Myocardial bridge detection method, device and equipment and storage medium

Abstract

The invention provides a myocardial bridge detection method and device, equipment and a storage medium. The method comprises the following steps: acquiring continuous multi-frame cardiac blood vessel images covering at least one cardiac cycle; for each frame of cardiac blood vessel image, identifying a blood vessel stenosis position and a stenosis rate corresponding to each blood vessel stenosis position; extracting preset feature points of blood vessels in each frame of cardiac blood vessel image, performing position alignment on each frame of cardiac blood vessel image based on the feature points, and determining a cardiac phase direction corresponding to each frame of cardiac blood vessel image; for each blood vessel stenosis position, based on the stenosis rate of the blood vessel stenosis position in each frame of cardiac blood vessel image and the corresponding cardiac phase direction of each frame of cardiac blood vessel image, generating a dynamic curve of the stenosis rate of each blood vessel stenosis position changing along with the cardiac phase direction; and detecting whether each vascular stenosis position is a myocardial bridge based on each dynamic curve. According to the application, accurate detection of the myocardial bridge is realized by tracking the stenosis rate change of the blood vessel stenosis position in different heart periods.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field, and particularly relates to a myocardial bridge detection method, device, equipment and storage medium. BACKGROUND

[0002] In clinical diagnosis of cardiovascular diseases, detection of blood vessel stenosis is a key link for evaluating blood vessel lesions and formulating treatment plans. Traditional detection methods generally rely on static analysis of single frame or part of frames of heart blood vessel images selected manually by doctors. Doctors judge whether there is stenosis and the degree of stenosis by observing the shape of the blood vessel in the image, measuring the diameter of the blood vessel and the like, and diagnose blood vessel related diseases on this basis.

[0003] However, it is found in clinical practice that some patients who are determined as 'no blood vessel stenosis' by traditional detection methods will still have typical symptoms of blood vessel insufficiency such as chest tightness and chest pain, and are finally diagnosed as myocardial bridge after further in-depth examination. This phenomenon prompts researchers to conduct targeted research on the pathological characteristics of myocardial bridge. The results show that myocardial bridge is a special type of blood vessel related lesion, and the core pathological mechanism is that a certain segment of the coronary blood vessel is wrapped by myocardial tissue, resulting in 'periodic dynamic change' characteristics of blood vessel stenosis. During the systolic period, the myocardium wrapped around the blood vessel contracts and presses the blood vessel, causing temporary stenosis of the blood vessel lumen. During the diastolic period, the myocardium relaxes, the blood vessel is no longer pressed, and the lumen returns to normal width.

[0004] This unique dynamic stenosis feature makes the traditional single frame or part of frame static detection scheme completely unsuitable. If the diastolic period image is captured during detection, it will be misjudged as 'no stenosis' because the blood vessel lumen is in a normal state, directly missing the diagnosis of myocardial bridge. If only part of the frame images of the systolic period are captured, although the blood vessel stenosis can be detected, the periodic change rule of the stenosis cannot be captured, and it is difficult to distinguish whether the stenosis is the'systolic period appears and diastolic period disappears' periodic stenosis caused by myocardial bridge or the persistent stenosis caused by ordinary blood vessel lesions, thereby causing misjudgment of the disease type. SUMMARY

[0005] To overcome the problems in the related art, the present application provides a myocardial bridge detection method, device, equipment and storage medium.

[0006] According to a first aspect of an embodiment of the present application, a myocardial bridge detection method is provided, and the method comprises: obtaining continuous multiple frames of heart blood vessel images covering at least one heart cycle; for each frame of the heart blood vessel images, identifying blood vessel stenosis positions and stenosis rates corresponding to each of the blood vessel stenosis positions; extract preset feature points of the blood vessels in each frame of the cardiac blood vessel images, perform position alignment on each frame of the cardiac blood vessel images based on the feature points, and determine a cardiac phase direction corresponding to each frame of the cardiac blood vessel images; For each of the blood vessel stenosis positions, generate a dynamic curve of the stenosis rate of each of the blood vessel stenosis positions varying with the cardiac phase direction based on the stenosis rate of the blood vessel stenosis position in each frame of the cardiac blood vessel images and the cardiac phase direction corresponding to each frame of the cardiac blood vessel images; Detect whether each of the blood vessel stenosis positions is a myocardial bridge based on the dynamic curves.

