Coronary angiography blood vessel contour stenosis detection method, device and equipment and storage medium
Through image segmentation and neural network model, coronary vascular images are processed, and the location and degree of stenosis are accurately positioned, which solves the problem of inaccurate positioning of coronary stenosis in the prior art and achieves efficient automatic detection.
Patent Information
- Application Number
- CN202510372178.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-27
- Publication Date
- 2025-07-18
AI Technical Summary
In prior art In coronary angiography, the positioning of the coronary stenosis position is not accurate enough, and the diagnosis is time-consuming and labor-intensive, with low reliability and poor repeatability.
The image segmentation network model and image skeletonization algorithm are used to obtain the blood vessel profile diameter, combine the preset neural network model to determine the narrow position points, and output the degree of narrowness information.
It improves the positioning accuracy of the stenosis position of coronary vascular stenosis and the accuracy of judging the degree of stenosis, and realizes automatic detection of the stenosis position and degree, and is robust.
Smart Images
Figure CN120339200A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of image processing technologies, and particularly relates to a coronary angiography blood vessel contour stenosis detection method, device, equipment and storage medium. Background Art
[0002] The improvement of the living standards and the change of living patterns of human beings have led to an increasing prevalence of coronary heart disease globally, seriously threatening human health. The stenosis of coronary arteries (referred to as coronary blood vessels for short) is the main factor causing coronary heart disease. At present, coronary angiography is the gold standard for diagnosing coronary heart disease at home and abroad. Through coronary angiography, doctors observe the vascular structure, analyze the contrast of pixels, and obtain information such as the location and degree of coronary stenosis lesions through an empirical diagnosis method. This diagnosis method has disadvantages such as time-consuming, laborious, low reliability, and poor repeatability.
[0003] With the development of image processing technologies, vascular image technologies can help doctors understand the vascular status of patients and contribute to the timely detection and diagnosis of various diseases. Therefore, they have important clinical value and practical significance for doctors. Some vascular image technologies on the market use the YOLOV3 object detection deep learning technology to directly obtain the stenosis location in the original image. Although this method can obtain the stenosis location of coronary angiography, the positioning of the stenosis in the blood vessel is not accurate enough. Summary of the Invention
[0004] Embodiments of the present application disclose a coronary angiography blood vessel contour stenosis detection method, device, equipment and storage medium, which are used to improve the positioning accuracy of the stenosis location in vascular images, and can also determine the degree of stenosis, with a relatively high accuracy rate.
[0005] In a first aspect, embodiments of the present application disclose a coronary angiography blood vessel contour stenosis detection method, which may include: Obtain a to-be-processed vascular image, where the to-be-processed vascular image includes coronary blood vessels; Process the to-be-processed vascular image by using an image segmentation network model and an image skeletonization algorithm to obtain a plurality of blood vessel contour diameters corresponding to a specified blood vessel segment, where the specified blood vessel segment is part or all of the coronary blood vessels, and one blood vessel contour diameter corresponds to one position point in the specified blood vessel segment; Process the blood vessel contour diameters through a preset neural network model, and determine stenosis position points from all the position points of the specified blood vessel segment, where the stenosis position points indicate that the coronary blood vessels at their corresponding positions have been stenosed; Output the stenosis position points.
[0006] In some optional embodiments, before outputting the stenosis position points, the method further includes: Based on the stenosis position points and the diameters of the vascular profiles, stenosis degree information is obtained, and the stenosis degree information at least includes the length and / or size of the stenosis segment formed by the stenosis position points, and the stenosis rate corresponding to the stenosis segment; The output of the stenosis position points includes: Outputting the stenosis position points and the stenosis degree information.
[0007] In a second aspect, an embodiment of the present application discloses a coronary angiography vascular profile stenosis detection device, which may include: An acquisition module for acquiring a vascular image to be processed, where the vascular image to be processed includes coronary blood vessels; A first processing module for processing the vascular image to be processed by using an image segmentation network model and an image skeletonization algorithm to obtain a plurality of vascular profile diameters corresponding to a specified vascular segment, where the specified vascular segment is part or all of the coronary blood vessels, and one vascular profile diameter corresponds to one position point in the specified vascular segment; A second processing module for processing the vascular image to be processed by using an image segmentation network model and an image skeletonization algorithm to obtain a plurality of vascular profile diameters corresponding to a specified vascular segment, where the specified vascular segment is part of the coronary blood vessels, and one vascular profile diameter corresponds to one position point in the specified vascular segment; An output module for outputting the stenosis position points.
[0008] In a third aspect, an embodiment of the present application discloses an electronic device, which may include: A memory storing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute a coronary angiography vascular profile stenosis detection method disclosed in the first aspect of the embodiments of the present application.
[0009] In a fourth aspect, an embodiment of the present application discloses a computer-readable storage medium, which stores a computer program, where the computer program enables a computer to execute a coronary angiography vascular profile stenosis detection method disclosed in the first aspect of the embodiments of the present application.
[0010] In a fifth aspect, an embodiment of the present application discloses a computer program product, when the computer program product runs on a computer, enabling the computer to execute some or all of the steps of any method in the first aspect.
[0011] Sixth aspect, an embodiment of the present application discloses an application publishing platform for publishing computer program products. When the computer program product runs on a computer, the computer is caused to execute some or all of the steps of any one of the methods in the first aspect.
[0012] Compared with the prior art, the embodiments of the present application have the following beneficial effects: In the embodiments of the present application, for a to-be-processed vascular image including coronary blood vessels, an image segmentation network model and an image skeletonization algorithm are first used to process the to-be-processed vascular image to obtain a plurality of vascular contour diameters corresponding to a specified vascular segment, where the specified vascular segment is a part of the coronary blood vessels, and one vascular contour diameter corresponds to one position point in the specified vascular segment. All the vascular contour diameters are processed by a preset neural network model, and a stenosis position point is determined from all the position points in the specified vascular segment. The stenosis position point indicates that the coronary blood vessels at its corresponding position have been stenosed. Finally, the stenosis position point is output. By implementing the embodiments of the present application, it is possible to analyze and process the vascular contour diameters by using a preset neural network model, accurately locate the stenosis position point indicating that the coronary blood vessels at its corresponding position in the vascular image have been stenosed, realize the automatic detection of the stenosis position, and have good robustness. Description of the Drawings
[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.
