A method and system for intelligent classification of cardiovascular angiography images

Abstract

The present application relates to the technical field of image processing, in particular to a kind of cardiovascular angiography image intelligent classification method and system, the method comprises: using each branch point after cardiovascular angiography image vascular skeletonization determines the corresponding blood vessel section of each branch;Using the deviation of the inner diameter of the blood vessel in the target blood vessel section determines its stenosis region and stenosis degree, and then determines its initial abnormality index;The feature registration of multiple frames of cardiovascular angiography is carried out to obtain registration target blood vessel section, and the change of initial abnormality index of registration target blood vessel section before and after the inflow of contrast medium is used to determine its stenosis relief degree;Determine the respective blood vessel improvement response degree of the upstream and downstream branches of the registration target blood vessel section after stenosis relief, and then determine the target stenosis index of the registration target blood vessel section in combination with the stenosis relief degree, and then classify and early warn based on the target stenosis index.Through the technical scheme of the present application, more accurate quantization representation of vascular stenosis and fine classification of vascular stenosis state are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a kind of cardiovascular angiography image intelligent classification method and system. BACKGROUND

[0002] As a kind of technology that can present the morphology in the blood vessel lumen by injecting contrast medium into blood vessel through imaging device, cardiovascular angiography can accurately reveal whether there is stenosis, occlusion or abnormal dilatation in coronary artery, and has irreplaceable value for helping doctors to diagnose cardiovascular diseases. When patients have chest tightness, chest pain and other symptoms, cardiovascular angiography can directly show the position of vascular stenosis, and provide key data basis for judging whether stent implantation or bypass surgery is needed. From early screening of occult vascular lesions to postoperative evaluation of vascular recanalization effect, cardiovascular angiography continues to provide scientific support for the diagnosis and treatment of cardiovascular diseases, and is an important technical guarantee in modern cardiovascular medicine.

[0003] The degree of vascular stenosis directly affects the normal flow of blood in the blood vessel, not only causes the increase of blood flow resistance, but also may induce blood vessel related diseases. Scientific classification based on the degree of vascular stenosis can provide convenience for clinicians to quickly judge the severity of the disease. Since the existing scheme only analyzes the stenosis of a single blood vessel in isolation, it is difficult to fully reflect the true state of blood in the blood vessel segment, resulting in low division fineness of blood vessel network in the image. SUMMARY

[0004] In order to solve the technical problem of low division fineness of blood vessel network in the existing scheme of cardiovascular image, the purpose of the present application is to provide a kind of cardiovascular angiography image intelligent classification method and system, and the technical scheme adopted is as follows:

[0005] The present application provides a kind of cardiovascular angiography image intelligent classification method, the method comprises:

[0006] Determine the blood vessel segment corresponding to each branch using each branch point after the blood vessel skeletonization of the cardiovascular angiography;

[0007] Determine the stenosis region and its stenosis degree in the target blood vessel segment using the blood vessel inner diameter deviation, and determine the initial abnormality index of the target blood vessel segment using each stenosis degree;

[0008] Perform feature registration on multiple frames of cardiovascular angiography using the blood vessel skeleton to obtain a registered target blood vessel segment, and determine the stenosis relief degree of the registered target blood vessel segment using the change of initial abnormality index before and after the contrast medium flows in;

[0009] Determine the blood vessel improvement response degree of the upstream branch and the downstream branch of the registered target blood vessel segment respectively after stenosis relief;

[0010] Determine a target stenosis index of the registration target vessel segment by using the respective blood vessel improvement response degree and the stenosis relief degree, and classify and warn the registration target vessel segment based on the target stenosis index.

[0011] Further, the step of determining the respective blood vessel segment corresponding to each branch point after the blood vessel skeletonization of the cardiovascular angiogram comprises:

[0012] Determine each skeleton node after the blood vessel skeletonization of the cardiovascular angiogram, and determine the cosine similarity between the front blood vessel direction vector and the rear blood vessel direction vector of the skeleton node;

[0013] Compare the cosine similarity with a preset similarity threshold to determine the real branch points in the skeleton node, and obtain the corresponding blood vessel segment by taking the branch point as the boundary of each branch.

[0014] Further, the step of determining the stenosis region and the stenosis degree thereof in the target vessel segment by using the blood vessel inner diameter deviation in the target vessel segment, and determining the initial abnormality index of the target vessel segment by using the respective stenosis degree comprises:

[0015] Determine the blood vessel inner diameter at the target sampling point in the target vessel segment, the average inner diameter of the window where the target sampling point is located, and the blood vessel inner diameter deviation of the blood vessel inner diameter relative to the average inner diameter;

[0016] Compare the blood vessel inner diameter deviation with a preset deviation threshold, and take the target sampling point greater than the preset deviation threshold as a stenosis sampling point;

[0017] Start traversing from the stenosis sampling point to both ends along the blood vessel axis until the blood vessel inner diameter deviations at both ends are less than or equal to the preset deviation threshold, and stop traversing, to obtain the stenosis region and the stenosis degree thereof;

[0018] Determine the difference ratio weight between the stenosis length of the stenosis region and the total blood vessel length of the target vessel segment, and determine the initial abnormality index of the target vessel segment by using the difference ratio weight and the stenosis degree.

[0019] Further, the step of obtaining the stenosis region and the stenosis degree thereof comprises:

[0020] Take the minimum blood vessel inner diameter in the stenosis region as the target pipe diameter of the stenosis region, and determine the normal pipe diameter mean of the normal blood vessel inner diameter of the target vessel segment;

[0021] Calculate the stenosis degree of the stenosis region by using the target pipe diameter and the normal pipe diameter mean.

