Intelligent segmentation method for coronary microangiography images
By using grayscale difference calculation and overlapping heatmap technology from multiple frames of time-series angiography images, the problem of difficult segmentation of coronary microvessels in coronary angiography images has been solved, achieving accurate segmentation of microvessels and providing reliable data for clinical diagnosis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- PEKING UNIVERSITY THIRD HOSPITAL (THE THIRD CLINICAL MEDICAL SCHOOL OF PEKING UNIVERSITY)
- Filing Date
- 2026-02-06
- Publication Date
- 2026-06-26
Smart Images
Figure CN121661651B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of image processing technology, specifically to an intelligent segmentation method for coronary microvascular angiography images. Background Technology
[0002] In coronary angiography, over 50% of patients with chest pain have no stenosis in the main coronary artery but exhibit microvascular lesions, which traditional angiography cannot assess. Precise microvascular segmentation is a prerequisite for quantifying the fractional flow reserve and microvascular resistance index, and is crucial for diagnosing coronary artery disease and diabetic cardiomyopathy. Furthermore, microvessels are the final pathway for myocardial oxygen supply, and their integrity directly determines myocardial survival. Segmentation allows for analysis of diastolic / systolic microvascular grayscale changes, assessing the risk of myocardial ischemia.
[0003] Current coronary microvessel segmentation primarily relies on traditional image processing algorithms (e.g., region growing or thresholding) and shallow machine learning methods. Some studies have attempted to use basic deep learning models such as U-Net, but these depend on static single-frame image input. Existing techniques largely focus on segmenting the main coronary arteries, neglecting the characteristics of microvessels and failing to fully utilize the temporal information of contrast agent dynamic perfusion. Furthermore, due to the small diameter of coronary microvessels and the low concentration of contrast agent, they are easily interfered with by surrounding tissues, resulting in insufficient sensitivity of traditional methods to low-contrast microvessels, making it difficult to distinguish between real microvessels and noise. In other words, the representation of coronary microvessels in coronary angiography images is relatively weak and easily obscured and affected by other tissues, making it difficult to directly obtain complete microvessel images using traditional methods. Summary of the Invention
[0004] The purpose of this application is to provide an intelligent segmentation method for coronary microvascular angiography images to solve the technical problem that coronary microvessels are poorly represented in coronary angiography images, making it difficult to obtain complete images of them.
[0005] To achieve the above objectives, this application provides the following technical solution:
[0006] A method for intelligent segmentation of coronary microvascular angiography images includes:
[0007] Acquire multi-frame temporal angiography images of coronary microvessels; these multi-frame temporal angiography images are acquired during coronary microvessel angiography.
[0008] Based on each frame of temporal contrast imaging images, multiple frames of grayscale difference images are obtained; the pixel values in the grayscale difference images are equal to the difference in grayscale values of corresponding pixels in the first image and the second image; the first image is any image in each frame of temporal contrast imaging images; the second image is an image in each frame of temporal contrast imaging images that is temporally adjacent to the first image, and the temporal sequence of the second image is after the temporal sequence of the first image;
[0009] Based on the grayscale difference images of each frame, multiple first regions are obtained; the first region is the region formed by the first feature pixels in the grayscale difference image; the first feature pixel is any pixel in the grayscale difference image whose absolute value of the difference between the pixel value and the average pixel value is greater than a first preset value; the average pixel value is the average value of the corresponding pixel values of all pixels in the grayscale difference image.
[0010] Based on each first region, the microvessel probability corresponding to each first region is obtained; the microvessel probability is at least used to characterize the probability that the corresponding first region in the gray-scale difference image is a region formed by microvessels.
[0011] Based on the probability of each microvessel, multiple second regions are obtained; the second region is any region in each first region whose corresponding microvessel probability is greater than or equal to a second preset value.
[0012] The second regions in each frame of grayscale difference image are superimposed to obtain an overlapping heatmap;
[0013] Based on the overlapping heatmap, coronary microvessels in each frame of time-series angiography images are segmented and labeled.
[0014] As a specific solution in this application, the acquisition of multi-frame temporal angiography images of coronary microvessels includes: acquiring images based on digital subtraction angiography technology, following these steps:
[0015] Configure the acquisition parameters to focus on the heart region with the preset highest spatial resolution and preset frame rate;
[0016] Acquire clean background images without artifacts and verify the quality of the background images through image playback;
[0017] Contrast agent is injected via coronary artery catheter. Simultaneously with the start of contrast agent injection, continuous image acquisition of the cardiac region is initiated to obtain multiple frames of sequential raw angiographic images.
[0018] Each original angiography image is preprocessed to obtain multi-frame temporal angiography images of the coronary microvessels.
[0019] As a specific solution in this application, the step of obtaining multi-frame grayscale difference images based on each frame of temporal angiography images includes:
[0020] Iterate through each frame of the time-series angiography images and acquire the first and second images one by one;
[0021] For each pair of obtained first and second images, the following steps are performed to obtain a corresponding grayscale difference image for any two temporally adjacent temporal contrast images in each frame of temporal contrast images;
[0022] The corner matching algorithm based on scale-invariant feature transform aligns the pixels in the first image and the second image to obtain multiple pixel pairs; each pixel pair includes a first pixel and a second pixel; the first pixel is a pixel in the first image; the second pixel is a pixel in the second image;
[0023] Based on each pixel pair, obtain the first pixel value that corresponds one-to-one with each pixel pair; the first pixel value is equal to the pixel value of the first pixel in the corresponding pixel pair minus the pixel value of the second pixel.
[0024] The grayscale difference image is obtained based on each first pixel value.
[0025] As a specific solution in this application, the step of obtaining multiple first regions based on the grayscale difference images of each frame includes:
[0026] Iterate through each frame of grayscale difference images and obtain the grayscale difference images one by one;
[0027] For each grayscale difference image, the following steps are performed to obtain multiple corresponding third regions for each grayscale difference image;
[0028] Based on the grayscale difference image, multiple first feature pixels are obtained;
[0029] Based on the clustering algorithm, each first feature pixel is clustered to obtain multiple first clusters;
[0030] Based on each first cluster, multiple third regions are obtained; each first cluster corresponds one-to-one with a third region.
[0031] After obtaining the corresponding third region for each grayscale difference image, multiple first regions are obtained based on the multiple third regions.