[0007] According to a second aspect of the embodiments of the present application, a myocardial bridge detection device is provided, and the device comprises: a blood vessel image acquisition module configured to acquire a plurality of continuous frames of cardiac blood vessel images covering at least one cardiac cycle; a blood vessel stenosis position identification module configured to identify, for each frame of the cardiac blood vessel images, a blood vessel stenosis position and a stenosis rate corresponding to each of the blood vessel stenosis positions; a blood vessel image cardiac phase direction determination module configured to extract preset feature points of the blood vessels in each frame of the cardiac blood vessel images, perform position alignment on each frame of the cardiac blood vessel images based on the feature points, and determine a cardiac phase direction corresponding to each frame of the cardiac blood vessel images; a dynamic curve generation module configured to, for each of the blood vessel stenosis positions, generate a dynamic curve of the stenosis rate of each of the blood vessel stenosis positions varying with the cardiac phase direction based on the stenosis rate of the blood vessel stenosis position in each frame of the cardiac blood vessel images and the cardiac phase direction corresponding to each frame of the cardiac blood vessel images; a myocardial bridge detection module configured to detect whether each of the blood vessel stenosis positions is a myocardial bridge based on the dynamic curves.

[0008] According to a third aspect of the embodiments of the present application, a computer device is provided, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method of the first aspect when executing the computer program.

[0009] According to a fourth aspect of the embodiments of the present application, a computer readable storage medium is provided, which stores a computer program, and the computer program is executable on a processor to implement the method of the first aspect.

[0010] The technical solutions provided by the embodiments of the present application can have the following beneficial effects: In this embodiment, by acquiring continuous multi-frame cardiac vascular images covering at least one cardiac cycle, the stenosis rate changes at the same vascular stenosis location in different cardiac phases are tracked and a dynamic curve is generated. Based on the dynamic curve, accurate detection of myocardial bridging is achieved, effectively overcoming the limitations of traditional static detection and improving the accuracy and reliability of myocardial bridging diagnosis.

[0011] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0012] The accompanying drawings, which are incorporated in and form part of this application, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0013] Figure 1 This is a schematic flowchart illustrating a myocardial bridging detection method according to an exemplary embodiment of this application.

[0014] Figure 2 This is a schematic diagram illustrating the distribution of cardiac diastolic feature points according to an exemplary embodiment of this application.

[0015] Figure 3 This is a schematic diagram illustrating the distribution of cardiac systolic feature points according to an exemplary embodiment of this application.

[0016] Figure 4 This is a dynamic curve diagram of a stenotic location corresponding to a myocardial bridge, according to an exemplary embodiment of this application.

[0017] Figure 5 This is a schematic diagram of a curve corresponding to a common continuous narrowing according to an exemplary embodiment of this application.

[0018] Figure 6 This is a schematic diagram illustrating the detection process of myocardial bridging using a detection model according to an exemplary embodiment of this application.

[0019] Figure 7 This is a schematic diagram of a myocardial bridging detection device according to an exemplary embodiment of this application.

[0020] Figure 8 This is a schematic diagram of the structure of a computer device according to an exemplary embodiment of this application. Detailed Implementation

[0021] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0022] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0023] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0024] In the clinical diagnosis of cardiovascular diseases, the detection of vascular stenosis is a key step in assessing vascular lesions and developing treatment plans. Traditional detection methods generally rely on static analysis of single frames or manually selected frames of cardiovascular images. Doctors determine the presence and degree of stenosis by observing the morphology of blood vessels in the images and measuring the diameter of blood vessels, and use this as a basis to diagnose vascular-related diseases.

[0025] However, in clinical practice, it has been found that some patients who were initially diagnosed as having "no vascular stenosis" using traditional testing still subsequently develop typical symptoms of insufficient blood supply, such as chest tightness and chest pain, and are later diagnosed with myocardial bridging after further examination. This phenomenon prompted researchers to conduct targeted studies on the pathological characteristics of myocardial bridging. The results showed that myocardial bridging is a special type of vascular-related lesion, and its core pathological mechanism is that a segment of the coronary artery is surrounded by myocardial tissue, resulting in a "periodic dynamic change" characteristic of vascular stenosis. During cardiac systole, the myocardium surrounding the vessel contracts and compresses the vessel, causing temporary narrowing of the lumen; while during cardiac diastole, the myocardium relaxes, the vascular compression is relieved, and the lumen returns to its normal width.

