation model to obtain a second coronary image.

[0106] Here, the hybrid coronary segmentation model can perform image segmentation on both the left coronary angiography image and the right coronary angiography image.

[0107] In one embodiment provided in the present application, a hybrid coronary artery segmentation model is constructed by the following steps: obtaining multiple unilateral coronary angiography images to be trained and coronary vessel segmentation images corresponding to each unilateral coronary angiography image to be trained; wherein the unilateral coronary angiography images to be trained include left coronary angiography images and right coronary angiography images; for each unilateral coronary angiography image to be trained, performing gradient domain processing on the image to determine a gradient domain mapping image of the unilateral coronary angiography image to be trained; performing channel stitching processing using the unilateral coronary angiography image to be trained and the gradient domain mapping image of the unilateral coronary angiography image to be trained to determine a fourth stitched coronary angiography image; using the fourth stitched coronary angiography image as input data of a second image segmentation neural network, using the coronary vessel segmentation images corresponding to the unilateral coronary angiography images to be trained as output data of the second image segmentation neural network, performing model iterative training, and stopping training when the model converges to obtain the hybrid coronary artery segmentation model.

[0108] Here, the constructed hybrid coronary artery segmentation model is used to learn some potential correlations between the left and right coronary arteries as a whole. Therefore, the multiple unilateral coronary angiography images to be trained include left coronary angiography images and right coronary angiography images.

[0109] Among them, the second image segmentation neural network can also adopt the Unet neural network, that is, the first image segmentation neural network and the second image segmentation neural network can be the same neural network, or the network parameters of the second image segmentation neural network directly adopt the neural network of the trained left and right coronary artery segmentation models, thereby simplifying the training process of the hybrid coronary artery segmentation model and reducing the training time.

[0110] S108: Fusing the first coronary artery image and the second coronary artery image to obtain a segmented target coronary artery image.

[0111] Here, the determined first coronary artery image can well preserve text information, and the determined second coronary artery image can better preserve texture information of the coronary artery edge.

[0112] In one embodiment provided in the present application, the fusing of the first coronary image and the second coronary image to obtain a segmented target coronary vessel image includes: performing pixel multiplication processing on the first coronary image and the second coronary image, and determining the image obtained after the pixel multiplication as the target coronary vessel image.

[0113] Here, when performing image fusion, pixel-level image fusion is used. In addition, other image fusion methods can also be used for fusion, such as feature-level image fusion and decision-level pixel fusion, which are not limited here.

[0114] An embodiment of the present application provides a method for segmenting coronary arteries, the method comprising: acquiring a unilateral coronary angiography image to be segmented and two frames of coronary angiography images before and after the unilateral coronary angiography image to be segmented; performing centerline extraction on the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary trunk of the unilateral coronary angiography image to be segmented; determining a target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary trunk; wherein the target coronary artery side is the left coronary artery or the right coronary artery; performing channel splicing processing on the unilateral coronary angiography image to be segmented and the two frames of coronary angiography images before and after the unilateral coronary angiography image to be segmented to determine a first spliced ​​coronary artery angiography image; performing gradient domain processing on the unilateral coronary angiography image to be segmented to obtain a gradient domain mapping image of the unilateral coronary angiography image to be segmented, and using the determined gradient domain mapping image and the unilateral coronary angiography image to be segmented to perform channel stitching processing to determine a second stitched coronary angiography image; inputting the first stitched coronary angiography image into a pre-trained unilateral coronary artery segmentation model corresponding to the target coronary artery side to obtain a first coronary artery image; inputting the second stitched coronary angiography image into a pre-trained hybrid coronary artery segmentation model to obtain a second coronary artery image; and fusing the first coronary artery image and the second coronary artery image to obtain a segmented target coronary vessel image.

[0115] In this way, the present application takes into account the differences in characteristic information of the left and right coronary arteries, and adopts a scheme of separate training and mixed training of the left and right coronary arteries. The model obtained by separate training of the left and right coronary arteries can better perform unilateral coronary artery segmentation; by introducing a gradient mapping constraint item at the channel position during the mixed training process to eliminate the accuracy problem caused by the large difference in the ratio of background pixels and blood vessel pixels, the model obtained by the mixed training can better retain the edge information of the coronary artery image, further improve the segmentation precision and accuracy of the coronary artery, and thus make the final coronary artery segmentation image obtained by fusing the segmentation results of the two models more accurate.