[0032] As a specific solution in this application, the step of obtaining the microvessel probability corresponding one-to-one with each first region includes:
[0033] Iterate through each first region and obtain the first region one by one;
[0034] For each first region, the following steps are performed to obtain the corresponding microvessel probability for each first region;
[0035] Based on the first region, the aspect ratio is obtained; the aspect ratio is equal to the ratio of the length to the width of the first region;
[0036] Based on the aspect ratio, obtain the first coefficient;
[0037] Based on the first coefficient, the probability of the microvessels is obtained.
[0038] As a specific solution in this application, obtaining the microvascular probability based on the first coefficient includes:
[0039] Based on the first region, a first average value and a first variance are obtained; the first average value is the average gray value of each pixel in the first region; the first variance is the variance of the gray value of each pixel in the first region.
[0040] Based on the grayscale difference image corresponding to the first region, a second average value is obtained; the second average value is the average grayscale value of each pixel in all regions except the first regions in the grayscale difference image.
[0041] A second coefficient is obtained based on the first average, the first variance, and the second average; the second coefficient is positively correlated with the first average and negatively correlated with both the first variance and the second average.
[0042] The microvascular probability is obtained based on the first coefficient and the second coefficient.
[0043] As a specific solution in this application, obtaining the microvascular probability based on the first coefficient and the second coefficient includes:
[0044] Based on the first region, multiple overlap rates are obtained; the overlap rate is equal to the ratio of the overlap area to its own area; the overlap area is the area of the overlap portion between the first region and any other first region; the own area is the area of the first region.
[0045] Based on the various overlap rates, obtain the third coefficient;
[0046] The microvascular probability is obtained based on the first coefficient, the second coefficient, and the third coefficient.
[0047] As a specific solution in this application, the step of superimposing the second regions in each frame of grayscale difference image to obtain an overlapping heatmap includes:
[0048] Based on each second region, multiple second feature pixels are obtained; the second feature pixels are any pixels in each second region that overlap with other second regions more than or equal to a third preset value.
[0049] The second feature pixels are clustered based on a clustering algorithm to obtain multiple second clusters.
[0050] Based on each second cluster, multiple fourth regions are obtained; each second cluster corresponds one-to-one with a fourth region.
[0051] Connect the outer contours of each fourth region to obtain multiple suspected coronary microvascular regions connected to the main coronary artery vessels;
[0052] Overlapping heatmaps were obtained based on suspected areas of coronary microvessels.
[0053] As a specific solution in this application, the step of segmenting and labeling coronary microvessels in each frame of time-series angiography images based on the overlapping heatmap includes:
[0054] Based on the overlapping heatmap, the main coronary arteries in each frame of time-series angiography images are partitioned, and the main coronary arteries are divided into multiple equal-length main artery partitions using a preset length as the partitioning standard.
[0055] The average inner diameter of the blood vessels in each main trunk region is calculated in multiple frames of time-series angiography images. The maximum and minimum values of the average inner diameter of the blood vessels in each main trunk region are recorded. The time-series angiography image interval corresponding to the change in inner diameter from the maximum value to the minimum value is recorded as the systolic phase of the main trunk region, and the time-series angiography image interval corresponding to the change in inner diameter from the minimum value to the maximum value is recorded as the diastolic phase of the main trunk region.
[0056] The grayscale values of the suspected coronary microvessel region corresponding to each main trunk partition are recorded synchronously during the systolic and diastolic phases to construct a corresponding sequence for each main trunk partition. The corresponding sequence includes the inner diameter of the main trunk partition and the average grayscale value of the suspected coronary microvessel region connected to the main trunk partition.
[0057] Calculate the Pearson correlation coefficient for the corresponding sequence during systole and diastole for each suspected coronary microvascular region.
[0058] Based on the Pearson correlation coefficient, the comprehensive synergy rate of each suspected coronary microvascular region is obtained;
[0059] The suspected coronary microvascular regions with a comprehensive synergy rate greater than or equal to the fourth preset value are marked as the actual coronary microvascular regions, thus completing the segmentation and labeling of coronary microvessels in each frame of time-series angiography images.
[0060] As a specific solution in this application, the step of obtaining the comprehensive synergy rate of each suspected coronary microvascular region based on the Pearson correlation coefficient includes:
[0061] The range of values for the systolic and diastolic Pearson correlation coefficients corresponding to each suspected coronary microvascular region was determined:
[0062] If either the systolic or diastolic Pearson correlation coefficient falls within the range of [-1, 0], the periodic synergy between the suspected coronary microvascular region and the main coronary artery region is deemed unqualified, and its overall synergy rate is directly set to 0; otherwise, the systolic and diastolic durations are calculated. The systolic duration is the product of the number of frames in the corresponding time-series angiography image interval during systole and the single-frame acquisition interval; the diastolic duration is the product of the number of frames in the corresponding time-series angiography image interval during diastole and the single-frame acquisition interval.
[0063] Based on the duration of systole and the duration of diastole, a first weighting coefficient and a second weighting coefficient are obtained.
[0064] Based on the first weighting coefficient and the second weighting coefficient, the comprehensive synergistic rate of the suspected coronary microvascular region is obtained.
[0065] Compared with the prior art, the beneficial effects of this application are:
[0066] The embodiment of the intelligent segmentation method for coronary microvascular angiography images proposed in this application captures the dynamic grayscale changes of microvessels during contrast agent perfusion by acquiring multiple frames of time-series angiography images and calculating grayscale difference images. This overcomes the limitations of traditional methods that rely on static single-frame images and ignore temporal information, effectively amplifying the weak signals of microvessels with small diameters and low contrast agent concentrations. Furthermore, by extracting a first region using first feature pixels to initially focus on the dynamic region, and then combining this with microvessel probability quantification assessment to screen the second region, background noise, tissue artifacts, and other interferences are eliminated layer by layer. This solves the technical problem of insufficient sensitivity to low-contrast microvessels and difficulty in distinguishing real blood vessels from noise in traditional methods. Finally, by superimposing multiple frames of the second region to generate an overlapping heatmap, the continuous presence of real microvessels in multiple frames is highlighted, avoiding misjudgment of isolated regions. Ultimately, complete and accurate segmentation of coronary microvessels is achieved, providing a reliable foundation for subsequent clinical quantitative analysis such as fractional flow reserve. Attached Figure Description
[0067] Figure 1 This is a flowchart illustrating an intelligent segmentation method for coronary microvascular angiography images proposed in an embodiment of this application.