[0026] This unique dynamic stenosis characteristic makes traditional single-frame or partial-frame static detection schemes completely inadequate: if a diastolic image is acquired during detection, it may be misjudged as "no stenosis" because the blood vessel lumen is in a normal state, directly missing the diagnosis of myocardial bridging; if only partial-frame images of the systolic phase are acquired, although vascular stenosis can be detected, the periodic change pattern of stenosis cannot be captured, making it difficult to distinguish whether the stenosis is a periodic stenosis caused by myocardial bridging that "appears during systole and disappears during diastole" or a persistent stenosis caused by ordinary vascular lesions, thus leading to misjudgment of the disease type.

[0027] Based on this, in order to solve the problems existing in the related technologies, this application provides a method for detecting myocardial bridging. This method acquires continuous multi-frame cardiac vascular images covering at least one cardiac cycle, tracks the stenosis rate changes of the same vascular stenosis location in different cardiac phases and generates a dynamic curve. Based on the dynamic curve, accurate detection of myocardial bridging is achieved, effectively overcoming the limitations of traditional static detection and improving the accuracy and reliability of myocardial bridging diagnosis.

[0028] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0029] Figure 1 This is a schematic flowchart illustrating a method for detecting myocardial bridging according to an exemplary embodiment of this application. Figure 1 As shown, the method includes steps S101 to S105.

[0030] Step S101: Acquire consecutive multi-frame cardiac vascular images covering at least one cardiac cycle.

[0031] The core pathological feature of myocardial bridging is the periodic dynamic change in vascular stenosis during cardiac contraction and relaxation. Relying solely on single frames or fragmented images that do not cover the entire cardiac cycle cannot capture the pattern of "stenosis during systole and normality during diastole," potentially leading to missed diastole or misdiagnosis. Therefore, the acquired image sequence must cover at least one complete cardiac cycle to fully record the vascular status throughout the entire cardiac cycle from contraction to relaxation. Considering that the normal human heart rate range is mostly 60-90 beats per minute, corresponding to a single cardiac cycle duration of approximately 0.67 to 1 second, to ensure that the image sequence fully encompasses all key stages of systole and diastole and avoids missing important vascular status information due to insufficient acquisition time, the acquisition time for coronary artery imaging in practical applications can be set to at least 1 second. This obtains a sufficient number of consecutive frames, providing complete and continuous data support for subsequent tracking of stenosis rate changes at the same stenosis location in different cardiac phases.

[0032] Step S102: For each frame of cardiac vascular image, identify the location of vascular stenosis and the stenosis rate corresponding to each stenosis location.

[0033] This step automatically identifies the location and degree of vascular stenosis from a single frame image through precise vascular region analysis and quantitative calculation, providing frame-level foundational data for subsequent dynamic curve generation. The specific implementation process is as follows: First, vascular region segmentation is performed on cardiac vascular images. The core objective is to separate the pure coronary artery regions from the original cardiac vascular images, which contain complex elements such as bone, soft tissue, and contrast agent background, thus eliminating irrelevant interference for subsequent analysis. In practical applications, the original cardiac vascular image (such as DSA image) can be input into a pre-trained end-to-end segmentation model (such as U-Net, nn-UNet, etc.). This type of model can automatically learn the gray-level features, edge texture, and topological structure of blood vessels through deep neural networks, and output a binary segmentation map that corresponds exactly to the size of the original image (pixel value 1 corresponds to the blood vessel region, and 0 corresponds to the background region), achieving preliminary localization of the blood vessel region. Since the preliminary segmentation may have problems such as missing small branches, incomplete edge segmentation, or misjudgment of background noise, a watershed algorithm can be further used for secondary optimization. That is, by combining the gray-level features of the original image with the prior information of the blood vessel region from the preliminary segmentation result, the blood vessel edges are accurately calibrated, broken blood vessel segments are filled, and discrete background noise points are removed, finally obtaining a blood vessel region image with clear boundaries, completeness, and accuracy.