[0022] Further, the step of performing feature registration on the plurality of cardiovascular angiograms by using the blood vessel skeleton to obtain the registration target vessel segment comprises:

[0023] taking a cardiovascular image when a blood vessel is not filled with contrast agent as a reference frame, and taking a subsequent cardiovascular image when the blood vessel is filled with contrast agent as a subsequent frame;

[0024] extracting a plurality of matched feature point pairs from the vessel skeletons of the reference frame and the subsequent frame, and minimizing spatial distance errors of the feature point pairs to obtain a registered target vessel segment after rigid registration.

[0025] Further, the minimizing spatial distance errors of the feature point pairs to obtain the registered target vessel segment after rigid registration further comprises:

[0026] sampling each vessel skeleton of the reference frame and the subsequent frame at a preset distance to obtain a plurality of skeleton sampling points, and determining coordinate differences of each skeleton sampling point between the subsequent frame and the reference frame;

[0027] solving the coordinate differences by using a thin plate spline difference function to obtain a global deformation field, and correcting pixel positions of the subsequent frame by using the global deformation field to obtain the registered target vessel segment after elastic registration.

[0028] Further, the determining the stenosis relief degree of the registered target vessel segment according to the change of the initial abnormality index before and after the contrast agent flows in comprises:

[0029] determining an initial abnormality index of the registered target vessel segment before the contrast agent flows in and a minimum abnormality index of the registered target vessel segment after the contrast agent flows in;

[0030] calculating the stenosis relief degree of the registered target vessel segment according to a change difference between the initial abnormality index before the contrast agent flows in and the minimum abnormality index after the contrast agent flows in.

[0031] Further, the determining the respective vascular improvement response degrees of the upstream branch and the downstream branch of the registered target vessel segment after the stenosis relief comprises:

[0032] calculating an abnormality index of the upstream branch of the target vessel segment according to a difference ratio weight of an abnormality degree of the upstream branch compared with an overall abnormality degree of all upstream branches, and combining the abnormality degree of the upstream branch; wherein the abnormality degree comprises a stenosis degree and a dilation degree;

[0033] determining a first abnormality index and a second abnormality index of the upstream branch of the registered target vessel segment before the contrast agent flows in and after the stenosis relief, respectively;

[0034] calculating the vascular improvement response degree of the upstream branch of the registered target vessel segment according to the first abnormality index and the second abnormality index.

[0035] Further, the determining the target stenosis index of the registered target vessel segment according to the respective vascular improvement response degrees and the stenosis relief degree comprises:

[0036] determining a target length ratio of the target blood vessel segment relative to a maximum relief length of all blood vessel segments in the target relief length of the stenosis relief process;

[0037] The target stenosis index of the target blood vessel segment is calculated by using the target length ratio, the stenosis relief degree, and the blood vessel improvement response degree of each of the upstream branch and the downstream branch.

[0038] The application also provides a cardiovascular angiography image intelligent classification system, which comprises a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the method according to any one of the above when executing the computer program.

[0039] The application has the following beneficial effects:

[0040] The application divides the entire blood vessel network into several functional blood vessel segments, and classifies the blood vessel segments as the basic unit. Since the blood vessel is a continuous circulatory system, it is difficult to comprehensively reflect the real state of blood in the blood vessel segment if only the stenosis condition of a single blood vessel is analyzed in isolation. Therefore, the application not only extracts the stenosis degree of the blood vessel segment itself, but also analyzes the features of the upstream and downstream branches of the blood vessel segment, realizes more accurate quantitative representation of the blood vessel stenosis and fine classification of the blood vessel stenosis state, provides more comprehensive data for the classification of the blood vessel segment, and is further beneficial to providing more scientific data reference for the diagnosis and treatment of the blood vessel disease. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0042] Figure 1 A step flowchart of a cardiovascular angiography image intelligent classification method provided by an embodiment of the application;

[0043] Figure 2 A detailed flowchart of step S1 in a cardiovascular angiography image intelligent classification method provided by an embodiment of the application;

[0044] Figure 3 A detailed flowchart of step S2 in a cardiovascular angiography image intelligent classification method provided by an embodiment of the application;

[0045] Figure 4A refinement flowchart of step S3 in a cardiovascular angiography image intelligent classification method provided by an embodiment of the present application;

[0046] Figure 5 A refinement flowchart of step S4 in a cardiovascular angiography image intelligent classification method provided by an embodiment of the present application;

[0047] Figure 6 A structural schematic diagram of a hardware running environment of a cardiovascular angiography image intelligent classification device related to an embodiment of the present application;

[0048] Figure 7 A framework structural schematic diagram of a cardiovascular angiography image intelligent classification system related to an embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to further clarify the technical means and effects taken by the present application to achieve the predetermined object of the application, the following describes in detail the specific implementation, structure, features and effects of a cardiovascular angiography image intelligent classification method according to the present application, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0051] Before unfolding each of the following embodiments, it needs to be explained that the main object of the present application is to:

[0052] An intelligent classification system of angiography images based on the stenosis degree of the blood vessel segments. After dividing the blood vessels in the angiography into numerous blood vessel segments, the changes of the features of each blood vessel segment itself and its upper and lower branches in the time sequence images are used as the basis to achieve the fine division of the blood vessel network in the images.