[0014] Figure 1 It is a schematic flowchart of a coronary angiography vascular contour stenosis detection method disclosed in Embodiment 1 of the present application; Figure 2 It is a schematic flowchart of a coronary angiography vascular contour stenosis detection method disclosed in Embodiment 2 of the present application; Figure 3 It is a schematic flowchart of a coronary angiography vascular contour stenosis detection method disclosed in Embodiment 3 of the present application; Figure 4 It is a schematic flowchart of a coronary angiography vascular contour stenosis detection method disclosed in Embodiment 4 of the present application; Figure 5 It is a schematic flowchart of a coronary angiography vascular contour stenosis detection method disclosed in Embodiment 5 of the present application; Figure 6 It is a schematic structural diagram of a coronary angiography vascular contour stenosis detection device disclosed in Embodiment 1 of the present application; Figure 7Schematic structural diagram of the coronary angiography vessel contour stenosis detection device disclosed in the second embodiment of the present application; Figure 8 Schematic structural diagram of the coronary angiography vessel contour stenosis detection device disclosed in the third embodiment of the present application; Figure 9 Schematic diagram of the skeletonized contour of the specified vessel segment disclosed in the embodiment of the present application; Figure 10 Schematic diagram of the normal vector of the centerline in the specified vessel segment disclosed in the embodiment of the present application; Figure 11 Schematic diagram of the normal vector of the centerline in the specified vessel segment disclosed in the embodiment of the present application; Figure 12 Schematic network structure diagram of the FC fully connected network model disclosed in the embodiment of the present application; Figure 13 Schematic diagram of the gradient change curve of the vessel contour diameter and the vessel contour reference curve disclosed in the embodiment of the present application; Figure 14 Schematic structural diagram of the electronic device disclosed in the embodiment of the present application. Detailed implementation manners
[0015] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying 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 the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.
[0016] It should be noted that the terms "first", "second", "third", and "fourth", etc. in the description and claims of the present application are used to distinguish different objects, rather than to describe a specific order. The terms "include" and "have" and any variations thereof in the embodiments of the present application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.
[0017] With the improvement of the living standards of humans and the change of living patterns, the prevalence of coronary heart disease has been increasing year by year, and there is also a trend of getting younger. In order to help doctors more accurately locate the stenosis position of coronary arteries and determine the degree of stenosis, the embodiments of the present application disclose a method, device, equipment and storage medium for detecting the stenosis of the coronary angiography vessel contour, which can improve the positioning accuracy of the stenosis position in the vessel image, and can also determine the degree of stenosis, realize the automatic detection of the stenosis position and the degree of stenosis, and has a relatively high determination rate and good robustness.
[0018] The technical solution of the present application will be introduced in detail through specific embodiments below.
[0019] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of the method for detecting the stenosis of the coronary angiography vessel contour disclosed in the first embodiment of the present application; as Figure 1 shown, the method for detecting the stenosis of the coronary angiography vessel contour may include: 101. Obtain a to-be-processed vessel image, which contains coronary arteries.
[0020] The execution subject of the embodiments of the present application is an electronic device or a device for detecting the stenosis of the coronary angiography vessel contour, and the device for detecting the stenosis of the coronary angiography vessel contour is built in the electronic device or is another device independent of the electronic device.
[0021] Among them, the to-be-processed vessel image of the present application, that is, the coronary angiography image of the coronary arteries, contains coronary arteries in its image. For example, the to-be-processed vessel image can be obtained by performing external direct irradiation on the coronary arteries to be measured through a coronary angiography system.
[0022] 102. Process the to-be-processed vessel image by using an image segmentation network model and an image skeletonization algorithm to obtain a plurality of vessel contour diameters corresponding to a specified vessel segment, where the specified vessel segment is part or all of the coronary arteries, and one vessel contour diameter corresponds to a position point in the specified vessel segment.
[0023] In step 102, the to-be-processed vessel image is segmented by using an image segmentation network model, and then the segmented image is skeletonized by using an image skeletonization algorithm, and then a plurality of vessel contour diameters corresponding to the specified vessel segment are obtained. The specified vessel segment refers to part or all of the coronary arteries included in the to-be-processed vessel image, and one vessel contour diameter corresponds to a position point in the specified vessel segment. Optionally, the position point can be obtained on the center line of the specified vessel contour diameter. More content about the image segmentation network model and the image skeletonization algorithm will be described in detail later.
[0024] 103. Process the vascular contour diameter through a preset neural network model, and determine the stenosis position points from all the position points of the specified vascular segment. The stenosis position points indicate that the coronary blood vessels at their corresponding positions have been stenosed.
[0025] After obtaining the vascular contour diameter in step 102, by analyzing and processing all the vascular contour diameters through a preset neural network model, the stenosis position points among all the position points of the specified vascular segment can be obtained, indicating that the coronary blood vessels at their corresponding positions have been stenosed.
[0026] 104. Output the stenosis position points.
[0027] Optionally, it can be understood that the position points are represented by their corresponding coordinates in the image coordinate system. The stenosis position points can be marked in the specified vascular segment of the to-be-processed vascular image after being processed by the image segmentation network model and the image skeletonization algorithm, and then the image marking the stenosis position is output for the user's reference.
[0028] Therefore, by implementing the above embodiments, obtain the to-be-processed vascular image containing coronary blood vessels. First, use the image segmentation network model and the image skeletonization algorithm to process the to-be-processed vascular image, and obtain multiple vascular contour diameters corresponding to the specified vascular segment. The specified vascular segment is a part of the coronary blood vessels, and one vascular contour diameter corresponds to one position point in the specified vascular segment. Process all the vascular contour diameters through a preset neural network model, and determine the stenosis position points from all the position points in the specified vascular segment. The stenosis position points indicate that the coronary blood vessels at their corresponding positions have been stenosed. Finally, output the stenosis position points. By implementing the embodiments of the present application, the vascular contour diameters can be analyzed and processed by using a preset neural network model, accurately locate the stenosis position points in the vascular image indicating that the coronary blood vessels at their corresponding positions have been stenosed, realize the automatic detection of the stenosis position, and have good robustness.
[0029] Please refer to Figure 2 , Figure 2 which is a schematic flowchart of the coronary angiography vascular contour stenosis detection method disclosed in the second embodiment of the present application. As Figure 2 shown, the coronary angiography vascular contour stenosis detection method may include: 201. Obtain the to-be-processed vascular image, which contains coronary blood vessels.
[0030] The execution subject of the embodiments of the present application is an electronic device or a coronary angiography vascular contour stenosis detection device, and the coronary angiography vascular contour stenosis detection device is built in the electronic device or is another device independent of the electronic device.