[0034] After obtaining a precise image of the vascular region, the centerlines of each branch vessel can be extracted using a morphological skeleton extraction algorithm. This algorithm can compress a two-dimensional vascular region into a one-dimensional continuous line while preserving the vascular topology (direction, branching relationships), accurately reflecting the core direction of the vessel. The entire process is automated, effectively avoiding subjective errors caused by manual centerline drawing. To further improve the quality of the centerlines, noise removal can be performed on the extraction results: a length threshold (e.g., 3 pixels) is set, and discrete line segments below the threshold are identified as noise and deleted. Only continuous lines with the required length are retained as valid vascular centerlines, ensuring a one-to-one correspondence with the actual vascular branches. Simultaneously, the flow characteristics of contrast agents can be combined to identify vascular branches: the starting point of contrast agent inflow (coronary ostium) in the initial image frame is identified as the root node, and all effective centerlines are traversed from this node to establish a complete set of vascular branches and clarify the connection relationship of each branch; furthermore, by analyzing the topological structure of the centerlines, when a node shows a bifurcation or intersection, the node is determined to be a vascular bifurcation point, while the part of the centerline that does not bifurcate and continues to extend continuously is the non-bifurcation point of the vascular branch, thus completing the accurate distinction between bifurcation points and non-bifurcation points, avoiding misjudging changes in vascular diameter caused by normal bifurcation as lesion stenosis.

[0035] Next, for each pixel of each valid centerline, the corresponding vessel diameter is precisely calculated. Specifically, the tangent vector of each pixel of the centerline is first determined. For non-endpoint pixels in the centerline pixel sequence, the initial tangent vector is calculated using the coordinate difference between its preceding and following pixels, and Gaussian filtering is used to smooth the gradient of adjacent vectors to avoid vector fluctuations caused by noise. For endpoint pixels, the coordinate difference of adjacent single pixels is directly used as the tangent vector to ensure that the tangent direction conforms to the vessel's orientation. Based on the tangent vector, a normal perpendicular to the vessel's orientation (i.e., the diameter measurement baseline) is generated. With the centerline pixel as the midpoint, rays of a preset length are generated along both sides of the normal (the length on one side can be set to 3-4 mm to adapt to the common thickness range of clinical coronary vessels, ensuring that the rays can completely cover the vessel edge). By identifying the two intersection points of the ray and the edge of the vessel region image, the straight-line distance between the two intersection points is calculated, which is the vessel diameter corresponding to that centerline pixel. By traversing all pixels of the entire centerline, the diameter distribution data of each position of the vessel branch can be obtained.

[0036] Finally, vascular stenosis is determined and the stenosis rate is calculated based on the diameter distribution data. Considering the physiological characteristic of blood vessels being wider proximally and narrowing distally, a large-window Gaussian smoothing method is first used to perform low-frequency smoothing on the diameter distribution data to eliminate local minor fluctuations and obtain a smooth curve reflecting the normal trend of blood vessel diameter changes. For each location at a non-bifurcation point, the average blood vessel diameter of its neighboring region (e.g., the blood vessel segment corresponding to 10 pixels before and after the bifurcation point) is calculated, and this average value is used as the reference standard for the normal blood vessel diameter at that location. A preset percentage threshold is set (e.g., 90%, which can be adjusted according to clinical diagnostic needs). If the actual blood vessel diameter at a non-bifurcation point is less than 90% of the average diameter of its neighboring region, then the location is determined to meet the preset stenosis condition and is identified as a vascular stenosis location. The stenosis rate is calculated as follows: Stenosis rate = (average diameter of neighboring region - actual blood vessel diameter at this location) / average diameter of neighboring region × 100%. For example, if the average diameter of the neighboring region is 4 mm and the actual diameter at this location is 3.6 mm, then the stenosis rate is 10%.

[0037] To better align with clinical diagnostic needs, stenosis locations can be merged and graded: The actual distance between adjacent stenosis locations is calculated according to the centerline pixel sequence, and a distance threshold (e.g., 2-3 mm) is set. If the distance between adjacent stenosis locations is less than this threshold, they are grouped into the same stenosis segment; if the distance is greater than or equal to the threshold, they are considered independent stenosis segments. Simultaneously, the severity of stenosis segments can be graded according to commonly used clinical standards: stenosis rate of 25% or less is Grade I (mild stenosis), 26%-50% is Grade II (moderate stenosis), 51%-75% is Grade III (severe stenosis), and 76% or more is Grade IV (extremely severe stenosis), providing a quantitative reference for subsequent dynamic tracking and clinical diagnosis.

[0038] Step S103: Extract the preset feature points of blood vessels in each frame of cardiac vascular images, align the positions of each frame of cardiac vascular images based on the feature points, and determine the cardiac phase corresponding to each frame of cardiac vascular images.