[0068] Figure 2 This is a schematic diagram illustrating the acquisition of grayscale difference images as proposed in the embodiments of this application. Detailed Implementation
[0069] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0070] The terms "first," "second," etc., used in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. For example, the first region and the second region mentioned below belong to different regions. It should be understood that such names can be used interchangeably where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or modules is not necessarily limited to those steps or modules explicitly listed, but may include other steps or modules not explicitly listed or inherent to these processes, methods, products, or devices. The division of modules in the embodiments of this application is merely a logical division. In actual applications, there may be other division methods. For example, multiple modules may be combined into or integrated into another system, or some features may be ignored or not performed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interface, and the indirect coupling or communication connection between modules may be electrical or other similar forms. None of these are limited in the embodiments of this application. Furthermore, the modules or sub-modules described as separate components may or may not be physically separated, may or may not be physical modules, or may be distributed among multiple circuit modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiments of this application.
[0071] To address the technical problem mentioned in the background art, where coronary microvessels are poorly represented in coronary angiography images, making it difficult to obtain complete images, this application proposes an intelligent segmentation method for coronary microvessel angiography images. Specifically, as... Figure 1 As shown, the intelligent segmentation method for coronary microvascular angiography images includes steps S100 to S700.
[0072] Step S100: Acquire multi-frame temporal angiography images of coronary microvessels.
[0073] In this embodiment, the multi-frame temporal angiography images are acquired during coronary microvascular angiography. Specifically, step S100, acquiring multi-frame temporal angiography images of coronary microvessels, includes: acquiring images based on digital subtraction angiography (DSA) according to the following steps:
[0074] Step S110: Configure the acquisition parameters to focus on the heart region with the preset highest spatial resolution and preset frame rate.
[0075] In this embodiment, the preset frame rate can be set according to requirements. For example, the preset frame rate can be 30 frames / second or 40 frames / second, etc.
[0076] Step S120: Acquire a clean background image without artifacts, and verify the quality of the background image through image playback.
[0077] In this embodiment, verifying the quality of the background image through image playback means confirming that the image is free of breathing artifacts.
[0078] Step S130: Inject contrast agent through coronary artery catheter. Simultaneously with the start of contrast agent injection, start continuous image acquisition of the cardiac region to obtain multiple frames of sequential angiographic raw images.
[0079] In this embodiment, a commonly used iodine-containing nonionic contrast agent (e.g., iohexol or iodixanol) is selected as the contrast agent. Its concentration needs to be adjusted according to the patient's weight, renal function, and vascular visualization requirements (typically 300 mgI / mL to 370 mgI / mL) to balance imaging clarity and patient tolerability. The injection is performed using a high-pressure injector at an injection rate of 3 mL / s to 5 mL / s, with a total injection dose controlled between 10 mL and 15 mL. This ensures rapid and uniform filling of the main coronary arteries and coronary microvessels at all levels, while avoiding the risk of contrast-induced nephropathy due to excessive dosage.
[0080] Step S140: Preprocess each original angiography image to obtain multi-frame time-series angiography images of the coronary microvessels.
[0081] In this embodiment, the purpose of preprocessing the original angiography images is to eliminate irrelevant interference, enhance microvascular signal characteristics, and unify image quality, laying the foundation for subsequent grayscale difference calculation and region extraction. In the embodiments of this application, any reasonable method can be used to preprocess the original angiography images. For example, preprocessing methods can include filtering and noise reduction, and signal enhancement. Filtering and noise reduction and signal enhancement are mature technologies and will not be elaborated upon here.
[0082] This concludes the introduction to the method for acquiring multiple temporal angiography images.
[0083] Step S200: Based on the temporal angiography images of each frame, obtain multi-frame grayscale difference images.
[0084] In this embodiment, the pixel value in the grayscale difference image is equal to the difference in grayscale values of corresponding pixels in the first image and the second image. The first image is any image in each frame of temporal contrast imaging. The second image is an image in each frame of temporal contrast imaging that is temporally adjacent to the first image, and the temporal sequence of the second image is after that of the first image. Specifically, step S200, based on each frame of temporal contrast imaging, obtains multiple frames of grayscale difference images, including steps S210 to S240.
[0085] Step S210: Traverse each frame of time-series angiography images and acquire the first image and the second image one by one.
[0086] In this embodiment, the following steps are performed for each pair of obtained first and second images to obtain a corresponding grayscale difference image for any two temporally adjacent temporal contrast images in each frame of temporal contrast images (the grayscale difference images obtained based on the first and second images are as follows). Figure 2 (As shown).
[0087] Step S220: Align each pixel in the first image and the second image based on the scale-invariant feature transform corner matching algorithm to obtain multiple pixel pairs.
[0088] It's important to understand that the Scale-Invariant Feature Transform (SIFT) corner matching algorithm is a classic image local feature extraction and matching algorithm. It can extract stable feature points from an image that are scale-invariant and rotation-invariant. Even if the image undergoes scaling, rotation, lighting changes, or slight viewpoint shifts, reliable matching of corresponding features between different images can still be achieved through the description and comparison of these feature points. The SIFT corner matching algorithm is a mature technology and will not be elaborated upon further.
[0089] In this embodiment, each pixel pair includes a first pixel and a second pixel, where the first pixel is a pixel in the first image and the second pixel is a pixel in the second image.
[0090] Step S230: Based on each pixel pair, obtain the first pixel value that corresponds one-to-one with each pixel pair.
[0091] In this embodiment, the first pixel value is equal to the pixel value of the first pixel in the corresponding pixel pair minus the pixel value of the second pixel.
[0092] Step S240: Obtain the grayscale difference image based on each first pixel value.
[0093] In this embodiment, the pixel value of each pixel in the grayscale difference image is equal to the first pixel value.
[0094] Step S300: Based on the grayscale difference images of each frame, obtain multiple first regions.