[0116] See also Figure 4 、 Figure 5 , Figure 4 This is one of the structural diagrams of a coronary artery segmentation device provided in an embodiment of the present application. Figure 5 This is a second structural diagram of a coronary artery segmentation device provided in an embodiment of the present application. Figure 4 As shown in , the segmentation device 400 includes:

[0117] An acquisition module 401 is configured to acquire a unilateral coronary angiography image to be segmented and two preceding and following frames of the unilateral coronary angiography image to be segmented;

[0118] An extraction module 402 is configured to extract the centerline of the unilateral coronary angiography image to be segmented, and obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented;

[0119] A determination module 403 is configured to determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary artery trunk; wherein the target coronary artery side is the left coronary artery or the right coronary artery;

[0120] A first stitching module 404 is configured to perform channel stitching processing on the unilateral coronary angiography image to be segmented and the two preceding and succeeding frames of the coronary angiography image to be segmented, to determine a first stitched coronary angiography image;

[0121] A second stitching module 405 is configured to perform gradient domain processing on the unilateral coronary angiography image to be segmented to obtain a gradient domain mapping image of the unilateral coronary angiography image to be segmented, and perform channel stitching processing on the unilateral coronary angiography image to be segmented using the determined gradient domain mapping image to determine a second stitched coronary angiography image.

[0122] A first input module 406 is configured to input the first spliced ​​coronary angiography image into a pre-trained unilateral coronary artery segmentation model corresponding to the target coronary artery side to obtain a first coronary artery image;

[0123] A second input module 407 is configured to input the second spliced ​​coronary angiography image into a pre-trained hybrid coronary segmentation model to obtain a second coronary image;

[0124] The fusion module 408 is configured to fuse the first coronary artery image and the second coronary artery image to obtain a segmented target coronary artery image.

[0125] Optional, such as Figure 5 As shown, the segmentation device 400 further includes a first model building module 409, and the first model building module 409 is used to:

[0126] Acquire multiple unilateral coronary angiography images to be trained on the same side, two frames of coronary angiography images before and after each unilateral coronary angiography image to be trained, and a coronary vessel segmentation image corresponding to each unilateral coronary angiography image to be trained;

[0127] For each unilateral coronary angiography image to be trained, performing channel splicing processing on the unilateral coronary angiography image to be trained and the two frames of coronary angiography images before and after the unilateral coronary angiography image to be trained to determine a third spliced ​​coronary angiography image;

[0128] The third spliced ​​coronary angiography image is used as input data of the first image segmentation neural network, and the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained is used as output data of the first image segmentation neural network. Model iterative training is performed. When the model converges, the training is stopped to obtain the unilateral coronary artery segmentation model.

[0129] Optionally, the segmentation device 400 further includes a second model building module 410, and the second model building module 410 is configured to:

[0130] Acquire multiple unilateral coronary angiography images to be trained and coronary vessel segmentation images corresponding to each unilateral coronary angiography image to be trained; wherein the unilateral coronary angiography images to be trained include a left coronary angiography image and a right coronary angiography image;

[0131] For each unilateral coronary angiography image to be trained, performing gradient domain processing on the image to determine a gradient domain mapping image of the unilateral coronary angiography image to be trained;

[0132] performing channel stitching processing using the to-be-trained unilateral coronary angiography image and the gradient domain mapping image of the to-be-trained unilateral coronary angiography image to determine a fourth stitched coronary angiography image;

[0133] The fourth spliced ​​coronary angiography image is used as input data of the second image segmentation neural network, and the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained is used as output data of the second image segmentation neural network. Model iterative training is performed. When the model converges, the training is stopped to obtain the hybrid coronary segmentation model.

[0134] Optionally, when the extraction module 402 is used to extract the centerline of the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented, the extraction module 402 is used to:

[0135] extracting a blood vessel skeleton from the unilateral coronary angiography image to be segmented;

[0136] The center line of the blood vessel segment included in the unilateral coronary angiography image to be segmented is extracted along the blood vessel skeleton to obtain the center line of the coronary artery trunk.

[0137] Optionally, when the determination module 403 is used to determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary artery trunk, the determination module 403 is used to:

[0138] According to the centerline of the coronary trunk, the starting position of the coronary blood vessel in the unilateral coronary angiography image to be segmented is determined; wherein the starting position of the coronary blood vessel is used to determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs.

[0139] Optionally, when the fusion module 408 is used to fuse the first coronary artery image and the second coronary artery image to obtain a segmented target coronary artery image, the fusion module 408 is used to:

[0140] The first coronary artery image and the second coronary artery image are used to perform pixel multiplication processing, and the image obtained after the pixel multiplication is determined as the target coronary artery image.