[0031] 202. Use an image segmentation network model and an image skeletonization algorithm to process the blood vessel image to be processed, and obtain multiple blood vessel contour diameters corresponding to a specified blood vessel segment, where the specified blood vessel segment is part or all of the coronary blood vessels, and one blood vessel contour diameter corresponds to a position point in the specified blood vessel segment.
[0032] In step 202, use the image segmentation network model to segment the blood vessel image to be processed, then use the image skeletonization algorithm to skeletonize the segmented image, and then obtain multiple blood vessel contour diameters corresponding to the specified blood vessel segment. The specified blood vessel segment refers to part or all of the coronary blood vessels included in the blood vessel image to be processed, and one blood vessel contour diameter corresponds to a position point in the specified blood vessel segment. Optionally, the position point can be obtained on the center line of the specified blood vessel contour diameter. More content about the image segmentation network model and the image skeletonization algorithm will be described in detail later.
[0033] 203. Process the blood vessel contour diameters through a preset neural network model, and determine the stenosis position points from all the position points of the specified blood vessel segment. The stenosis position points indicate that the coronary blood vessels at their corresponding positions have been stenosed.
[0034] 204. Obtain stenosis degree information based on the stenosis position points and the blood vessel contour diameters. The stenosis degree information at least includes the length and / or size of the stenosis segment formed by the stenosis position points, and the stenosis rate corresponding to the stenosis segment.
[0035] In step 204, further obtain stenosis degree information based on the stenosis position points and the blood vessel contour diameters. The stenosis degree information at least includes the length and / or size of the stenosis segment formed by the stenosis position points, and the stenosis rate corresponding to the stenosis segment. Among them, the length and / or size of the stenosis segment can accurately indicate the degree and / or size of the stenosed coronary blood vessels in the specified blood vessel segment, and the stenosis rate is the degree of stenosis severity.
[0036] 205. Output the stenosis position points and the stenosis degree information.
[0037] It can be understood that step 205 outputs the stenosis position points and the stenosis degree information to the user, such as a doctor, so that the user can accurately know the situation of the coronary blood vessels based on the stenosis position points and the stenosis degree information, and give a more reasonable and effective treatment plan for the patient.
[0038] Optionally, it can be understood that the stenosis position points can be marked in the specified blood vessel segment of the blood vessel image to be processed after being processed by the image segmentation network model and the image skeletonization algorithm, and the stenosis degree information can be attached, and then output for the user to refer to.
[0039] Implementing the above embodiments can process the diameter of the blood vessel contour using a preset neural network model, accurately locate the stenosis position points in the blood vessel image, and further obtain stenosis degree information with a relatively high accuracy based on the stenosis position points and the blood vessel contour diameter. It can improve the positioning accuracy of the stenosis position and the determination rate of the stenosis degree in the blood vessel image, realize the automatic detection of the stenosis position and the stenosis degree, and has good robustness.
[0040] Please refer to Figure 3 , Figure 3 which is a schematic flowchart of the coronary angiography blood vessel contour stenosis detection method disclosed in Embodiment 3 of this application; as Figure 3 shown, the coronary angiography blood vessel contour stenosis detection method may include: 301. Obtain a blood vessel image to be processed, and the blood vessel image to be processed includes coronary blood vessels.
[0041] The execution subject of the embodiment of this application is an electronic device or a coronary angiography blood vessel contour stenosis detection device, and the coronary angiography blood vessel contour stenosis detection device is built in the electronic device or is other equipment independent of the electronic device.
[0042] 302. Use an image segmentation network model to perform segmentation processing on the blood vessel image to be processed to obtain a segmented image.
[0043] Optionally, step 302 may include: inputting the blood vessel image to be processed, as well as the starting point and the ending point, into the image segmentation network model, and then after being processed by the image segmentation network model, outputting a segmented image of a specified blood vessel segment, where the starting point and the ending point are the starting point and the ending point indicating the segmented image in the blood vessel image to be processed.
[0044] Among them, the image segmentation network model is pre-trained. For example, it can be a UNet image segmentation network model, which is a convolutional neural network (CNN) architecture pre-trained for image segmentation tasks, mainly including an encoder and a decoder. After the blood vessel image to be processed, the starting point and the ending point are input into the UNet image segmentation network model, in the encoder, through operations such as multi-layer convolution, activation, and max pooling of the convolutional neural network, deeper feature information is extracted from the image, and then the decoder processes the feature information extracted by the encoder, segments out the segmented image that meets the starting point and the ending point, and then outputs it. The segmented image has the same size as the blood vessel image to be processed.
[0045] 303. Use an image skeletonization algorithm to perform skeletonization processing on the specified blood vessel segment in the segmented image.
[0046] The coronary blood vessels in the segmented image are skeletonized using an image skeletonization algorithm to obtain a skeletonized contour map of the specified blood vessel segment. The skeletonized contour of the specified blood vessel segment consists of skeletonized lines, including the skeletonized contour boundary line and the skeletonized center line.
[0047] Exemplarily, as Figure 9 shown, Figure 9 is a schematic diagram of the skeletonized contour of the specified blood vessel segment disclosed in the embodiment of the present application; it should be noted that Figure 9 is a schematic diagram obtained by further performing black-and-white processing on the obtained skeletonized contour, only for better showing the specified blood vessel segment. In fact, black-and-white processing is not required in the processing flow of the embodiment of the present application. Figure 9 The skeletonized contour 91 of the specified blood vessel segment 90 in
[0048] 304. Sample position points at a preset interval distance for the skeletonized lines in the specified blood vessel segment, and determine the center line in the specified blood vessel segment according to the position points.
[0049] Sampling is performed at a preset interval distance, combined with Figure 9 , that is, sampling the obtained center line 92 by skeletonization at a preset interval distance to obtain a plurality of position points, and then using two adjacent position points for linear fitting. It can be using the least squares method for linear fitting, and then connecting the fitted straight lines based on each position point, so as to determine the center line in the specified blood vessel segment.
[0050] Optionally, the preset interval distance is set in advance, and it can be a relatively reasonable interval distance obtained according to actual data analysis, which will not be too dense, but can improve the smooth transition and stability of the center line.
[0051] 305. Obtain the normal vector of each position point according to the center line.