[0095] It is important to note that coronary angiography is a crucial technique for diagnosing coronary heart disease, accurately detecting the location and degree of blockage in coronary arteries. However, in monitoring coronary microvessels, because the diameter of coronary microvessels is much smaller than that of the main coronary arteries, less contrast agent flows through them, resulting in lighter contrast and making them more susceptible to interference from other structures in the background myocardial tissue. Furthermore, the continuous contraction and relaxation of the myocardium further complicates the segmentation of coronary microvessels, making identification even more difficult. Based on this, the inventors of this application discovered that in sequential coronary microvessel angiography images captured by DSA, the grayscale values in both coronary microvessels and the main coronary arteries continuously increase due to the increasing concentration of contrast agent in the body. Based on these characteristics, the inventors of this application found that preliminary judgment can be made based on the grayscale value changes of adjacent DSA frames to screen out candidate regions (i.e., the first region) for suspected coronary microvessels. In other words, in this embodiment, the first region is the area formed by the first feature pixel in the grayscale difference image. The first feature pixel is any pixel in the grayscale difference image whose absolute value of the difference between its pixel value and the average pixel value is greater than a first preset value. The average pixel value is the average value of the corresponding pixel values of all pixels in the grayscale difference image. In other words, in this embodiment, the core purpose of obtaining the first region is to initially screen out candidate regions suspected of being coronary microvessels from the grayscale difference image. The grayscale difference image reflects the dynamic changes of pixels in the time-series angiography image. The first region is formed by clustering pixels with significant grayscale differences, which can focus on dynamic regions related to contrast agent flow. It provides a basis for subsequent calculation of microvessel probability and screening of the second region, and can eliminate a large amount of background noise and static interference, narrowing the scope of subsequent accurate identification of coronary microvessels.
[0096] Specifically, in this embodiment, step S300 involves obtaining multiple first regions based on the grayscale difference images of each frame, including steps S310 to S340.
[0097] Step S310: Traverse each frame of grayscale difference image and obtain the grayscale difference image one by one.
[0098] In this embodiment, the following steps are performed for each grayscale difference image to obtain a plurality of corresponding third regions for each grayscale difference image.
[0099] Step S320: Based on the grayscale difference image, obtain multiple first feature pixels.
[0100] As mentioned above, the first feature pixel is any pixel in the grayscale difference image whose absolute value of the difference between the pixel value and the average pixel value is greater than a first preset value.
[0101] In this embodiment, the first preset value can be set according to requirements, for example, the first preset value can be 15 or 25, etc.
[0102] Step S330: Cluster each first feature pixel point based on the clustering algorithm to obtain multiple first clusters.
[0103] In this embodiment, any reasonable clustering algorithm can be used to cluster the first feature pixels. For example, the clustering algorithm can be the DBSCAN algorithm or the OPTICS algorithm. It should be noted that clustering multiple pixels using clustering algorithms is a mature technology, and will not be elaborated here.
[0104] Step S340: Based on each first cluster, obtain multiple third regions.
[0105] In this embodiment, each first cluster corresponds one-to-one with a third region. That is, each first feature pixel in each first cluster can form a third region.
[0106] After obtaining the corresponding third region for each grayscale difference image, multiple first regions are obtained based on the multiple third regions.
[0107] In this embodiment, the third region and the first region can also correspond one-to-one. That is, one third region is one first region. Of course, in other embodiments, if two third regions overlap, they can be merged to form one first region.
[0108] Step S400: Based on each first region, obtain the microvessel probability corresponding to each first region.
[0109] It is important to note that because the color representation of coronary microvessels in the image after angiography is relatively weak, it is easily confused with random grayscale variations in the background noise. Therefore, to eliminate background interference, it is necessary to obtain the microvessel probability of each first region in the grayscale difference image. In this embodiment, the microvessel probability is used at least to characterize the probability that the corresponding first region in the grayscale difference image is a region formed by microvessels.
[0110] In this embodiment, the probability of microvessels corresponding to the first region can be obtained in any reasonable way. For example, since coronary microvessels appear as slender linear structures in images, the larger the aspect ratio of a certain first region, the closer the shape of the first region is to linearity, which better matches the image representation of coronary microvessels. Based on this, the probability of microvessels in each first region can be obtained according to the shape characteristics of the first region. Specifically, step S400, based on each first region, obtains the probability of microvessels corresponding to each first region, including steps S401 to S404.
[0111] Step S401: Traverse each first region and obtain the first region one by one.
[0112] In this embodiment, the following steps are performed for each first region to obtain the corresponding microvascular probability for each first region.
[0113] Step S402: Based on the first region, obtain the aspect ratio.
[0114] In this embodiment, the aspect ratio is equal to the ratio of the length to the width of the first region. It should be noted that obtaining the length-to-width ratio of a region is a well-established technique in computer image processing, and will not be elaborated upon here.
[0115] Step S403: Obtain the first coefficient based on the aspect ratio.
[0116] In this embodiment, the first coefficient can be equal to the aspect ratio, or it can be equal to the normalized value of the aspect ratio. In the field of artificial intelligence, normalizing a value (i.e., the aspect ratio) is a mature technology, which will not be elaborated here.
[0117] Specifically, the formula for calculating the first coefficient can be as follows:
[0118] ;
[0119] in, Indicates the first coefficient; Indicates the length of the first region; Indicates the width of the first region; This indicates that the normalization function is used to normalize the values within the parentheses to the range [0, 1].
[0120] Step S404: Based on the first coefficient, obtain the probability of the microvessels.
[0121] In this embodiment, the normalized aspect ratio can be used as the probability of the microvessels.
[0122] It is important to note that while calculating the microvessel probability using the first coefficient corresponding to the aspect ratio can screen out regions that match the microvessel morphology based on linear shape features, it is difficult to completely eliminate background interference with similar shapes (e.g., elongated tissue artifacts, linear noise clusters formed by chaotic gray-level fluctuations, etc.). Combined with the angiographic characteristics of coronary microvessels, it is known that the diffusion of contrast agent within the microvessels results in relatively uniform gray-level changes within the region, and the gray-level value increases significantly with increasing contrast agent concentration, showing a significant difference in gray-level from the surrounding non-vascular areas. To more accurately assess the microvessel probability of the first region, it is necessary to supplement and integrate the gray-level statistical features of the first region and its gray-level contrast relationship with the background. Therefore, a second coefficient needs to be introduced to further optimize the calculation of the microvessel probability. Specifically, step S404 obtains the microvessel probability based on the first coefficient, including steps S405 to S408.