[0053] The specific scenario targeted by the present application can be:

[0054] When a doctor performs clinical diagnosis on a patient according to the angiography images, the classification results generated according to the stenosis degree of the blood vessels can assist the doctor to quickly lock the position of the blood vessel segment with higher stenosis degree, thereby improving the efficiency and accuracy of the diagnosis process.

[0055] The following specifically describes the specific scheme of a cardiovascular angiography image intelligent classification method provided by the present application in combination with the accompanying drawings.

[0056] Embodiment one:

[0057] For the cardiovascular angiogram intelligent classification method provided by the application, please refer to Figure 1 , which shows the step flow chart of the cardiovascular angiogram intelligent classification method provided by an embodiment of the application.

[0058] The cardiovascular angiogram intelligent classification method comprises:

[0059] Step S1, determining the blood vessel segments corresponding to each branch by using each branch point after blood vessel skeletonization of the cardiovascular angiogram;

[0060] In this embodiment, the contrast catheter is placed into the target cardiovascular region through peripheral blood vessels by interventional technology. While injecting contrast agent, the filling process of the contrast agent in the blood vessel is dynamically captured by using a CT (Computed Tomography) angiography device to obtain multiple continuous high-resolution blood vessel lumen shape images, i.e., a cardiovascular angiogram. The contrast agent injection rate and dose are accurately controlled throughout the process, and the device shooting parameters are adjusted synchronously to ensure that the image clearly presents the details such as blood vessel shape, pipe diameter change and stenosis, thereby providing a high-quality original data basis for subsequent blood vessel segment analysis and classification.

[0061] In order to facilitate blood vessel classification, the complex network structure of the blood vessel needs to be divided. Therefore, the blood vessel centerline is extracted by blood vessel skeletonization, and the skeleton nodes are used as landmarks to divide the blood vessel segments, which can not only realize the unit disassembly of the blood vessel network, but also establish a standardized spatial reference for subsequent stenosis degree evaluation, i.e., taking each blood vessel segment as an independent analysis unit, and quantifying the blood vessel stenosis degree by combining the feature of the upstream and downstream branches.

[0062] Specifically, please refer to Figure 2 , the step S1 comprises:

[0063] Step S11, determining each skeleton node after blood vessel skeletonization of the cardiovascular angiogram, and determining the cosine similarity of the included angle between the front blood vessel direction vector and the rear blood vessel direction vector of the skeleton node;

[0064] Step S12, comparing the cosine similarity with a preset similarity threshold to determine each branch point in the skeleton node and obtaining the corresponding blood vessel segment by taking the branch point as the boundary of each branch.

[0065] In this embodiment, when different vessels intersect in space, their projections will present similar patterns at the branch points on the image, but these "pseudo" branch points do not have the structural characteristics of real branch points, so it is necessary to remove these "pseudo" branch points from the skeleton nodes before formal segmentation. Since the vessel directions on both sides of the intersection point present continuity in space (for example, the main trend of the vessels before the intersection point is left and right, and the branches after the node still maintain this trend), the direction vector v1 of the vessel before the intersection point and the direction vector v2 of the vessel after the intersection point are calculated, and a discrimination formula based on the cosine similarity of the included angle (the existing cosine similarity formula) is constructed: When , it is determined that the node is a pseudo node generated by vessel intersection and is excluded; if , it is confirmed as a real branch node, which is a boundary point limit for vessel segment division ( is a preset similarity threshold, which can be 0.7-0.9, and is adjusted according to the actual situation).

[0066] After screening and removing the pseudo nodes, the skeleton nodes correspond to the real branch points in the vessel network. The vessel network is segmented using these branch nodes as segmentation points, and the corresponding vessel segments are obtained, which can provide a structural division basis for subsequent vessel classification based on the degree of stenosis.

[0067] In step S2, the stenosis region and the degree of stenosis in the target vessel segment are determined by the vessel inner diameter deviation, and the initial abnormality index of the target vessel segment is determined by the degree of stenosis.

[0068] Before quantifying the stenosis of the target vessel segment (any vessel segment), the normal region of the vessel diameter in the vessel segment (i.e., the normal diameter reference D0) needs to be calculated. This is because the normal diameter is the core reference for judging the stenosis of the vessel segment. Only by determining the normal diameter of the vessel in the physiological state can the severity of stenosis be quantified by the deviation of the actual diameter from the normal diameter. Without this reference, the judgment of stenosis will lack a unified reference and cannot accurately distinguish between normal physiological fluctuations and stenosis regions.

[0069] Specifically, please refer to Figure 3 , the step S2 comprises:

[0070] In step S21, the vessel inner diameter at the target sampling point in the target vessel segment, the average inner diameter of the window where the target sampling point is located, and the vessel inner diameter deviation of the vessel inner diameter relative to the average inner diameter are determined.

[0071] In step S22, the vessel inner diameter deviation is compared with the preset deviation threshold, and the target sampling point greater than the preset deviation threshold is taken as a stenosis sampling point.

[0072] Step S23: Starting from the narrow sampling point, traverse along the blood vessel axis to both ends until the deviation of the inner diameter of the blood vessels at both ends is less than or equal to the preset deviation threshold, and stop traversing to obtain the narrow region and its degree of narrowing.

[0073] Specifically, step S23, which obtains the narrow region and its degree of narrowness, includes:

[0074] The minimum vascular diameter in the narrow region is taken as the target vascular diameter of the narrow region, and the mean normal vascular diameter of the target vascular segment is determined.

[0075] The degree of narrowing in the narrow area is calculated using the average of the target pipe diameter and the normal pipe diameter.