[0141] Optionally, the unilateral coronary artery segmentation model includes a left coronary artery segmentation model and a right coronary artery segmentation model.

[0142] See also Figure 6 , Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 6 As shown in FIG, the electronic device 600 includes a processor 610 , a memory 620 and a bus 630 .

[0143] The memory 620 stores machine-readable instructions executable by the processor 610. When the electronic device 600 is running, the processor 610 communicates with the memory 620 via the bus 630. When the machine-readable instructions are executed by the processor 610, the above-mentioned Figure 1 The specific implementation of the steps of the segmentation method in the illustrated method embodiment can be found in the method embodiment and will not be described in detail here.

[0144] The embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program can execute the above-mentioned Figure 1 The specific implementation of the steps of the segmentation method in the illustrated method embodiment can be found in the method embodiment and will not be described in detail here.

[0145] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0146] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. There may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed may be through some communication interface, indirect coupling or communication connection of devices or units, which may be electrical, mechanical or other forms.

[0147] The units described as separate components may or may not be physically separate, and 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 these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0148] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0149] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0150] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The scope of protection of the present application is not limited thereto. Although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-mentioned embodiments within the technical scope disclosed in the present application, or perform equivalent replacements for some of the technical features thereof. These modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A coronary artery segmentation method, characterized in that: The segmentation method includes: Acquire a unilateral coronary angiography image to be segmented and two frames of coronary angiography images before and after the unilateral coronary angiography image to be segmented; Performing centerline extraction on the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented; Determining the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary artery trunk; wherein the target coronary artery side is the left coronary artery or the right coronary artery; performing channel splicing processing on the unilateral coronary angiography image to be segmented and the two frames of coronary angiography images preceding and following the unilateral coronary angiography image to be segmented to determine a first spliced ​​coronary angiography image; performing gradient domain processing on the unilateral coronary angiography image to be segmented to obtain a gradient domain mapping image of the unilateral coronary angiography image to be segmented, and performing channel stitching processing on the unilateral coronary angiography image to be segmented using the determined gradient domain mapping image and the unilateral coronary angiography image to determine a second stitched coronary angiography image; Inputting the first spliced ​​coronary angiography image into a pre-trained unilateral coronary artery segmentation model corresponding to the target coronary artery side to obtain a first coronary artery image; inputting the second spliced ​​coronary angiography image into a pre-trained hybrid coronary segmentation model to obtain a second coronary artery image; fusing the first coronary artery image and the second coronary artery image to obtain a segmented target coronary artery image; The unilateral coronary artery segmentation model is constructed by the following steps: Acquire multiple unilateral coronary angiography images to be trained on the same side, two frames of coronary angiography images before and after each unilateral coronary angiography image to be trained, and a coronary vessel segmentation image corresponding to each unilateral coronary angiography image to be trained; For each unilateral coronary angiography image to be trained, performing channel splicing processing on the unilateral coronary angiography image to be trained and the two frames of coronary angiography images before and after the unilateral coronary angiography image to be trained to determine a third spliced ​​coronary angiography image; Using the third spliced ​​coronary angiography image as input data of a first image segmentation neural network, using the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained as output data of the first image segmentation neural network, performing model iterative training, and stopping training when the model converges to obtain the unilateral coronary artery segmentation model; The hybrid coronary segmentation model is constructed by the following steps: Acquire multiple unilateral coronary angiography images to be trained and coronary vessel segmentation images corresponding to each unilateral coronary angiography image to be trained; wherein the unilateral coronary angiography images to be trained include a left coronary angiography image and a right coronary angiography image; For each unilateral coronary angiography image to be trained, performing gradient domain processing on the image to determine a gradient domain mapping image of the unilateral coronary angiography image to be trained; performing channel stitching processing using the to-be-trained unilateral coronary angiography image and the gradient domain mapping image of the to-be-trained unilateral coronary angiography image to determine a fourth stitched coronary angiography image; The fourth spliced ​​coronary angiography image is used as input data of the second image segmentation neural network, and the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained is used as output data of the second image segmentation neural network. Model iterative training is performed. When the model converges, the training is stopped to obtain the hybrid coronary segmentation model.