[0123] Step S405: Based on the first region, obtain the first average value and the first variance.
[0124] In this embodiment, the first average value is the average grayscale value of each pixel in the first region. The first variance is the variance of the grayscale values of each pixel in the first region. It should be noted that obtaining the average and variance of numerical values is a mature technology in the computer field, and will not be elaborated upon here.
[0125] Step S406: Obtain a second average value based on the grayscale difference image corresponding to the first region.
[0126] In this embodiment, the second average value is the average gray value of each pixel in all regions except the first regions in the gray-scale difference image.
[0127] Step S407: Obtain the second coefficient based on the first average value, the first variance, and the second average value.
[0128] In this embodiment, the second coefficient is positively correlated with the first average value and negatively correlated with both the first variance and the second average value. The second coefficient can be obtained based on the first average value, the first variance, and the second average value in any reasonable manner. For example, step S407: The calculation formula for obtaining the second coefficient based on the first average value, the first variance, and the second average value can be as follows:
[0129] ;
[0130] in, Indicates the second coefficient; This represents the first average value; This represents the second average value; Indicates the first variance; The normalization function is used to normalize the values within the parentheses to the range [0, 1]. Alternatively, step S407: Based on the first average, the first variance, and the second average, the formula for calculating the second coefficient can be as follows:
[0131] ;
[0132] in, Indicates the second coefficient; This represents the first average value; This represents the second average value; Indicates the first variance; This represents an exponential function with the natural constant e as its base.
[0133] Step S408: Obtain the microvascular probability based on the first coefficient and the second coefficient.
[0134] In this embodiment, the microvessel probability can be obtained based on the first coefficient and the second coefficient in any reasonable manner. For example, the microvessel probability can be the average of the first coefficient and the second coefficient. It should be noted that in this embodiment, although calculating the microvessel probability using the first coefficient and the second coefficient can screen candidate regions (i.e., the second region) from the perspective of spatial morphology and grayscale characteristics, it does not consider the dynamic stability characteristics of coronary microvessels in the temporal dimension. Due to the continuous perfusion of contrast agent, the corresponding regions of real coronary microvessels will continuously appear in multiple frames of grayscale difference images. That is, if the formation of a certain first region is caused by coronary microvessels, then the first region will have a high overlap rate with other first regions. Regions formed by random background noise or transient artifacts are mostly isolated in a single frame, with extremely low overlap rates. In order to further optimize the evaluation of microvessel probability by combining temporal consistency characteristics, in one embodiment of this application, the microvessel probability is obtained based on the first coefficient and the second coefficient in step S408, including steps S409 to S411.
[0135] Step S409: Based on the first region, obtain multiple overlap rates.
[0136] In this embodiment, the overlap rate is equal to the ratio of the overlapping area to the area of the first region itself. The overlapping area is the area of the portion overlapping between the first region and any other first region. The area of the first region itself is the area of the first region. Specifically, the formula for calculating the overlap rate is as follows:
[0137] ;
[0138] in, Indicates the overlap rate; This represents the area of the overlapping portion formed by the first region (hereinafter referred to as the fifth region) and the i-th first region among the other first regions; This indicates the area of the fifth region itself.
[0139] Step S410: Obtain the third coefficient based on each overlap rate.
[0140] In this embodiment, the third coefficient can be obtained based on each overlap rate using any reasonable method. For example, step S410: The calculation formula for obtaining the third coefficient based on each overlap rate is as follows:
[0141] ;
[0142] in, Indicates the third coefficient; This indicates the number of non-zero overlap regions corresponding to the first region; This represents the j-th non-zero overlap rate corresponding to the first region. Alternatively, step S410: Based on each overlap rate, the calculation formula for the third coefficient is as follows:
[0143] ;
[0144] in, Indicates the third coefficient; This indicates the number of non-zero overlap regions corresponding to the first region; This represents the j-th non-zero overlap rate corresponding to the first region; This represents the variance of each non-zero overlap rate corresponding to the first region; Zero-prevention coefficient, used to prevent the denominator from being 0. It can be any positive number close to 0, for example, the zero-prevention coefficient. It can be 0.01 or 0.001; This indicates that the normalization function is used to normalize the values within the parentheses to the range [0, 1].
[0145] Step S411: Obtain the microvascular probability based on the first coefficient, the second coefficient, and the third coefficient.
[0146] In this embodiment, the microvascular probability can be obtained based on the first coefficient, the second coefficient, and the third coefficient using any reasonable method. For example, the microvascular probability can be the average of the first coefficient, the second coefficient, and the third coefficient. Alternatively, step S411: The calculation formula for the microvascular probability based on the first coefficient, the second coefficient, and the third coefficient is as follows:
[0147] ;
[0148] in, Indicates the probability of microvessels; Indicates the first weight; Indicates the first coefficient; Indicates the second weight; Indicates the second coefficient; Indicates the third weight; This represents the third coefficient; the first, second, and third weights are set according to requirements, and the sum of the first, second, and third weights is 1.
[0149] Step S500: Based on the probability of each microvessel, obtain multiple second regions.
[0150] In this embodiment, the second region is any region in each of the first regions where the probability of microvessels is greater than or equal to a second preset value.
[0151] In this embodiment, a second preset value can be set according to requirements. For example, the second preset value can be 0.6 or 0.7, etc.
[0152] Step S600: Overlay the second regions in each frame of grayscale difference image to obtain an overlapping heatmap.
[0153] In this embodiment, any reasonable method can be used to superimpose the second regions in each frame of grayscale difference images to obtain an overlapping heatmap. For example, the overlapping heatmap can be obtained by directly aligning and overlapping the grayscale difference images. It should be noted that if the overlapping heatmap is obtained by directly aligning and overlapping the grayscale difference images, there is a lack of clustering and integration, and it is impossible to aggregate the discretely distributed effective pixels into a continuous structure. Since microvessels have branching and continuous anatomical features, direct superposition will make them present a fragmented form, completely losing the true vascular course and hierarchical relationship. Based on this, step S600, superimposing the second regions in each frame of grayscale difference images to obtain an overlapping heatmap, includes steps S610 to S650.
[0154] Step S610: Based on each second region, obtain multiple second feature pixels.
[0155] In this embodiment, the second feature pixel is any pixel in each second region that overlaps with other second regions a number greater than or equal to a third preset value.