[0076] Step S24: Determine the difference ratio weight between the stenosis length of the stenosis region and the total vascular length of the target vascular segment, and use the difference ratio weight and the degree of stenosis to determine the initial abnormality index of the target vascular segment.

[0077] In this embodiment, sampling points are evenly distributed within the blood vessel segment (the number of sampling points can be selected as 10-20, and can be adjusted according to the actual situation). At each sampling point P... h At this location, the inner diameter of the blood vessel perpendicular to the vessel axis is measured and recorded as follows. Because blood vessels are tubular structures, their diameter does not change drastically under normal physiological conditions. Therefore, abnormal points that deviate from the local trend can be identified through local smoothing.

[0078] Choose a window width *l*, for example, *l=3* means that the local reference value for each sampling point is calculated from itself and one point before and after it. For the *h*th sampling point (as the target sampling point, referring to any sampling point), the local smoothing value (the average inner diameter of the window it belongs to) is... (Edge points can reduce window size, such as the first point using...) ), calculate the absolute deviation (vessel diameter deviation) between the actual diameter and the local smoothed value at all sampling points. The 90th quantile of the deviation value is taken as the threshold E for local fluctuation (a preset deviation threshold, which can be adjusted), that is, the deviation of 90% of normal (sampling) points is ≤ E. For any sampling point, if If the diameter is ≤E, it is considered a normal pipe diameter, and the corresponding target sampling point is a normal sampling point; otherwise, it is a narrow pipe diameter and a narrow sampling point.

[0079] For the sampling point P that is determined to be narrow h From P h Starting from a point, traverse along the vessel axis (obtained through skeletal mapping) towards both ends, recording the vessel diameter along the path. Stop traversing when both ends of the traversal have normal vessel diameters, thus identifying the corresponding narrowed region. Record the smallest diameter D within this narrowed region during the traversal. minThe minimum vessel diameter is taken as the representative diameter of the stenosis, i.e. the target diameter. A spatial weight factor (difference ratio weight) is introduced For the i-th stenosis region, the length of the traversed stenosis region is denoted as the stenosis length L i The total length of the vessel segment is denoted as the total vessel length L A The weight formula is For the abnormal (stenosis) case of the vessel segment stenosis diameter It can be assigned as:

[0080]

[0081] In the formula, The stenosis degree of the minimum stenosis diameter in the stenosis region i can be taken as the stenosis degree of the stenosis region i: Wherein is the average value of all diameters meeting the normal diameter requirement (normal diameter mean value), The minimum vessel diameter in the stenosis region is taken as the target diameter. The presence of stenosis in the vessel segment is a pathological abnormality, which is assigned as negative: For the case where multiple stenosis regions may occur in the vessel segment, the assignment of the vessel with multiple stenosis regions can be obtained by superposition: Where n represents the total number of stenosis regions in the target vessel segment. Here, The initial stenosis index of the target vessel segment is taken as the initial abnormal index. The initial abnormal index can be divided into the initial stenosis index and the initial expansion index. In fact, the abnormality in each embodiment refers to stenosis or expansion.

[0082] Step S3, using the vessel skeleton to perform feature registration on multiple frames of cardiovascular angiography to obtain a registered target vessel segment, and using the change of the initial abnormal index of the registered target vessel segment before and after the inflow of the contrast agent to determine the stenosis relief degree thereof;

[0083] In the above embodiments, the abnormality indexes of each blood vessel segment extracted based on a single frame image are used to construct a quantitative basis for the blood vessel structure. On this basis, multi-frame tracking analysis is performed on these indicators to further capture the evolution law of the blood vessels over time: when the contrast agent flows in, the blood flow impact can cause the originally narrow area to appear a temporary physiological expansion, the degree of stenosis is temporarily alleviated, and the corresponding abnormality index is lowered; if the blood vessel segment stenosis is temporarily alleviated under the impact of blood flow, the expansion state of the upstream branch formed due to blood flow congestion will also be alleviated, and the abnormality index of the expansion will be lowered; when the blood vessel segment stenosis is improved due to blood flow, the blood flow of the downstream branch increases, and the small stenosis caused by ischemia is alleviated, and the abnormality index of the stenosis is lowered. However, before capturing these dynamic changes, it is necessary to exclude the interference of physiological activities such as heartbeat, which can cause non-pathological fluctuations in the blood vessel image and may mask the real blood flow changes when the contrast agent passes through.

[0084] Heartbeat can cause the overall displacement or local deformation of blood vessels, and such movement is manifested in multi-frame images as the spatial position and morphology of the same blood vessel structure being inconsistent in different frames. For example, the stenosis region of a blood vessel segment in a certain frame of image may be misjudged as a normal region with position deviation or as a change in the degree of stenosis due to deformation in another frame. In order to eliminate the spatial position deviation of the blood vessel structure in multi-frame images due to physiological movement (such as heartbeat), the deviation needs to be calibrated to align the blood vessel structure in space in multi-frame images.

[0085] Specifically, referring to Figure 4 , the step S3 of obtaining a registration target blood vessel segment by feature registration of the multi-frame cardiovascular angiograms using the blood vessel skeleton includes:

[0086] Step S31, taking the cardiovascular angiogram when the contrast agent is not filled in the blood vessel as a reference frame, and taking the subsequent cardiovascular angiogram when the contrast agent is filled in the blood vessel as a subsequent frame;

[0087] Step S32, extracting a plurality of matched feature point pairs from the blood vessel skeletons of the reference frame and the subsequent frame, and minimizing the spatial distance error of the feature point pairs to obtain the registration target blood vessel segment after rigid registration.