[0052] After obtaining the center line of the specified blood vessel segment, a plurality of normal vectors are obtained based on the center line, and the normal vector is perpendicular to the center line. In the embodiment of the present application, it is preferable to obtain a normal vector for each position point.
[0053] Exemplarily, please refer to Figure 10 , Figure 10 is a schematic diagram of the normal vector of the center line in the specified blood vessel segment disclosed in the embodiment of the present application. It should be noted that Figure 10 only some position points are given as a schematic, not all the position points sampled according to the preset interval distance of the present application, and Figure 10It is obtained after further black-and-white processing. It can be understood that the black-and-white processing does not affect the display effect, but the black-and-white processing is not required in the implementation process of the embodiments of the present application. Figure 10 In Figure 10 , 11 is the center line of the specified blood vessel segment, and 12 with an indicating arrow is the normal vector. Further explanation is as follows. Figure 10 Only the center line 11 and the normal vector 12 are displayed, and the skeletonized contour 91 is not shown.
[0054] 306. Obtain the blood vessel contour diameter according to the normal vector and the skeletonized contour of the specified blood vessel segment.
[0055] Combining the normal vector and the skeletonized contour of the specified blood vessel segment, a line segment from one side line of the skeletonized contour to the other side line is determined at each position point, and the line segment overlaps with the straight line corresponding to the normal vector of the position point. The obtained line segment is the blood vessel contour diameter of the position point.
[0056] Exemplarily, as Figure 11 shown, Figure 11 is a schematic diagram of the blood vessel contour diameter disclosed in the embodiments of the present application; it should be noted that Figure 11 is a schematic diagram after further black-and-white processing. The black-and-white processing does not affect the display effect, but the black-and-white processing is not required in the implementation process of the embodiments of the present application. 90 is the specified blood vessel segment, 11 is the center line, and 13 is the blood vessel contour diameter.
[0057] 307. Input the change curve of the blood vessel contour diameter into a preset neural network model to obtain the probability value corresponding to each output blood vessel contour diameter. The probability value is used to indicate the probability that the corresponding position point of the blood vessel contour diameter is a stenosis position point.
[0058] In step 307, the change curve of the blood vessel contour diameter is input into a preset neural network model. The preset neural network model is a fully connected network model. Each blood vessel contour diameter is extracted to obtain an output of a probability value. The probability value is used to indicate the probability that the corresponding position point of the blood vessel contour diameter is a stenosis position point, that is, to indicate whether the coronary blood vessel at this position point has been stenosed.
[0059] Therefore, the preset neural network model can classify the blood vessel contour diameter into two categories: stenosis and non-stenosis. The preset neural network model can adopt a length of 128 points, that is, 128 blood vessel contour diameters can be input and processed at one time, and the corresponding output is also a length of 128 points, that is, 128 probability values will also be correspondingly output. Each probability value corresponds to a blood vessel contour diameter. It should be noted that since the coordinates corresponding to the change curve of the input blood vessel contour diameter have the value of the blood vessel contour diameter on the vertical axis, in order to reduce the length of the coordinate system, the abscissa can be established based on the number of points of the position points on the center line. The number of points refers to the number of the position points on the center line where the position point belongs. By representing the position points according to the number of points of the position points at a certain distance, thus, there is a one-to-one correspondence between the position points, the number of points, and the blood vessel contour diameter, etc. After being input into the preset neural network model, the preset neural network model will extract the blood vessel contour diameter corresponding to the position points from the change curve of the blood vessel contour diameter. Since the number of points of the position points on the change curve of the blood vessel contour diameter may exceed 128 or be less than 128. If it exceeds 128, the sliding window method is adopted in the change curve of the blood vessel contour diameter for extraction, and each time 128 points are extracted for one processing, and then the processing of the change curve of the blood vessel contour diameter can be completed in at least one time. If it is less than 128, the change curve of the blood vessel contour diameter can be extended to 128 points by means of reflection filling. For example, if the actual number of points is only 100, after extracting the blood vessel contour diameter of the 100th position point, the blood vessel contour diameter of the first position point in the actual change curve of the blood vessel contour diameter can be retrieved and filled as the blood vessel contour diameter of the 101st position point, and the blood vessel contour diameter of the second position point in the actual change curve of the blood vessel contour diameter can be retrieved and filled as the blood vessel contour diameter of the 102nd position point, and so on, until it is filled to 128 points to obtain the corresponding probability value of 128 points, but only the probability values of the first 100 position points are taken for subsequent processing.
[0060] Therefore, the number of points in the embodiment of the present application refers to the total number of position points.
[0061] In some alternative embodiments, the above-mentioned inputting the change curve of the blood vessel contour diameter into the preset neural network model to obtain the probability value corresponding to each blood vessel contour diameter of the output includes: Taking the change curve of the blood vessel contour diameter as the input of the FC fully connected network model to map each blood vessel contour diameter to between 0 and 1 by using the FC fully connected network model to obtain a probability value. Among them, the larger the probability value, the greater the probability that the corresponding position point of the blood vessel contour diameter is a stenosis position point.
[0062] In an embodiment of the present application, preferably, the preset neural network model is an FC fully connected network model. In the FC fully connected network model, the sigmoid activation function is used to map the vascular contour diameter to between 0 and 1 to obtain a probability value. The larger the probability value, the greater the probability that the corresponding position point of the vascular contour diameter is a stenosis position point. Conversely, the smaller the probability value, the smaller the probability that the corresponding position point of the vascular contour diameter is a stenosis position point.
[0063] Further optionally, in an embodiment of the present application, a large number of change samples of the vascular contour diameter of position points marked with stenosis categories or non-stenosis categories can be analyzed and processed to obtain a preset value for effectively comparing the probability value with the preset value to distinguish whether the position point is a stenosis position point.
[0064] Exemplarily, as Figure 12 shown, Figure 12 is a schematic diagram of the network structure of the FC fully connected network model disclosed in an embodiment of the present application. The input is a point length of 128, passes through a point length of 512 in the middle, then a point length of 256, and finally outputs a point length of 128.
[0065] 308. Determine the position points with a probability value greater than the preset value from all the position points of the specified vascular segment as the stenosis position points.
[0066] In step 308, the probability value obtained through step 207 is used to determine the stenosis position points from all the position points on the center line of the specified vascular segment in the image, and the stenosis position points can be indicated in the skeletonized contour diagram of the specified vascular segment, or in the segmented image, or even in the to-be-processed vascular image, realizing the determination of the stenosis position in the image.