[0156] In this embodiment, the third preset value can be set according to requirements, for example, the third preset value can be 3, 4 or 5, etc.
[0157] Step S620: Cluster each second feature pixel point based on the clustering algorithm to obtain multiple second clusters.
[0158] It is important to understand that clustering multiple pixels (i.e., each second feature pixel) using clustering algorithms is a mature technique, which will not be elaborated on here.
[0159] Step S630: Based on each second cluster, obtain multiple fourth regions.
[0160] In this embodiment, each second cluster corresponds one-to-one with a fourth region. That is, in this embodiment, one second cluster corresponds to one fourth region.
[0161] Step S640: Connect the outer contours of each fourth region to obtain multiple suspected coronary microvascular regions connected to the main coronary artery vessels.
[0162] In this embodiment, the outer contours of each fourth region can be connected in any reasonable manner to obtain multiple suspected coronary microvascular regions connected to the main coronary arteries. For example, a region growing algorithm can be used to connect the outer contours of each fourth region to obtain multiple suspected coronary microvascular regions connected to the main coronary arteries. Region growing algorithms are mature technologies and will not be elaborated upon here.
[0163] Step S650: Obtain an overlapping heat map based on the suspected areas of each coronary microvessel.
[0164] The overlapping heatmap obtained in this embodiment filters out low-quality interference such as single-frame artifacts and isolated noise by selecting second feature pixels that meet the overlap count standard, thereby improving the signal-to-noise ratio. Then, clustering processing aggregates discrete pixels into a fourth region, restoring the coherent anatomical features of microvascular branches and avoiding the fragmentation issues caused by direct superposition. Finally, by connecting the outer contour and associating it with the main coronary arteries, the microvascular region is accurately located, effectively distinguishing real blood vessels from noise clusters. Overall, it achieves interference removal, structural integration, and authenticity identification, outputting an overlapping heatmap with clear features and a complete structure.
[0165] Step S700: Based on the overlapping heatmap, segment and label the coronary microvessels in each frame of time-series angiography images.
[0166] In this embodiment, the suspected coronary microvessel regions in the overlapping heatmap can be directly used as the coronary microvessels in each frame of time-series angiography images.
[0167] It is important to understand that although the overlapping heatmap obtained in step S600 has identified suspected areas of coronary microvessels connected to the main coronary arteries, false positive areas may still exist due to uneven distribution of contrast agent and residual tissue artifacts. It is also important to note that coronary microvessels undergo dynamic changes with myocardial contraction and relaxation, and these changes are synergistic with the cyclical rhythm of the main coronary arteries. That is, during myocardial contraction, the inner diameter of the main coronary arteries narrows, resulting in reduced blood flow in the corresponding coronary microvessels (i.e., a decrease in grayscale value in the sequential angiography image); conversely, during myocardial relaxation, the inner diameter of the main coronary arteries widens, resulting in increased blood flow in the corresponding coronary microvessels (i.e., an increase in grayscale value in the sequential angiography image). Based on this, step S700, using the overlapping heatmap, segments and labels the coronary microvessels in each frame of the sequential angiography image, including steps S710 to S760.
[0168] Step S710: Based on the overlapping heatmap, the coronary trunk vessels in each frame of time-series angiography image are partitioned, and the coronary trunk vessels are divided into multiple equal-length trunk partitions using a preset length as the partitioning standard.
[0169] As mentioned earlier, since the overlay heatmap contains the main coronary arteries, comparing each frame of time-series angiography images with the overlay heatmap allows us to identify the main coronary arteries in each frame. In other words, the suspected area in each frame of time-series angiography images that roughly overlaps with the main coronary arteries in the overlay heatmap is the main coronary artery in that time-series angiography image.
[0170] In this embodiment, the main coronary artery can be divided into multiple main artery sections using any reasonable preset length. For example, the preset length can be 128 pixels, 256 pixels, or 512 pixels, etc.
[0171] Step S720: Calculate the average inner diameter of blood vessels in each main trunk region in multi-frame time-series angiography images, record the maximum and minimum values of the average inner diameter of blood vessels in each main trunk region, record the time-series angiography image interval corresponding to the change in inner diameter from the maximum to the minimum value as the systolic phase of the main trunk region, and record the time-series angiography image interval corresponding to the change in inner diameter from the minimum to the maximum value as the diastolic phase of the main trunk region.
[0172] It is important to understand that in the field of image processing, obtaining the average diameter of a certain region (i.e., the backbone partition) is a mature technology, which will not be elaborated here.
[0173] Step S730: Simultaneously record the gray values of the suspected coronary microvascular area corresponding to each main trunk partition during systole and diastole, and construct the corresponding sequence for each main trunk partition.
[0174] In this embodiment, the corresponding sequence includes the inner diameter of the main trunk partition and the average gray value of the suspected coronary microvascular region connected to the main trunk partition.
[0175] Step S740: Calculate the Pearson correlation coefficient for the corresponding sequence during systole and diastole for each suspected coronary microvascular region.
[0176] It is important to note that the Pearson correlation coefficient, also known as the Pearson product-moment correlation coefficient (PCC), is a statistic used to measure the strength and direction of the linear correlation between two continuous random variables. The core definition of the Pearson correlation coefficient can be summarized as follows: by calculating the ratio of the product of the covariance and the standard deviation of each variable, the correlation is standardized to the interval [-1, 1], thus intuitively reflecting the strength and direction of the linear association between the variables. In other words, obtaining the Pearson correlation coefficients for the inner diameter of the main coronary artery region and the average gray value of the suspected coronary microvascular region in the corresponding systolic sequence, as well as obtaining the Pearson correlation coefficients for the inner diameter of the main coronary artery region and the average gray value of the suspected coronary microvascular region in the corresponding diastolic sequence, are both mature techniques and will not be elaborated upon here.
[0177] Step S750: Based on the Pearson correlation coefficient, obtain the comprehensive synergy rate of each suspected coronary microvascular region.
[0178] In this embodiment, the larger the two Pearson correlation coefficients, the greater the probability that the corresponding suspected coronary microvascular region is caused by coronary microvascular disease. Specifically, step S750 involves obtaining the comprehensive synergy rate of each suspected coronary microvascular region based on the Pearson correlation coefficients, including steps S751 to S754.