[0088] In this embodiment, the cardiovascular angiogram when the contrast agent is not filled is selected as the reference frame with relatively stable blood vessel morphology, and the blood vessel segment at this time is not subjected to the dynamic image of the contrast agent flow, which is close to the natural static physiological state of the blood vessel and can be used as a spatial template for the subsequent frame. Correspondingly, the cardiovascular angiogram when the contrast agent is filled in the blood vessel is taken as the subsequent frame.

[0089] Because the vascular skeleton can stably represent the vascular direction and branch connection relationship, even if the blood vessels are displaced or deformed due to heartbeat, the structure of the vascular skeleton can remain consistent. Therefore, the vascular skeleton (the central axis of the blood vessels) is selected as the registration feature.

[0090] The overall displacement caused by heartbeat is the most obvious error source. Because the vascular structure has an unchanged structure (such as branch connection and skeleton direction) in different frames, only the spatial position is different. A rigid registration can be used to align the reference frame and the subsequent frame through geometric changes: q matching feature point pairs are extracted from the vascular skeletons of the reference frame and the subsequent frame, the reference frame feature point set is represented as , and the subsequent frame feature point set is represented as . The goal is to minimize the spatial distance error between the feature point pairs, and the vascular displacement registration error function is defined as:

[0091]

[0092] In the formula, , the vascular displacement registration error function is represented, and the rotation angle, the x-axis direction translation, and the y-axis direction translation in the brackets respectively represent the rotation angle, the x-axis direction translation, and the y-axis direction translation that need to be optimized; here, q represents the number of feature point pairs; represents the coordinates of the kth feature point in the reference frame; represents the coordinates of the corresponding kth feature point in the subsequent frame after rotation and translation transformation; represents the square of the Euclidean distance between the two coordinates (any feature point pair). The is solved by the least square method when E is minimum. The vascular pixels in the subsequent frame are subjected to overall registration, and the registration target vascular segment after rigid registration is obtained.

[0093] After the step S32, the method further includes:

[0094] Each skeleton sampling point is sampled along the vascular skeleton of the reference frame and the subsequent frame at a preset distance, and the coordinate difference of each skeleton sampling point between the subsequent frame and the reference frame is determined;

[0095] The global deformation field is obtained by solving the coordinate difference using a thin plate spline difference function, and the pixel position of the subsequent frame is corrected using the global deformation field to obtain the registration target vascular segment after elastic registration.

[0096] In the present embodiment, the rigid registration mentioned in the above embodiment cannot handle the local deformation caused by heart beat (e.g. the periodic expansion and contraction of the vessel wall with blood pressure fluctuation, the slight bending of branch vessels). These deformations will cause the same vessel segment to have different vessel morphology in the reference frame and the subsequent frame, which needs to be further corrected by elastic registration. In the subsequent frame and the reference frame after rigid registration, sample every preset fixed distance (e.g. 0.5mm, which can be adjusted) along the vessel skeleton. For the coordinate difference of each skeleton sampling point (in order to distinguish from the above sampling points, hereinafter referred to as skeleton sampling points) in the subsequent frame and the reference frame, solve the global deformation field by TPS (Thin Plate Spline, thin plate spline) difference function . Apply the global deformation field to each vessel pixel in the subsequent frame to obtain the corrected pixel position by resampling , and finally make the local morphology of the vessel in the subsequent frame consistent with that in the reference frame to obtain the registered target vessel segment after elastic registration.

[0097] In addition, the step S3 determines the stenosis relief degree of the registered target vessel segment by using the change of the initial abnormal index before and after the contrast agent inflow, specifically comprising:

[0098] Determine the initial abnormal index before the contrast agent inflow and the minimum abnormal index after the contrast agent inflow of the registered target vessel segment respectively;

[0099] Calculate the stenosis relief degree of the registered target vessel segment by using the change difference between the initial abnormal index before the contrast agent inflow and the minimum abnormal index after the contrast agent inflow.

[0100] For the registered target vessel segment itself, after the contrast agent inflow, the stenosis index of the vessel segment decreases from the original value to the minimum value, and the relative amplitude in this process reflects the expansion effect of the blood flow on the stenosis region (referred to as stenosis relief degree): , wherein is the initial stenosis index of the vessel segment before the contrast agent inflow (initial abnormal index); is the minimum value of the stenosis index during the contrast agent inflow or after the inflow, i.e. the minimum abnormal index. represents the change difference between the two. The stenosis relief degree The greater the value, the more obvious the stenosis relief. It should be noted that in order to prevent the case where the denominator is zero, a very small value a can be added, for example a = 0.000001. In addition, for the denominator in each formula in the context, in order to avoid the extreme case of the denominator being 0, the very small value a can be set.

[0101] Step S4, respectively determine the blood vessel improvement response degree of the upstream branch and the downstream branch of the registered target vessel segment after stenosis relief.

[0102] In the classification of the blood vessel segment in the cardiovascular angiogram, the integrity of the feature representation is crucial. Since the vascular system is an interconnected cycle as a whole, if only the features of a single blood vessel segment are extracted, the features may not be comprehensive enough. Therefore, when extracting the features of a single blood vessel segment, in addition to focusing on the features of the blood vessel segment itself, the structural features of the upstream and downstream branches thereof also need to be considered to achieve a comprehensive representation of the functional state of the blood vessel.