[0067] In an embodiment of the present application, by using the preset neural network model, a probability value with a higher accuracy can be obtained, and then combined with the preset value, the stenosis position points can be accurately identified in the specified vascular segment.
[0068] 309. Obtain the gradient change curve of the vascular contour diameter according to the stenosis position points and the vascular contour diameter.
[0069] 310. Obtain the stenosis degree information based on the gradient change curve.
[0070] In steps 309-310, the stenosis position points in the segmented image can be obtained as described above, and then combined with the stenosis position points and the vascular contour diameter. First, the gradient change curve of the vascular contour diameter is obtained, and then based on this gradient change curve, the stenosis degree information is obtained.
[0071] 311. Output the stenosis position points and stenosis degree information.
[0072] Implementing the above embodiments can process the vascular contour diameter using a preset neural network model, accurately locate the stenosis position points in the vascular image, and further obtain relatively accurate stenosis degree information based on the stenosis position points and the vascular contour diameter. It can improve the positioning accuracy of the stenosis position and the determination rate of the stenosis degree in the vascular image, realize the automatic detection of the stenosis position and stenosis degree, and has good robustness.
[0073] Please refer to Figure 4 , Figure 4 which is a schematic flow chart of the coronary angiography vascular contour stenosis detection method disclosed in the fourth embodiment of the present application; Figure 4 In the coronary angiography vascular contour stenosis detection method shown, the above step 309 can further include the following implementation steps: 401. Obtain the change curve of the vascular contour diameter according to the vascular contour diameter.
[0074] The coordinates corresponding to the change curve of the vascular contour diameter can use the position points on the center line of the specified vascular segment as the abscissa, and the ordinate is the value of the vascular contour diameter. Or, in order to reduce the length of the coordinate system, the abscissa can be established based on the number of points of the position points on the center line. The number of points refers to the nth position point where the position point belongs on the center line. By representing the position point according to the number of points of the position point at a certain distance, thus, there is a one-to-one correspondence relationship among the position point, the number of points, and the vascular contour diameter.
[0075] 402. Transform the change curve of the vascular contour diameter to obtain the gradient change curve of the vascular contour diameter. The ordinate of the coordinate system where the gradient change curve is located is the physical space diameter, and the abscissa is the position point corresponding to the vascular contour diameter. One physical space diameter corresponds to one vascular contour diameter.
[0076] By transforming the change curve of the vascular contour diameter, the vascular contour diameter is correspondingly mapped to another coordinate system. The abscissa on this coordinate system remains unchanged, and the vascular contour diameter is transformed to obtain the corresponding physical space diameter.
[0077] Through the above embodiments, the change curve of the vascular contour diameter can be transformed into a gradient transformation curve with relatively high accuracy, so as to obtain the stenosis degree information based on this gradient change curve.
[0078] Please refer to Figure 5 , Figure 5 which is a schematic flow chart of the coronary angiography vascular contour stenosis detection method disclosed in the fifth embodiment of the present application; Figure 5 In the coronary angiography vascular contour stenosis detection method shown, the above step 310 can further include the following implementation steps: 501. Determine the length and / or size of the narrow segment corresponding to the narrow position point according to the gradient change curve, where the narrow segment is the narrow interval formed by all narrow position points on the gradient change curve.
[0079] In some alternative embodiments, step 501 further includes: determining a target narrow position point on the gradient change curve according to the gradient change curve, where the target narrow position point is the narrow position point with the maximum probability value on the gradient change curve; determining a first longest continuous interval with a gradient value less than 0 in front of the target narrow position point and a second longest continuous region with a gradient value greater than 0 behind the target narrow position point; combining the first longest continuous interval and the second longest continuous interval to obtain the narrow segment; and obtaining the length and / or size of the narrow segment.
[0080] Among them, the first longest continuous interval with a gradient value less than 0 indicates that the first longest continuous interval is in a downhill state, and the second longest continuous region with a gradient value greater than 0 indicates that the second longest continuous region is in an uphill state. It should be noted that the probability value of the corresponding position point within the narrow segment should be greater than the preset value, that is, it is determined as a narrow position point. Through the above embodiments, its length and / or size can be determined.
[0081] 502. Obtain the stenosis rate corresponding to the narrow segment according to the non-narrow segment and the narrow segment of the gradient change curve, where the non-narrow segment is other line segments on the gradient change curve except the narrow segment.
[0082] Optionally, step 502 may further include: performing least-squares curve fitting on the data information of the gradient change curve segment corresponding to the non-narrow segment to obtain a reference blood vessel contour curve; and obtaining the stenosis rate corresponding to the narrow segment according to the reference blood vessel contour curve and the data information of the gradient change curve segment corresponding to the narrow segment.
[0083] Further, the above-mentioned performing least-squares curve fitting on the data information of the gradient change curve segment corresponding to the non-narrow segment to obtain a reference blood vessel contour curve includes: Extracting the first physical space diameter of each corresponding position point on the non-narrow segment; Based on the criterion of the minimum sum of squared deviations, performing curve fitting on two adjacent first physical space diameters to obtain the reference blood vessel contour curve, where each position point on the reference blood vessel contour curve corresponds to a reference physical space diameter.
[0084] In the above embodiments, the data information of the gradient change curve segment corresponding to the non-narrow segment, that is, the coordinate information of the number of points (position points) on the coordinate system, includes the first physical space diameter on the ordinate and the number of points (position points) on the abscissa, and then the least squares curve fitting is performed to obtain a straight line corresponding to a linear equation. The fitting is performed based on the criterion of the minimum sum of squared deviations.
[0085] Exemplarily, the fitting equation is: f(x) = kx + b, where k is the slope, b is a constant, x is the number of points, and the criterion for the minimum sum of squared deviations is as follows:
[0086] where y i is the first physical space diameter corresponding to the i-th number of points, L is the sum of squared deviations. The smaller the value of L, the higher the accuracy of the straight line corresponding to the linear equation obtained by fitting.
[0087] Further, obtaining the stenosis rate corresponding to the stenosis segment based on the data information of the blood vessel contour reference curve and the gradient change curve segment corresponding to the stenosis segment includes: Extracting the second physical space diameter of each corresponding position point on the stenosis segment; Calculating the difference between the reference physical space diameter and each second physical space diameter; Obtaining the target difference with the largest value among all the differences; Calculating the ratio of the target difference to the reference physical space diameter to obtain the stenosis rate corresponding to the stenosis segment.