2. The segmentation method according to claim 1, characterized in that The step of extracting the centerline of the unilateral coronary angiography image to be segmented to obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented includes: extracting a blood vessel skeleton from the unilateral coronary angiography image to be segmented; The center line of the blood vessel segment included in the unilateral coronary angiography image to be segmented is extracted along the blood vessel skeleton to obtain the center line of the coronary artery trunk.

3. The segmentation method according to claim 1, wherein: The step of determining the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs based on the centerline of the coronary artery trunk includes: According to the centerline of the coronary trunk, the starting position of the coronary blood vessel in the unilateral coronary angiography image to be segmented is determined; wherein the starting position of the coronary blood vessel is used to determine the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs.

4. The segmentation method according to claim 1, wherein: The fusing the first coronary artery image and the second coronary artery image to obtain a segmented target coronary artery image includes: The first coronary artery image and the second coronary artery image are used to perform pixel multiplication processing, and the image obtained after the pixel multiplication is determined as the target coronary artery image.

5. The segmentation method according to claim 1, wherein: The unilateral coronary artery segmentation model includes a left coronary artery segmentation model and a right coronary artery segmentation model.

6. A coronary artery segmentation device, characterized in that: The segmentation device comprises: an acquisition module, configured to acquire a unilateral coronary angiography image to be segmented and two frames of coronary angiography images preceding and following the unilateral coronary angiography image to be segmented; an extraction module, configured to extract the centerline of the unilateral coronary angiography image to be segmented, and obtain the centerline of the coronary artery trunk of the unilateral coronary angiography image to be segmented; a determination module, configured to determine, based on the centerline of the coronary artery trunk, the target coronary artery side to which the unilateral coronary angiography image to be segmented belongs; wherein the target coronary artery side is the left coronary artery or the right coronary artery; a first splicing module, configured to perform channel splicing processing on the unilateral coronary angiography image to be segmented and the two preceding and succeeding frames of coronary angiography images of the unilateral coronary angiography image to be segmented, to determine a first spliced ​​coronary angiography image; a second stitching module, configured to perform gradient domain processing on the unilateral coronary angiography image to be segmented to obtain a gradient domain mapping image of the unilateral coronary angiography image to be segmented, and perform channel stitching processing on the unilateral coronary angiography image to be segmented using the determined gradient domain mapping image to determine a second stitched coronary angiography image; a first input module, configured to input the first spliced ​​coronary angiography image into a pre-trained unilateral coronary artery segmentation model corresponding to the target coronary artery side to obtain a first coronary artery image; a second input module, configured to input the second spliced ​​coronary angiography image into a pre-trained hybrid coronary segmentation model to obtain a second coronary image; a fusion module, configured to fuse the first coronary artery image and the second coronary artery image to obtain a segmented target coronary artery image; The segmentation device further includes a first model building module, wherein the first model building module is configured to: Acquire multiple unilateral coronary angiography images to be trained on the same side, two frames of coronary angiography images before and after each unilateral coronary angiography image to be trained, and a coronary vessel segmentation image corresponding to each unilateral coronary angiography image to be trained; For each unilateral coronary angiography image to be trained, performing channel splicing processing on the unilateral coronary angiography image to be trained and the two frames of coronary angiography images before and after the unilateral coronary angiography image to be trained to determine a third spliced ​​coronary angiography image; Using the third spliced ​​coronary angiography image as input data of a first image segmentation neural network, using the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained as output data of the first image segmentation neural network, performing model iterative training, and stopping training when the model converges to obtain the unilateral coronary artery segmentation model; The segmentation device further includes a second model building module, wherein the second model building module is configured to: Acquire multiple unilateral coronary angiography images to be trained and coronary vessel segmentation images corresponding to each unilateral coronary angiography image to be trained; wherein the unilateral coronary angiography images to be trained include a left coronary angiography image and a right coronary angiography image; For each unilateral coronary angiography image to be trained, performing gradient domain processing on the image to determine a gradient domain mapping image of the unilateral coronary angiography image to be trained; performing channel stitching processing using the to-be-trained unilateral coronary angiography image and the gradient domain mapping image of the to-be-trained unilateral coronary angiography image to determine a fourth stitched coronary angiography image; The fourth spliced ​​coronary angiography image is used as input data of the second image segmentation neural network, and the coronary vessel segmentation image corresponding to the unilateral coronary angiography image to be trained is used as output data of the second image segmentation neural network. Model iterative training is performed. When the model converges, the training is stopped to obtain the hybrid coronary segmentation model.

7. An electronic device, characterized in that: include: 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 and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the segmentation method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the segmentation method according to any one of claims 1 to 5 are executed.

Images