[0179] Step S751: Determine the range of values for the systolic and diastolic Pearson correlation coefficients corresponding to each suspected coronary microvascular region.
[0180] Step S752: If either the systolic Pearson correlation coefficient or the diastolic Pearson correlation coefficient is in the range of [-1, 0], then the periodic synergy between the suspected coronary microvascular region and the main trunk region is deemed unqualified, and its overall synergy rate is directly set to 0; otherwise, the systolic duration and diastolic duration are calculated.
[0181] In this embodiment, the duration of the systolic phase is the product of the number of frames in the time-series angiography image interval corresponding to the systolic phase and the single-frame acquisition time interval. The duration of the diastolic phase is the product of the number of frames in the time-series angiography image interval corresponding to the diastolic phase and the single-frame acquisition interval.
[0182] Step S753: Based on the duration of the systolic phase and the duration of the diastolic phase, obtain the first weighting coefficient and the second weighting coefficient.
[0183] Specifically, the formula for calculating the first weighting coefficient is as follows:
[0184] ;
[0185] in, Indicates the first weighting coefficient; Indicates the duration of the systolic phase; Indicates the duration of diastole.
[0186] The formula for calculating the second weighting coefficient is as follows:
[0187] ;
[0188] in, This represents the second weighting coefficient; Indicates the duration of the systolic phase; Indicates the duration of diastole.
[0189] Step S754: Based on the first weighting coefficient and the second weighting coefficient, obtain the comprehensive synergistic rate of the suspected coronary microvascular region.
[0190] In this embodiment, based on the first weighting coefficient and the second weighting coefficient, the formula for calculating the comprehensive synergistic rate of the suspected coronary microvascular region is as follows:
[0191] ;
[0192] in, Indicates the overall coordination rate; Indicates the first weighting coefficient; This represents the Pearson correlation coefficient corresponding to the systolic phase; This represents the second weighting coefficient; This represents the Pearson correlation coefficient corresponding to the diastolic phase.
[0193] Step S760: Mark the suspected coronary microvascular regions with a comprehensive synergy rate greater than or equal to the fourth preset value as the real coronary microvascular regions, and complete the segmentation and labeling of coronary microvessels in each frame of time-series angiography images.
[0194] In this embodiment, a fourth preset value can be set according to requirements. For example, the fourth preset value can be 0.5 or 0.6, etc.
[0195] The embodiment of the intelligent segmentation method for coronary microvascular angiography images proposed in this application captures the dynamic grayscale changes of microvessels during contrast agent perfusion by acquiring multiple frames of time-series angiography images and calculating grayscale difference images. This overcomes the limitations of traditional methods that rely on static single-frame images and ignore temporal information, effectively amplifying the weak signals of microvessels with small diameters and low contrast agent concentrations. This embodiment also uses a first feature pixel to extract a first region to initially focus on the dynamic region, and then combines this with a microvessel probability quantification assessment to screen a second region, eliminating background noise, tissue artifacts, and other interferences layer by layer. This solves the technical problems of insufficient sensitivity to low-contrast microvessels and difficulty in distinguishing real blood vessels from noise in traditional methods. Furthermore, this embodiment generates an overlapping heatmap by superimposing multiple frames of the second region, highlighting the continuous presence of real microvessels in multiple frames of images, avoiding misjudgment of isolated regions, and ultimately achieving complete and accurate segmentation of coronary microvessels, providing a reliable foundation for subsequent clinical quantitative analysis such as fractional flow reserve.
[0196] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.
[0197] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the methods, apparatuses, and devices described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0198] In the several embodiments provided in this application, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or modules may be electrical, mechanical, or other forms.
[0199] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.
[0200] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.
[0201] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.
[0202] The computer program product includes one or more computer instructions. When the computer program is loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital video optical disc), or a semiconductor medium (e.g., solid-state disk (SSD)).
[0203] Although embodiments of this application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles of this application.
Claims
1. An intelligent segmentation method for coronary microvascular angiography images, characterized in that, include: Acquire multi-frame temporal angiography images of coronary microvessels; The multi-frame temporal angiography images were acquired during coronary microvascular angiography. Based on each frame of temporal contrast imaging images, multiple frames of grayscale difference images are obtained; the pixel values in the grayscale difference images are equal to the difference in grayscale values of corresponding pixels in the first image and the second image; the first image is any image in each frame of temporal contrast imaging images; the second image is an image in each frame of temporal contrast imaging images that is temporally adjacent to the first image, and the temporal sequence of the second image is after the temporal sequence of the first image; Based on the grayscale difference images of each frame, multiple first regions are obtained; the first region is the region formed by the first feature pixels in the grayscale difference image; the first feature pixel is any pixel in the grayscale difference image whose absolute value of the difference between the pixel value and the average pixel value is greater than a first preset value; the average pixel value is the average value of the corresponding pixel values of all pixels in the grayscale difference image. Based on each first region, the microvessel probability corresponding to each first region is obtained; the microvessel probability is at least used to characterize the probability that the corresponding first region in the gray-scale difference image is a region formed by microvessels. Based on the probability of each microvessel, multiple second regions are obtained; The second region is any region in each of the first regions where the probability of microvessels is greater than or equal to a second preset value; The second regions in each frame of grayscale difference image are superimposed to obtain an overlapping heatmap; Based on the overlapping heatmap, the main coronary arteries in each frame of time-series angiography images are partitioned, and the main coronary arteries are divided into multiple equal-length main artery partitions using a preset length as the partitioning standard. The average inner diameter of the blood vessels in each main trunk region is calculated in multiple frames of time-series angiography images. The maximum and minimum values of the average inner diameter of the blood vessels in each main trunk region are recorded. The time-series angiography image interval corresponding to the change in inner diameter from the maximum value to the minimum value is recorded as the systolic phase of the main trunk region, and the time-series angiography image interval corresponding to the change in inner diameter from the minimum value to the maximum value is recorded as the diastolic phase of the main trunk region. The grayscale values of the suspected coronary microvessel region corresponding to each main trunk partition are recorded synchronously during the systolic and diastolic phases to construct a corresponding sequence for each main trunk partition. The corresponding sequence includes the inner diameter of the main trunk partition and the average grayscale value of the suspected coronary microvessel region connected to the main trunk partition. Calculate the Pearson correlation coefficient for the corresponding sequence during systole and diastole for each suspected coronary microvascular region. Based on the Pearson correlation coefficient, the comprehensive synergy rate of each suspected coronary microvascular region is obtained; The suspected coronary microvascular regions with a comprehensive synergy rate greater than or equal to the fourth preset value are marked as the actual coronary microvascular regions, thus completing the segmentation and labeling of coronary microvessels in each frame of time-series angiography images.