[0088] Among them, in the above embodiments, the physical space diameter of the number of points (position points) on the stenosis segment on the vertical coordinate axis is used as the second physical space diameter, and the physical space diameter corresponding to the number of points (position points) on the blood vessel contour reference curve is used as the reference physical space diameter. Calculate the difference between each reference physical space diameter and each second physical space diameter respectively, and then obtain the difference with the largest value as the target difference. Then calculate the ratio of the target difference to the reference physical space diameter to obtain the stenosis rate corresponding to the stenosis segment.
[0089] Optionally, first calculate all the reference physical space diameters to obtain a standard physical space diameter, then calculate the difference between the standard physical space diameter and each second physical space diameter, select the difference with the largest value among them as the target difference, and then calculate the ratio of the target difference to the standard physical space diameter, multiply by 100% to obtain the stenosis rate.
[0090] Exemplarily, the calculation formula for the stenosis rate is as follows:
[0091] Wherein, S represents the stenosis rate of the stenosis segment, ref_y is the reference physical space diameter or the standard physical space diameter, and y is the second physical space diameter.
[0092] Exemplarily, please refer to Figure 13 , Figure 13 which is a schematic diagram of the gradient change curve of the blood vessel contour diameter and the blood vessel contour reference curve disclosed in the embodiment of the present application. In Figure 13 , the values on the horizontal axis represent the number of position points, and the values on the vertical axis represent the physical space diameter corresponding to the blood vessel contour diameter, with the unit of mm. Among them, the curve is the gradient change region. Among them, the lowest point between point A and point B on the horizontal axis is the point with the largest probability value, that is, the target stenosis position point. The interval between point A and the target stenosis position point is the first longest continuous interval with a gradient value less than 0, and the interval between the target stenosis position point and point B is the second longest continuous region with a gradient value greater than 0. The continuous interval from point A to point B is the stenosis segment, which also indicates the size of the stenosis segment. The dotted line represents the blood vessel contour reference curve fitted in the embodiment of the present application, and the thick black represents the non-stenosis segment. It should be noted that, in order to better distinguish the non-stenosis segment and the stenosis segment, in Figure 13 , the curve corresponding to the non-stenosis segment is further thickened, which may cause a deviation from the original curve.
[0093] 503. Obtain stenosis degree information by combining the length and / or size of the stenosis segment and the stenosis rate corresponding to the stenosis segment.
[0094] Please refer to Figure 6 , Figure 6 which is a schematic structural diagram of the coronary angiography blood vessel contour stenosis detection device disclosed in Embodiment 1 of the present application; as Figure 6 shown, the coronary angiography blood vessel contour stenosis detection device may include: An acquisition module 601, configured to acquire a blood vessel image to be processed, where the blood vessel image to be processed includes coronary blood vessels; A first processing module 602, configured to process the blood vessel image to be processed by using an image segmentation network model and an image skeletonization algorithm to obtain a plurality of blood vessel contour diameters corresponding to a specified blood vessel segment, where the specified blood vessel segment is part or all of the coronary blood vessels, and one blood vessel contour diameter corresponds to one position point in the specified blood vessel segment; A second processing module 603, configured to process the blood vessel image to be processed by using an image segmentation network model and an image skeletonization algorithm to obtain a plurality of blood vessel contour diameters corresponding to a specified blood vessel segment, where the specified blood vessel segment is part of the coronary blood vessels, and one blood vessel contour diameter corresponds to one position point in the specified blood vessel segment; An output module 604, configured to output the stenosis position point.
[0095] Please refer to Figure 7 , Figure 7 , which is a schematic structural diagram of the coronary angiography vascular contour stenosis detection device disclosed in the second embodiment of the present application; as Figure 7 shown, the coronary angiography vascular contour stenosis detection device further includes: A third processing module 705, configured to obtain stenosis degree information according to the stenosis position points and the vascular contour diameter before the output module 604 outputs the stenosis position points and the stenosis degree information, where the stenosis degree information at least includes the length and / or size of the stenosis segment formed by the stenosis position points, and the stenosis rate corresponding to the stenosis segment; Furthermore, an output module 604, configured to output the stenosis position points and the stenosis degree information.
[0096] Further, please refer to Figure 8 , Figure 8 , which is a schematic structural diagram of the coronary angiography vascular contour stenosis detection device disclosed in the third embodiment of the present application; in Figure 8 , the coronary angiography vascular contour stenosis detection device may include: The first processing module 602 may include: a segmentation sub-module 801, a skeletonization sub-module 802, and a first determination sub-module 803. The second processing module 603 may include: a second processing sub-module 804 and a second determination sub-module 805. The third processing module 705 may include: a conversion sub-module 806 and a third determination sub-module 807.
[0097] Among them, the segmentation sub-module 801 is configured to perform segmentation processing on the to-be-processed vascular image by using the image segmentation network model to obtain a segmented image; The skeletonization sub-module 802 is configured to perform skeletonization processing on a specified vascular segment in the segmented image by using an image skeletonization algorithm; The first determination sub-module 803 is configured to sample the position points of the skeletonized lines in the specified vascular segment at a preset interval distance, determine the center line in the specified vascular segment according to the position points; and, obtain a normal vector of each position point according to the center line; and, obtain the vascular contour diameter according to the normal vector and the skeletonized contour of the specified vascular segment.
[0098] In some embodiments, the second processing sub-module 804 is configured to input all the vascular contour diameters into the preset neural network model to obtain a probability value corresponding to each output vascular contour diameter, where the probability value is used to indicate the probability that the corresponding position point of the vascular contour diameter is the stenosis position point.
[0099] The second determination sub-module 805 is configured to determine the stenosis position points from all the position points of the specified blood vessel segment according to the probability value, where the position points are the position points on the center line of the specified blood vessel segment.
[0100] Further, in some alternative embodiments, the manner in which the second processing sub-module 704 is configured to input all the blood vessel contour diameters into the preset neural network model and obtain the probability value corresponding to each of the output blood vessel contour diameters specifically includes: Taking all the blood vessel contour diameters as the input of the FC fully connected network model, so as to use the FC fully connected network model to map each of the blood vessel contour diameters to a value between 0 and 1, obtaining a probability value, where the larger the probability value, the greater the probability that the corresponding position point of the blood vessel contour diameter is the stenosis position point.
[0101] Furthermore, the manner in which the second determination sub-module 805 is configured to determine the stenosis position points from all the position points of the specified blood vessel segment according to the probability value is specifically: Determining the position points with the probability value greater than a preset value from all the position points of the specified blood vessel segment as the stenosis position points.