[0103] In the angiogram, the upstream branches of the blood vessel segment are the starting points of blood flow, and the state of these branches is directly related to the degree of stenosis of the downstream blood vessel segment. When there is stenosis in the downstream blood vessel segment, the upstream branches may expand due to blood congestion. The degree of stenosis of the blood vessel segment itself can be directly reflected by the difference between the stenosis region and the normal region. The state of the downstream branch is also affected by the stenosis of the upstream blood vessel segment and may appear mild stenosis due to reduced blood perfusion. These three types of features are interconnected: the expansion of the upstream branch is a compensatory response to the stenosis of the downstream blood vessel segment, the stenosis of the blood vessel segment itself directly determines the degree of influence, and the stenosis of the downstream branch is a chain result caused by the upstream stenosis. Extracting features in these three dimensions can more comprehensively represent the stenosis state of the blood vessel.

[0104] Specifically, referring to Figure 5 , the step S4 comprises:

[0105] Step S41, using the difference between the abnormality degree of the upstream branch of the target blood vessel segment and the overall abnormality degree of all upstream branches as a difference ratio weight, combining the abnormality degree of the upstream branch to obtain an abnormality index of the upstream branch; wherein the abnormality degree includes the stenosis degree and the expansion degree;

[0106] Step S42, determining the first abnormality index and the second abnormality index of the upstream branch of the registration target blood vessel segment before and after the contrast agent flows in and the stenosis is relieved, respectively;

[0107] Step S43, using the first abnormality index and the second abnormality index to calculate the blood vessel improvement response degree of the upstream branch of the registration target blood vessel segment.

[0108] In the present embodiment, the stenosis of the blood vessel segment will cause different effects on the upper and lower branches. The upstream branches (branches into which blood flows into the blood vessel segment) will be blocked by the stenosis of the blood vessel segment, and will expand; the downstream branches (branches from which blood flows out of the blood vessel segment) will have reduced blood supply due to the stenosis of the blood vessel segment, and will shrink slightly. Therefore, the effects are quantified in combination with the number of branches and the primary and secondary differences. The expansion or contraction at the branch is directly related to the blood vessel segment in which the branch blood vessel is located (for example, the branch blood vessel at the upper branch essentially belongs to the lower half of the region of the previous blood vessel segment of the current blood vessel segment), and there is not only one branch blood vessel at the branch, and the differences in the diameters of the branch blood vessels need to be superimposed to accurately quantify the abnormality degree (expansion is positive, and stenosis is negative) at the branch.

[0109] For each branch (whether upper or lower), the difference ratio (i.e., the abnormality degree, which is analogous to the stenosis degree of the target blood vessel segment) of the u-th branch blood vessel and the normal branch blood vessel in the main blood vessel (the blood vessel segment to which the branch blood vessel belongs) is calculated: abnormality degree wherein represents the mean value of the diameters of the normal branch blood vessels in the blood vessel segment to which the u-th branch blood vessel belongs, represents the diameter of the abnormal region, which is obtained in the same manner as the stenosis region described above, and the corresponding diameter can be the mean value of the diameters of the multiple sampling points in the abnormal region, or the minimum diameter. Since there is not only one branch blood vessel at the branch, the abnormality degree (expansion or contraction) at the branch needs to be quantified in combination with all the branch blood vessels. The greater the difference between the diameter of the branch blood vessel and the diameter of the normal branch blood vessel in the main blood vessel, the greater the contribution of the branch to the overall abnormality. Therefore, the difference ratio weight is designed as wherein m herein represents the number of branch blood vessels at the branch, represents the overall abnormality degree of all upstream branches or downstream branches at the branch (in the case of being in the denominator, the minimum value a described above can be set to prevent the denominator from being 0), and the abnormality index of the corresponding upstream branch or downstream branch at the branch is

[0110]

[0111] wherein represents the abnormality degree of the upstream branch or the downstream branch, represents the difference ratio weight of the upstream branch or the downstream branch; represents a direction factor, for the upstream branch (the main trend is expansion), the value is +1; for the downstream branch (the main trend is contraction), the value is -1. It needs to be noted again that the direction factor herein ​Uniform refers to the inner diameter of the sampling point of the target vessel segment and the upstream and downstream branches, respectively Refers to the mean of the normal tube diameter of the respective vessel segment (the normal tube diameter of the respective branch vessel of the upstream and downstream branches).

[0112] By extracting the characteristics of the upstream branch of the blood vessel, the blood vessel segment itself and the downstream branch, an evaluation system covering the range of the blood vessel segment is formed. The three mutually confirm the quantitative evaluation of the health status of the blood vessel in a single frame image. Through multi-frame comparative analysis, the dynamic changes of the blood vessel structure over time can be further captured, providing a more scientific basis for clinical decision-making.

[0113] Further quantifying the dynamic characteristics at the upper branch. After the relief of the stenosis of the blood vessel segment, the decrease in the expansion index of the upstream branch due to the relief of blood flow congestion reflects the response degree of the upstream to the improvement of the downstream blood flow, that is, the blood vessel improvement response degree of the upstream , wherein is the expansion index of the upstream branch before the contrast agent flows into the target blood vessel segment (registration), which is denoted as the first abnormal index, is the expansion index of the upstream branch after the relief of the stenosis of the target blood vessel segment, which is denoted as the second abnormal index. The greater the value, the more obvious the improvement of the expansion state of the upstream branch.