[0039] Because heartbeats cause periodic displacement of blood vessel positions across different image frames, directly comparing the stenosis rate of narrowing locations in different frames will introduce errors due to inconsistent spatial locations. Furthermore, the stenosis rate of myocardial bridging is directly related to cardiac phase (systole and diastole). Only by clearly defining the cardiac phase for each frame can the correlation between the stenosis rate and cardiac phase be established, thereby detecting the presence of myocardial bridging. Therefore, to accurately track the stenosis rate changes of narrowing locations across different image frames, feature points can be pre-aligned across each frame to ensure that the same narrowing location is at the same spatial coordinates across multiple frames, and cardiac phase determination can be performed to provide a temporal label for subsequent dynamic curve generation.

[0040] Specifically, vascular bifurcation points, vascular terminals, points of abrupt change in vascular diameter, and vascular bends can be selected as preset feature points. This is because these feature points are inherent anatomical structures of blood vessels, do not undergo essential structural changes with cardiac pulsation, and have obvious grayscale or morphological differences in the image, making them easy to extract accurately. More importantly, the relative positional relationship of these feature points remains fixed under the same angiography angle and will not undergo drastic changes due to cardiac pulsation. This provides a reliable benchmark for positional alignment, and their spatial distribution can directly reflect changes in cardiac pulsation.

[0041] In the specific implementation of position alignment, the aforementioned preset feature points can first be extracted from the vascular region of each frame image. A feature vector for each feature point is generated using SIFT (Scale Invariant Feature Transform) or ORB (Fast Feature Extraction and Description) algorithms. This vector uniquely represents the local grayscale texture information of the feature point. Using one frame image (such as the starting frame of the sequence or the frame with the clearest vascular morphology) as the reference frame, the feature points of other frames (target frames) are matched with those of the reference frame. For example, based on feature vector similarity calculation, a one-to-one correspondence between feature points in the target frame and the reference frame is established. For each successfully matched feature point pair, the spatial coordinates of the vascular region in the target frame are corrected by solving the affine transformation matrix, ensuring that the positions of all feature points in the target frame are completely aligned with those in the reference frame. This method achieves spatial uniformity of the vascular structure in all frames, ensuring that subsequent analyses focus on changes in vascular stenosis at the same physical location, and avoiding misjudgments of stenosis rate due to displacement.

[0042] In determining cardiac systole, quantitative analysis can be performed based on the dispersion of feature points. During cardiac systole, myocardial contraction compresses coronary arteries, causing the overall vascular morphology to contract and become compact. This reduces the spatial distance between feature points, resulting in feature point clustering and low dispersion. Figure 2 As shown; however, during diastole, the heart relaxes, blood vessels dilate fully, and the spatial distance between the feature points increases, resulting in the feature points being dispersed and highly discrete, such as... Figure 3 As shown. In the specific implementation process, the spatial distribution density (number of feature points per unit area) of all feature points in each frame of the image or the average Euclidean distance between any two feature points can be calculated as a quantitative indicator of the degree of dispersion: a dispersion threshold is set, and when the average Euclidean distance is greater than the threshold, the corresponding frame is determined to be in the diastolic phase; when the average Euclidean distance is less than or equal to the threshold, the corresponding frame is determined to be in the systolic phase. Through this quantification method, the cardiac phase label can be accurately labeled for each frame of the image, providing a clear time dimension basis for the subsequent generation of dynamic curves.

[0043] Step S104: For each vascular stenosis location, based on the stenosis rate of the vascular stenosis location in each frame of cardiac vascular images and the corresponding cardiac phase of each frame of cardiac vascular images, generate a dynamic curve showing the change of the stenosis rate of each vascular stenosis location with the cardiac phase.

[0044] After single-frame stenosis identification in step S102 and position alignment and cardiac orientation determination in step S103, a four-dimensional data association can be established between vascular stenosis location, frame number, stenosis rate, and cardiac orientation. For each independent vascular stenosis location (including merged stenosis segments), its corresponding stenosis rate data can be extracted from all frame images and bound one-to-one with the cardiac orientation label of each frame to form a set of stenosis rate-cardiac orientation data pairs for that stenosis location.