[0102] In some embodiments, the above conversion sub-module 806 is configured to obtain the gradient change curve of the blood vessel contour diameter according to the stenosis position points and the blood vessel contour diameter; The third determination sub-module 807 is configured to obtain the stenosis degree information based on the gradient change curve.
[0103] In some alternative embodiments, the manner in which the conversion sub-module 806 is configured to obtain the gradient change curve of the blood vessel contour diameter according to the stenosis position points and the blood vessel contour diameter is specifically: Obtaining the change curve of the blood vessel contour diameter according to the blood vessel contour diameter; Performing a transformation on the change curve of the blood vessel contour diameter to obtain the gradient change curve of the blood vessel contour diameter, where the ordinate of the coordinate system where the gradient change curve is located is the physical space diameter, the abscissa is the position point corresponding to the blood vessel contour diameter, and one physical space diameter corresponds to one blood vessel contour diameter.
[0104] In some alternative embodiments, the manner in which the third determination sub-module 807 is configured to obtain the stenosis degree information based on the gradient change curve is specifically: Determining the length and / or size of the stenosis segment corresponding to the stenosis position points according to the gradient change curve, where the stenosis segment is the stenosis interval formed by all the stenosis position points on the gradient change curve; Obtain the stenosis rate corresponding to the stenosis segment based on the non-stenosis segment and the stenosis segment of the gradient change curve, where the non-stenosis segment is other line segments on the gradient change curve except the stenosis segment; Combine the length and / or size of the stenosis segment and the stenosis rate corresponding to the stenosis segment to obtain the stenosis degree information.
[0105] Among them, in some implementable ways, the specific way for the third determination sub-module 807 to obtain the stenosis rate corresponding to the stenosis segment according to the non-stenosis segment and the stenosis segment of the gradient change curve is as follows: Perform least squares curve fitting on the data information of the gradient change curve segment corresponding to the non-stenosis segment to obtain a blood vessel contour reference curve; Obtain the stenosis rate corresponding to the stenosis segment based on the blood vessel contour reference curve and the data information of the gradient change curve segment corresponding to the stenosis segment.
[0106] Furthermore, in some implementable ways, the specific way for the third determination sub-module 807 to determine the length and / or size of the stenosis segment corresponding to the stenosis position point according to the gradient change curve is as follows: Determine the target stenosis position point on the gradient change curve according to the gradient change curve, where the target stenosis position point is the stenosis position point with the largest probability value on the gradient change curve; Determine the first longest continuous interval with a gradient value less than 0 in front of the target stenosis position point and the second longest continuous region with a gradient value greater than 0 behind the target stenosis position point; Combine the first longest continuous interval and the second longest continuous interval to obtain the stenosis segment; Obtain the length and / or size of the stenosis segment.
[0107] Furthermore, in some implementable ways, the specific way for the third determination sub-module 807 to perform least squares curve fitting on the data information of the gradient change curve segment corresponding to the non-stenosis segment to obtain a blood vessel contour reference curve is as follows: Extract the first physical space diameter of each corresponding position point on the non-stenosis segment; Based on the criterion of the minimum sum of squared deviations, perform curve fitting on two adjacent first physical space diameters to obtain the blood vessel contour reference curve, and each position point on the blood vessel contour reference curve corresponds to a reference physical space diameter.
[0108] Further, in some implementable manners, the manner in which the third determination sub-module 807 obtains the stenosis rate corresponding to the stenosis segment according to the data information of the vascular profile reference curve and the gradient change curve segment corresponding to the stenosis segment is specifically as follows: Extract the second physical space diameter of each corresponding position point on the stenosis segment; Calculate the difference between the reference physical space diameter and each of the second physical space diameters; Obtain the target difference with the largest value among all the differences; Calculate the ratio of the target difference to the reference physical space diameter to obtain the stenosis rate corresponding to the stenosis segment.
[0109] Please refer to Figure 14 , Figure 14 which is a schematic structural diagram of the electronic device disclosed in the embodiment of the present application; Figure 14 The illustrated electronic device may include: A memory 1401 storing executable program code; A processor 1402 coupled to the memory 1401; Wherein, the processor 1402 calls the executable program code stored in the memory 1401 and executes Figures 1 to 5 Some steps of any coronary angiography vascular profile stenosis detection method.
[0110] The embodiment of the present application also discloses a computer-readable storage medium, which stores a computer program, wherein the computer program enables a computer to execute Figures 1 to 5 Any coronary angiography vascular profile stenosis detection method disclosed.
[0111] The embodiment of the present application also discloses a computer program product, when the computer program product runs on a computer, enabling the computer to execute Figures 1 to 5 Some or all steps of any method disclosed.
[0112] The embodiment of the present application also discloses an application publishing platform, the application publishing platform is used to publish a computer program product, wherein, when the computer program product runs on a computer, enabling the computer to execute Figures 1 to 5 Some or all steps of any method disclosed.
[0113] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and this program can be stored in a computer-readable storage medium. The storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disc memories, tape memories, or any other medium that can be used to carry or store data and is computer-readable.
[0114] The above has introduced in detail a coronary angiography blood vessel contour stenosis detection method, device, equipment and storage medium disclosed in the embodiments of the present application. Specific examples are used herein to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present application.
Claims
1. A method for detecting the stenosis of the coronary angiography vascular contour, characterized in that Including: Obtain a vascular image to be processed, where the vascular image to be processed includes coronary blood vessels; Process the vascular image to be processed by using an image segmentation network model and an image skeletonization algorithm to obtain a plurality of vascular contour diameters corresponding to a specified vascular segment, where the specified vascular segment is part or all of the coronary blood vessels, and one vascular contour diameter corresponds to one position point in the specified vascular segment; Process the vascular contour diameters through a preset neural network model, and determine stenosis position points from all the position points of the specified vascular segment, where the stenosis position points indicate that the corresponding positions of the coronary blood vessels are stenosed; Output the stenosis position points.
2. The method according to claim 1, characterized in that, Before outputting the stenosis position points, the method further includes: Obtain stenosis degree information according to the stenosis position points and the vascular contour diameters, where the stenosis degree information at least includes the length and / or size of the stenosis segment formed by the stenosis position points, and the stenosis rate corresponding to the stenosis segment; The outputting the stenosis position points includes: Outputting the stenosis position points and the stenosis degree information.