2. The intelligent segmentation method for coronary microvascular angiography images according to claim 1, characterized in that, The acquisition of multi-frame temporal angiography images of coronary microvessels includes: acquisition based on digital subtraction angiography technology, following the steps below: Configure the acquisition parameters to focus on the heart region with the preset highest spatial resolution and preset frame rate; Acquire clean background images without artifacts and verify the quality of the background images through image playback; Contrast agent is injected via coronary artery catheter. Simultaneously with the start of contrast agent injection, continuous image acquisition of the cardiac region is initiated to obtain multiple frames of sequential raw angiographic images. Each original angiography image is preprocessed to obtain multi-frame temporal angiography images of the coronary microvessels.
3. The intelligent segmentation method for coronary microvascular angiography images according to claim 1, characterized in that, The step of obtaining multi-frame grayscale difference images based on each frame of temporal angiography images includes: Iterate through each frame of the time-series angiography images and acquire the first and second images one by one; For each pair of obtained first and second images, the following steps are performed to obtain a corresponding grayscale difference image for any two temporally adjacent temporal contrast images in each frame of temporal contrast images; The corner matching algorithm based on scale-invariant feature transform aligns the pixels in the first image and the second image to obtain multiple pixel pairs; each pixel pair includes a first pixel and a second pixel; the first pixel is a pixel in the first image; the second pixel is a pixel in the second image; Based on each pixel pair, obtain the first pixel value that corresponds one-to-one with each pixel pair; the first pixel value is equal to the pixel value of the first pixel in the corresponding pixel pair minus the pixel value of the second pixel. The grayscale difference image is obtained based on each first pixel value.
4. The intelligent segmentation method for coronary microvascular angiography images according to claim 1, characterized in that, The process of obtaining multiple first regions based on the grayscale difference images of each frame includes: Iterate through each frame of grayscale difference images and obtain the grayscale difference images one by one; For each grayscale difference image, the following steps are performed to obtain multiple corresponding third regions for each grayscale difference image; Based on the grayscale difference image, multiple first feature pixels are obtained; Based on the clustering algorithm, each first feature pixel is clustered to obtain multiple first clusters; Based on each first cluster, multiple third regions are obtained; each first cluster corresponds one-to-one with a third region. After obtaining the corresponding third region for each grayscale difference image, multiple first regions are obtained based on the multiple third regions.
5. The intelligent segmentation method for coronary microvascular angiography images according to claim 1, characterized in that, The step of obtaining the microvessel probability corresponding one-to-one with each first region includes: Iterate through each first region and obtain the first region one by one; For each first region, the following steps are performed to obtain the corresponding microvessel probability for each first region; Based on the first region, the aspect ratio is obtained; the aspect ratio is equal to the ratio of the length to the width of the first region; Based on the aspect ratio, obtain the first coefficient; Based on the first coefficient, the probability of the microvessels is obtained.
6. The intelligent segmentation method for coronary microvascular angiography images according to claim 5, characterized in that, The step of obtaining the microvascular probability based on the first coefficient includes: Based on the first region, a first average value and a first variance are obtained; the first average value is the average gray value of each pixel in the first region; the first variance is the variance of the gray value of each pixel in the first region. Based on the grayscale difference image corresponding to the first region, a second average value is obtained; the second average value is the average grayscale value of each pixel in all regions except the first regions in the grayscale difference image. A second coefficient is obtained based on the first average, the first variance, and the second average; the second coefficient is positively correlated with the first average and negatively correlated with both the first variance and the second average. The microvascular probability is obtained based on the first coefficient and the second coefficient.
7. The intelligent segmentation method for coronary microvascular angiography images according to claim 6, characterized in that, The step of obtaining the microvascular probability based on the first coefficient and the second coefficient includes: Based on the first region, multiple overlap rates are obtained; the overlap rate is equal to the ratio of the overlap area to its own area; the overlap area is the area of the overlap portion between the first region and any other first region; the own area is the area of the first region. Based on the various overlap rates, obtain the third coefficient; The microvascular probability is obtained based on the first coefficient, the second coefficient, and the third coefficient.
8. The intelligent segmentation method for coronary microvascular angiography images according to claim 7, characterized in that, The step of overlaying the second regions in each frame of grayscale difference image to obtain an overlapping heatmap includes: Based on each second region, multiple second feature pixels are obtained; the second feature pixels are any pixels in each second region that overlap with other second regions more than or equal to a third preset value. The second feature pixels are clustered based on a clustering algorithm to obtain multiple second clusters. Based on each second cluster, multiple fourth regions are obtained; each second cluster corresponds one-to-one with a fourth region. Connect the outer contours of each fourth region to obtain multiple suspected coronary microvascular regions connected to the main coronary artery vessels; Overlapping heatmaps were obtained based on suspected areas of coronary microvessels.
9. The intelligent segmentation method for coronary microvascular angiography images according to claim 1, characterized in that, The method of obtaining the comprehensive synergy rate of each suspected coronary microvascular region based on the Pearson correlation coefficient includes: The range of values for the systolic and diastolic Pearson correlation coefficients corresponding to each suspected coronary microvascular region was determined: If either the systolic or diastolic Pearson correlation coefficient falls within the range of [-1, 0], the periodic synergy between the suspected coronary microvascular region and the main coronary artery region is deemed unqualified, and its overall synergy rate is directly set to 0; otherwise, the systolic and diastolic durations are calculated. The systolic duration is the product of the number of frames in the corresponding time-series angiography image interval during systole and the single-frame acquisition interval; the diastolic duration is the product of the number of frames in the corresponding time-series angiography image interval during diastole and the single-frame acquisition interval. Based on the duration of systole and the duration of diastole, a first weighting coefficient and a second weighting coefficient are obtained. Based on the first weighting coefficient and the second weighting coefficient, the comprehensive synergistic rate of the suspected coronary microvascular region is obtained.