[0045] When generating a dynamic curve, the cardiac phase can be used as the horizontal axis (two representations are possible: one is to directly label the phase as "systole" and "diastole," and the other is to use a time axis based on the frame number to indirectly reflect the changes in the cardiac cycle), and the stenosis rate (percentage) as the vertical axis. The data points are plotted in the coordinate system according to the chronological order of the cardiac cycle, and then connected by a smooth curve to form a dynamic curve showing the change in the stenosis rate at that location with the cardiac phase. For example, as shown... Figure 4 As shown, the dynamic curve at the stenotic location corresponding to the myocardial bridge exhibits a clear "periodic fluctuation" characteristic: during the systolic phase ( Figure 4 The area indicated by the middle arrow shows a significantly increased stenosis rate, forming the peak of the curve; during diastole, the stenosis rate drops to a normal level (close to 0%), forming the trough of the curve; while the curve corresponding to ordinary persistent stenosis shows no significant fluctuation in the stenosis rate during systole and diastole, remaining at the same high level throughout.Figure 5 As shown in the figure, this dynamic curve visually presents the stenosis rate variation pattern at each stenosis location, providing core data support for the accurate identification of subsequent myocardial bridging.

[0046] Step S105: Based on each dynamic curve, detect whether each vascular stenosis location is a myocardial bridge.

[0047] Dynamic curves visually present the variation of stenosis rate at various vascular stenosis locations with cardiac phase, effectively distinguishing myocardial bridging from ordinary stenosis. However, traditional manual curve analysis is inefficient, subjective, and struggles to accurately capture the subtle fluctuations of minor myocardial bridging. Therefore, in this embodiment, a pre-trained detection model can be further employed to achieve automated and precise detection of myocardial bridging based on dynamic curve features and original cardiac vascular image features. The specific implementation process is as follows: First, data preprocessing is performed to transform the dynamic curves and original cardiac vascular images into an input format recognizable by the model. For the dynamic curve corresponding to each vascular stenosis location, since it is essentially one-dimensional time-series data, directly inputting it into the model is insufficient to fully extract its periodic fluctuation features. Therefore, it can be upscaled using methods such as Generative Adversarial Networks (GANs) or depooling to transform it into a two-dimensional feature image, expanding the one-dimensional stenosis rate time-series data into a two-dimensional matrix (i.e., a feature image) that perfectly matches the size of the region image at that stenosis location. During this process, the core features of the dynamic curve, such as the peak value (systolic stenosis rate), trough value (diastolic stenosis rate), and fluctuation amplitude, are mapped to the grayscale distribution or texture features of the two-dimensional image, preserving the temporal variation pattern while adapting to the model's feature extraction capabilities for two-dimensional data. Simultaneously, the region image corresponding to the vascular stenosis location in each frame of cardiac vascular images (i.e., the stenosis location identified in step S102) is obtained and arranged in frame sequence to form a region image sequence. This sequence contains the morphological features of the stenosis location in different cardiac phases (such as the clarity of the vessel edge, the presence of occlusion, and changes in lumen width), providing the model with intuitive image evidence.

[0048] Subsequently, the two-dimensional feature image and the sequence of regional images corresponding to each vascular stenosis location are used as a set of input data and fed into a pre-trained detection model (preferably a convolutional neural network CNN). This model extracts features from both types of data simultaneously through a deep network structure: it captures the periodic fluctuation features of the stenosis rate from the two-dimensional feature image (such as the unique "systolic peak-diastolic trough" fluctuation pattern of myocardial bridging), and extracts the dynamic change features of vascular morphology from the regional image sequence (such as the difference in morphology between systolic compression and narrowing of the vessel and diastolic restoration). The two types of features are then fused to achieve a comprehensive judgment based on multi-dimensional information.

[0049] The output layer of the detection model can contain three functional neurons, each corresponding to one of the three output categories, and each neuron uses a specific activation function to adapt to the output requirements: The first neuron (corresponding to the first determination result): can be activated using the Sigmoid function, with an output value ranging from 0 to 1, used to determine whether the corresponding vascular stenosis location is a myocardial bridging. A preset threshold (e.g., 0.5) is set; when the output value is greater than the threshold, the location is determined to be a myocardial bridging; when the output value is less than or equal to the threshold, it is determined to be a non-myocardial bridging (e.g., ordinary persistent stenosis). The determination logic of this neuron is based on the model's comprehensive learning of "fluctuation characteristics + morphological characteristics," enabling it to accurately distinguish between the periodic dynamic stenosis of myocardial bridging and the persistent stenosis of ordinary stenosis. The second neuron (corresponding to the second judgment result): It can also be activated by the Sigmoid function, with an output value ranging from 0 to 1, used to determine whether the stenosis location of the blood vessel is a true stenosis location without obstruction. When the output value is greater than a preset threshold (e.g., 0.5), it indicates that the stenosis location has no obvious obstruction in multiple frames of images, the blood vessel morphology is clear, and it is judged as a true stenosis; when the output value is less than or equal to the threshold, it indicates that the stenosis location has obstruction (e.g., obstruction by bone or soft tissue) or angular interference, and it is judged as a suspected false stenosis, avoiding misjudgment caused by obstruction; The third neuron (corresponding to the third output): can be activated using the ReLU function, with an output range of 0-100%, used to output the true stenosis rate at the location of the vascular stenosis. The model automatically selects the image frame with the best observation effect from the regional image sequence (i.e., the frame with the clearest blood vessel edges, no occlusion, and the most accurate lumen measurement), and uses the stenosis rate calculated in this frame as the true stenosis rate output. In this way, even if some frames are occluded, the quantification accuracy of the stenosis rate can be ensured by using the optimal frame, providing a reliable quantitative indicator for clinical diagnosis. Specifically, the detection process of the detection model is as follows: Figure 6 As shown.