[0114] Similarly, the dynamic characteristics at the lower branch are quantified. After the relief of the stenosis of the blood vessel segment, the decrease in the stenosis index of the downstream branch due to the increase in blood flow can reflect the transmission effect of the improvement of blood flow: the blood vessel improvement response degree of the downstream , wherein is the stenosis index of the downstream branch before the contrast agent flows into the target blood vessel segment. It can be denoted as the third abnormal index; is the stenosis index of the downstream branch after the relief of the stenosis of the target blood vessel segment, which can be denoted as the fourth abnormal index. The greater the value, the more obvious the improvement of the ischemic state of the downstream branch.

[0115] In addition, in order to prevent the extreme case that and are 0, a small value a, such as a=0.000001, can be added.

[0116] Step S5, determining the target stenosis index of the registration target blood vessel segment by using the respective blood vessel improvement response degree and the stenosis relief degree, and classifying and warning the registration target blood vessel segment based on the target stenosis index.

[0117] Specifically, the step S5 of determining the target stenosis index of the registration target blood vessel segment by using the respective blood vessel improvement response degree and the stenosis relief degree comprises:

[0118] determining a target duration ratio of the target relief duration of the registered target vessel segment in the stenosis relief process relative to the maximum relief duration of all vessel segments;

[0119] using the target duration ratio, the stenosis relief degree, and the respective vessel improvement response degree of the upstream branch and the downstream branch, a target stenosis index of the registered target vessel segment is calculated.

[0120] In this embodiment, the time length of the stenosis index of the vessel segment maintained in the relief state (target relief duration ) can reflect the persistence of the blood flow impact effect:

[0121]

[0122] In the formula, represents the time point at which the stenosis index of the (registered) target vessel segment recovers to more than 90% of the original value; represents the time point at which the stenosis index reaches the minimum value. The greater the value, the more persistent the expansion effect of blood flow on the stenosis region.

[0123] By fusing the three characteristic change rates of the target vessel segment, its upstream and downstream branches, a comprehensive evaluation index can be constructed, which can more completely depict the change rule of the vessel state over time. Considering the complexity of the vascular network structure, some vessel segments may not have upstream branches (such as the starting end of the aorta) or downstream branches (such as the terminal small arterioles), so the formula should be designed to recognize the absence of branches, and calculate the target stenosis index :

[0124]

[0125] In the formula, represents the basic stenosis index calculated according to the target vessel segment itself: . Wherein, represents the maximum relief duration among all vessel segments, which is used for normalization and is generally not 0. If there is an extreme case of 0, the above minimum value a can be added to prevent the denominator from being 0, represents the target duration ratio. The stenosis relief degree is greater, indicating that the stenosis relief is more obvious, so is positively correlated with the stenosis degree. The target relief duration is greater, indicating that the relief duration is longer, and the stenosis is more likely to be temporary, so is positively correlated with the stenosis degree.

[0126] represents the upstream branch vessel improvement response degree correction term: ;

[0127] denotes the correction term of the downstream branch vessel improvement response degree:

[0128] If there is a branch or The more obvious the branch improvement (the greater the vessel improvement response degree , the smaller the correction term, and the lower the final stenosis index.

[0129] Calculate the target stenosis index of the target vessel segment By fusing the vessel segment itself and the changes in the upstream and downstream branches during the contrast agent flow process, the dynamic stability of the vascular system from blood flow input to output can be comprehensively captured, providing a quantitative tool that integrates multi-dimensional information for vascular health assessment.

[0130] Based on the final target stenosis index Classify the vessel segment. Draw the cumulative distribution curve of each vessel segment , and divide the vessel segment into three categories using the cumulative distribution curve. Take the 75th percentile as the high-risk boundary threshold , i.e. the value of 75% of the samples in the cumulative distribution is less than this value; take the 25th percentile as the low-risk boundary threshold , i.e. the value of 25% of the samples in the cumulative distribution is less than this value.

[0131] When the value is greater than or equal to 0.75, the corresponding vessel segment is marked as a high-risk stenosis and needs to be paid special attention by the doctor; When the value is between 0.5 and 0.75, it is marked as a medium-risk stenosis and needs to be regularly monitored; When the value is less than 0.5, it is marked as a low-risk stenosis and does not require special attention.

[0132] When classifying the vessel segment as the basic unit, since the blood vessels are continuous and circulatory, it is difficult to fully reflect the true state of blood in the vessel segment if only the stenosis of a single vessel is analyzed in isolation. Therefore, the present application not only extracts the stenosis degree of the vessel segment itself, but also analyzes the features of the upstream and downstream branches of the vessel segment, achieving more accurate quantification of vessel stenosis and fine classification of vessel stenosis state, providing more comprehensive data for the classification of the vessel segment, and further facilitating the provision of more scientific data reference for the diagnosis and treatment of vascular diseases.

[0133] Example Two:

[0134] ​The embodiment of the present application also provides a cardiovascular imaging image intelligent classification device. The device can be a computer, a server or a combination of multiple data analysis and calculation devices.

[0135] As shown in Figure 6 , Figure 6 is a structural schematic diagram of a hardware running environment of a cardiovascular imaging image intelligent classification device related to the embodiment of the present application.

[0136] As shown in Figure 6 , the cardiovascular imaging image intelligent classification device can include a processor 1001 such as a CPU, a network interface 1004, a user interface 1003, a memory 1005 and a communication bus 1002. The communication bus 1002 is used to realize the connection and communication among the components. The user interface 1003 can include a display, an input unit such as a control panel, and the optional user interface 1003 can also include a standard wired interface and a wireless interface. The network interface 1004 can optionally include a standard wired interface and a wireless interface (such as a WIFI interface). The memory 1005 can be a high-speed RAM memory or a stable memory (non-volatile memory) such as a disk memory. The memory 1005 can also be an independent storage device from the aforementioned processor 1001. The memory 1005 as a computer storage medium can include a cardiovascular imaging image intelligent classification program.