3. The method according to claim 1 or 2, characterized in that, The processing the vascular contour diameters through a preset neural network model and determining stenosis position points from all the position points of the specified vascular segment includes: Input the change curve of the vascular contour diameters into the preset neural network model to obtain a probability value corresponding to each output vascular contour diameter, where the probability value is used to indicate the probability that the corresponding position point of the vascular contour diameter is the stenosis position point; Determine the stenosis position points from all the position points of the specified vascular segment according to the probability values, where the position points are the position points on the center line of the specified vascular segment.
4. The method according to claim 3, wherein The inputting the change curve of the vascular contour diameters into the preset neural network model to obtain a probability value corresponding to each output vascular contour diameter includes: Taking the change curve of the vascular contour diameters as the input of the FC fully connected network model, so as to use the FC fully connected network model to map each vascular contour diameter to between 0 and 1 to obtain a probability value, where the larger the probability value, the greater the probability that the corresponding position point of the vascular contour diameter is the stenosis position point; The determining the stenosis position points from all the position points of the specified vascular segment according to the probability values includes: Determining the position points with probability values greater than a preset value from all the position points of the specified vascular segment as the stenosis position points.
5. The method according to claim 2, wherein The obtaining stenosis degree information according to the stenosis position points and the vascular contour diameters includes: Obtain the gradient change curve of the vascular contour diameters according to the stenosis position points and the vascular contour diameters; Obtain the stenosis degree information based on the gradient change curve.
6. The method according to claim 5, wherein The obtaining the stenosis degree information based on the gradient change curve includes: Determine the length and / or size of the stenosis segment corresponding to the stenosis position points according to the gradient change curve, where the stenosis segment is the stenosis interval formed by all the stenosis position points on the gradient change curve; Obtain the stenosis rate corresponding to the stenosis segment according to the non-stenosis segment and the stenosis segment of the gradient change curve, where the non-stenosis segment is other line segments on the gradient change curve except the stenosis segment; Combine the length and / or size of the stenosis segment and the stenosis rate corresponding to the stenosis segment to obtain the stenosis degree information.
7. The method according to claim 6, wherein The obtaining the stenosis rate corresponding to the stenosis segment according to the non-stenosis segment and the stenosis segment of the gradient change curve includes: Perform least squares curve fitting on the data information of the gradient change curve segment corresponding to the non-stenosis segment to obtain a blood vessel contour reference curve; Obtain the stenosis rate corresponding to the stenosis segment according to the blood vessel contour reference curve and the data information of the gradient change curve segment corresponding to the stenosis segment.
8. The method according to claim 7, wherein The obtaining the gradient change curve of the blood vessel contour diameter according to the stenosis position point and the blood vessel contour diameter includes: Obtain the change curve of the blood vessel contour diameter according to the blood vessel contour diameter; Transform the change curve of the blood vessel contour diameter to obtain the gradient change curve of the blood vessel contour diameter. The ordinate of the coordinate system where the gradient change curve is located is the physical space diameter, and the abscissa is the position point corresponding to the blood vessel contour diameter. One physical space diameter corresponds to one blood vessel contour diameter.
9. The method according to claim 8, wherein The determining the length and / or size of the stenosis segment corresponding to the stenosis position point according to the gradient change curve includes: Determine the target stenosis position point on the gradient change curve according to the gradient change curve, where the target stenosis position point is the stenosis position point with the largest probability value on the gradient change curve; Determine the first longest continuous interval with a gradient value less than 0 in front of the target stenosis position point and the second longest continuous region with a gradient value greater than 0 behind the target stenosis position point; Combine the first longest continuous interval and the second longest continuous interval to obtain the stenosis segment; Obtain the length and / or size of the stenosis segment.
10. The method according to claim 8, characterized in that The performing least squares curve fitting on the data information of the gradient change curve segment corresponding to the non-stenosis segment to obtain a blood vessel contour reference curve includes: Extract the first physical space diameter of each corresponding position point on the non-stenosis segment; Based on the criterion of the minimum sum of squared deviations, perform curve fitting on two adjacent first physical space diameters to obtain the blood vessel contour reference curve. Each position point on the blood vessel contour reference curve corresponds to a reference physical space diameter.
11. The method according to claim 10, wherein, The obtaining the stenosis rate corresponding to the stenosis segment according to the blood vessel contour reference curve and the data information of the gradient change curve segment corresponding to the stenosis segment includes: Extract the second physical space diameter of each corresponding position point on the stenosis segment; Calculate the difference between the reference physical space diameter and each second physical space diameter; Obtain the target difference with the largest value among all the differences; Calculate the ratio of the target difference to the reference physical space diameter to obtain the stenosis rate corresponding to the stenosis segment.
12. The method according to claim 1 or 2, characterized in that, Processing the to-be-processed vascular image by using the image segmentation network model and the image skeletonization algorithm to obtain a plurality of vascular contour diameters corresponding to a specified vascular segment, including: Performing segmentation processing on the to-be-processed vascular image by using the image segmentation network model to obtain a segmented image; Performing skeletonization processing on the specified vascular segment in the segmented image by using the image skeletonization algorithm; Sampling the position points at a preset interval distance along the skeletonized lines in the specified vascular segment, and determining the center line in the specified vascular segment according to the position points; Obtaining a normal vector of each of the position points according to the center line; Obtaining the vascular contour diameter according to the normal vector and the skeletonized contour of the specified vascular segment.
13. A coronary angiography vascular contour stenosis detection device, characterized in that, Including: An acquisition module, configured to acquire a to-be-processed vascular image, where the to-be-processed vascular image includes coronary blood vessels; A first processing module, configured to process the to-be-processed vascular image by using an image segmentation network model and an image skeletonization algorithm to obtain a plurality of vascular contour diameters corresponding to a specified vascular segment, where the specified vascular segment is part or all of the coronary blood vessels, and one vascular contour diameter corresponds to one position point in the specified vascular segment; A second processing module, configured to process the to-be-processed vascular image by using an image segmentation network model and an image skeletonization algorithm to obtain a plurality of vascular contour diameters corresponding to a specified vascular segment, where the specified vascular segment is part of the coronary blood vessels, and one vascular contour diameter corresponds to one position point in the specified vascular segment; An output module, configured to output the stenosis position point.
14. An electronic device, characterized in that, Including: A memory storing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory and executes the coronary angiography vascular contour stenosis detection method according to any one of claims 1-12.
15. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, the steps of the method according to any one of claims 1-12 are implemented.