[0050] Finally, based on the model's three types of output results, the final detection conclusion is formed: if the first determination result is myocardial bridging, a "positive myocardial bridging" conclusion is drawn, along with the corresponding dynamic curve (for doctors to verify fluctuation characteristics) and the true stenosis rate (for assessing the severity of the lesion) in the third output result; if the first determination result is not myocardial bridging, but the second determination result is true stenosis, a "common vascular stenosis" conclusion and the true stenosis rate are drawn; if the second determination result is suspected false stenosis, a "false stenosis" conclusion is drawn, which can indicate to clinicians that no intervention is needed for this location. Through this comprehensive detection process, both automated and accurate identification of myocardial bridging is achieved, and problems such as misjudgment due to occlusion and inaccurate quantification of stenosis rate are solved, effectively adapting to the diverse practical needs of clinical diagnosis.

[0051] Corresponding to the embodiments of the aforementioned myocardial bridging detection method, this application also provides a myocardial bridging detection device. Figure 7 This is a schematic diagram of a myocardial bridging detection device according to an exemplary embodiment of this application. Figure 7 As shown, the device includes: The vascular image acquisition module 701 is used to acquire a series of multiple frames of cardiac vascular images covering at least one cardiac cycle. The vascular stenosis location identification module 702 is used to identify the location of vascular stenosis and the stenosis rate corresponding to each vascular stenosis location for each frame of cardiac vascular image; The cardiac phase determination module 703 for vascular images is used to extract preset feature points of blood vessels in each frame of cardiac vascular images, perform position alignment of each frame of cardiac vascular images based on the feature points, and determine the cardiac phase corresponding to each frame of cardiac vascular images. The dynamic curve generation module 704 is used to generate a dynamic curve of the stenosis rate of each stenosis location as a function of the cardiac phase, based on the stenosis rate of the stenosis location in each frame of cardiac vascular images and the corresponding cardiac phase of each frame of cardiac vascular images. The myocardial bridging detection module 705 is used to detect whether the stenosis location of each blood vessel is a myocardial bridging based on each dynamic curve.

[0052] The specific implementation process of the functions and roles of each module in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0053] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and 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 application according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0054] Corresponding to the embodiments of the aforementioned myocardial bridging detection method, this application also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor; wherein, when the processor executes the computer program, it implements the steps of the myocardial bridging detection method described in any of the above embodiments.

[0055] For example, processors include, but are not limited to, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), application-specific integrated circuits (ASICs), or field-programmable gate arrays (FPGAs).

[0056] For example, the memory may include at least one type of storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc.

[0057] Figure 8 This is a schematic diagram illustrating the structure of a computer device according to an exemplary embodiment of this application. Figure 8 As shown, at the hardware level, the computer device includes a processor 801, an internal bus 802, a network interface 803, memory 804, and non-volatile memory 805, and may also include other hardware required for business operations. One or more embodiments of this application can be implemented in software, for example, the processor 801 reads the corresponding computer program from the non-volatile memory 805 into the memory 804 and then runs it. Of course, in addition to software implementation, one or more embodiments of this application do not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the above processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0058] Corresponding to the embodiments of the aforementioned myocardial bridging detection method, this application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the myocardial bridging detection method described in any of the above embodiments.

[0059] Corresponding to the embodiments of the aforementioned myocardial bridging detection method, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the myocardial bridging detection method described in any of the above embodiments.

[0060] The foregoing has described specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0061] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention filed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not claimed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the foregoing claims.

[0062] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

[0063] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for detecting myocardial bridging, characterized in that, Comprising: acquiring a plurality of continuous cardiac vessel images covering at least one cardiac cycle; for each frame of the cardiac vessel images, identifying a blood vessel stenosis position and a stenosis rate corresponding to each of the blood vessel stenosis positions; extracting a preset feature point of the blood vessel in each frame of the cardiac vessel images, aligning the positions of each frame of the cardiac vessel images based on the feature points, and determining a cardiac phase direction corresponding to each frame of the cardiac vessel images; for each of the blood vessel stenosis positions, generating a dynamic curve of the stenosis rate of each of the blood vessel stenosis positions varying with the cardiac phase direction based on the stenosis rate of the blood vessel stenosis position in each frame of the cardiac vessel images and the corresponding cardiac phase direction of each frame of the cardiac vessel images; based on each of the dynamic curves, detecting whether each of the blood vessel stenosis positions is a myocardial bridge.

2. The myocardial bridge detection method of claim 1, wherein, The detection of whether each of the blood vessel stenosis positions is a myocardial bridge based on each of the dynamic curves comprises: performing dimensionality increasing processing on each of the dynamic curves respectively to obtain a feature image matching the size of the region corresponding to the blood vessel stenosis position; acquiring a region image corresponding to each of the blood vessel stenosis positions in each frame of the cardiac vessel images to form a region image sequence; inputting each of the feature images and the corresponding region image sequence into a pre-trained detection model, and determining whether each of the blood vessel stenosis positions is a myocardial bridge based on a first determination result output by the detection model.

3. The myocardial bridge detection method of claim 2, wherein, The output of the detection model further includes a second determination result, which is used to determine whether each of the blood vessel stenosis positions is a real stenosis position without occlusion.

4. The myocardial bridge detection method of claim 3, wherein, The output of the detection model further includes a third output result, which is a real stenosis rate of each of the blood vessel stenosis positions, and the real stenosis rate is determined by the detection model from an image frame with the best observation effect in each frame of the cardiac vessel images.

5. The myocardial bridge detection method of claim 1, wherein, The detection of whether each of the blood vessel stenosis positions is a myocardial bridge based on each of the dynamic curves comprises: performing blood vessel region segmentation and blood vessel centerline extraction on the cardiac vessel images to identify the bifurcation and non-bifurcation of the blood vessel; calculating the blood vessel diameter corresponding to each position of the blood vessel centerline, determining the non-bifurcation position of the blood vessel as the blood vessel stenosis position when the blood vessel diameter is less than the average blood vessel diameter of the adjacent region by a preset proportion threshold, and calculating the stenosis rate of the blood vessel stenosis position.

6. The myocardial bridge detection method of claim 1, wherein, The determination of the cardiac phase direction corresponding to each frame of the cardiac vessel images based on the feature points comprises: determining the cardiac phase direction corresponding to each frame of the cardiac vessel images based on the discrete degree of the feature points in each frame of the cardiac vessel images.

7. The myocardial bridge detection method of claim 1, wherein, The feature points include blood vessel bifurcation points, blood vessel tips, blood vessel diameter mutation points, and blood vessel bending points.

8. A myocardial bridge detection apparatus characterized by, Comprising: a blood vessel image acquisition module configured to acquire a plurality of continuous cardiac vessel images covering at least one cardiac cycle; a blood vessel stenosis position identification module configured to identify a blood vessel stenosis position and a stenosis rate corresponding to each of the blood vessel stenosis positions for each frame of the cardiac vessel images; a blood vessel image cardiac phase direction determination module configured to extract a preset feature point of the blood vessel in each frame of the cardiac vessel images, align the positions of each frame of the cardiac vessel images based on the feature points, and determine a cardiac phase direction corresponding to each frame of the cardiac vessel images; a dynamic curve generation module, configured to generate, for each of the vessel stenosis positions, a dynamic curve of the stenosis rate of the vessel stenosis position varying with the heart phase direction based on the stenosis rate of the vessel stenosis position in each frame of the heart vessel images and the corresponding heart phase direction of each frame of the heart vessel images; a myocardial bridge detection module, configured to detect, based on each of the dynamic curves, whether each of the vessel stenosis positions is a myocardial bridge.

9. A computer device, comprising: A computer program product, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the method of any one of claims 1 to 7. The computer program is executed by the processor to implement the method of any one of claims 1 to 7.

Images