[0137] Those skilled in the art can understand that Figure 6 the hardware structure shown in the foregoing embodiments does not constitute a limitation on the device, and can include more or fewer components than those shown, or combine certain components, or different component arrangements.

[0138] Continuing to refer to Figure 6 , Figure 6 the memory 1005 as a computer readable storage medium can include an operating system, a user interface module, a network communication module and a cardiovascular imaging image intelligent classification program.

[0139] In Figure 7 , the network communication module is mainly used to connect a server and can communicate data with the server; and the processor 1001 can call the cardiovascular imaging image intelligent classification program stored in the memory 1005 and execute the steps in the foregoing embodiments.

[0140] Based on the hardware structure of the cardiovascular imaging image intelligent classification device, various embodiments of the cardiovascular imaging image intelligent classification method of the present application are realized.

[0141] In addition, the present application also provides a cardiovascular imaging image intelligent classification system, please refer to​ The cardiovascular angiography image intelligent classification system comprises:

[0142] A blood vessel region division module A10 is configured to determine a blood vessel segment corresponding to each branch by using each branch point after the blood vessel skeletonization of the cardiovascular angiography image; determine a stenosis region and a stenosis degree thereof in the target blood vessel segment by using a blood vessel inner diameter deviation in the target blood vessel segment; and determine an initial abnormality index of the target blood vessel segment by using each stenosis degree.

[0143] A blood vessel change analysis module A20 is configured to determine a registration target blood vessel segment by using the blood vessel skeleton to perform feature registration on a plurality of cardiovascular angiography images; determine a stenosis relief degree of the registration target blood vessel segment by using a change in the initial abnormality index of the registration target blood vessel segment before and after the inflow of the contrast agent; and determine a blood vessel improvement response degree of each of an upstream branch and a downstream branch of the registration target blood vessel segment after the stenosis relief of the registration target blood vessel segment, respectively.

[0144] A blood vessel classification and early warning module A30 is configured to determine a target stenosis index of the registration target blood vessel segment by using the blood vessel improvement response degree and the stenosis relief degree; and perform classification and early warning on the registration target blood vessel segment based on the target stenosis index.

[0145] Further, the blood vessel region division module A10 is further configured to:

[0146] determine each skeleton node after the blood vessel skeletonization of the cardiovascular angiography image; and determine a cosine similarity of an included angle between a front blood vessel direction vector and a rear blood vessel direction vector of the skeleton node.

[0147] compare the cosine similarity with a preset similarity threshold value; determine each branch point in the skeleton node; and obtain the corresponding blood vessel segment by taking the branch point as a boundary of each branch.

[0148] Further, the blood vessel region division module A10 is further configured to:

[0149] determine a blood vessel inner diameter at a target sampling point in the target blood vessel segment, an average inner diameter of a window in which the target sampling point is located, and a blood vessel inner diameter deviation of the blood vessel inner diameter relative to the average inner diameter.

[0150] compare the blood vessel inner diameter deviation with a preset deviation threshold value; and take the target sampling point greater than the preset deviation threshold value as a stenosis sampling point.

[0151] start traversing from the stenosis sampling point to both ends along a blood vessel axis until the blood vessel inner diameter deviations of both ends are less than or equal to the preset deviation threshold value, to stop the traversal, and obtain the stenosis region and the stenosis degree thereof.

[0152] determine a difference ratio weight between a stenosis length of the stenosis region and a total blood vessel length of the target blood vessel segment; and determine the initial abnormality index of the target blood vessel segment by using the difference ratio weight and the stenosis degree.

[0153] Further, the blood vessel region division module A10 is further configured to:

[0154] determine a normal lumen diameter mean value of the normal lumen diameter of the target blood vessel segment, wherein the minimum lumen diameter in the stenosis region is taken as a target lumen diameter of the stenosis region;

[0155] calculate the stenosis degree of the stenosis region by using the target lumen diameter and the normal lumen diameter mean value.

[0156] Further, the blood vessel change analysis module A20 is further configured to:

[0157] take the angiogram when the blood vessel is not filled with the contrast agent as a reference frame, and take the subsequent angiogram when the blood vessel is filled with the contrast agent as a subsequent frame;

[0158] extract a plurality of matched feature point pairs from the blood vessel skeletons of the reference frame and the subsequent frame, and obtain the registration target blood vessel segment after the rigid registration by minimizing the spatial distance error of the feature point pairs.

[0159] Further, the blood vessel change analysis module A20 is further configured to:

[0160] sample each skeleton sampling point along the blood vessel skeletons of the reference frame and the subsequent frame at a preset distance to obtain a coordinate difference between the subsequent frame and the reference frame of each skeleton sampling point;

[0161] obtain the global deformation field by solving the coordinate difference by using a thin plate spline interpolation function, and obtain the registration target blood vessel segment after the elastic registration by correcting the pixel position of the subsequent frame by using the global deformation field.

[0162] Further, the blood vessel change analysis module A20 is further configured to:

[0163] determine an initial abnormality index of the registration target blood vessel segment before the contrast agent flows in and a minimum abnormality index of the registration target blood vessel segment after the contrast agent flows in;

[0164] obtain the stenosis relief degree of the registration target blood vessel segment by using the difference between the initial abnormality index before the contrast agent flows in and the minimum abnormality index after the contrast agent flows in.

[0165] Further, the blood vessel change analysis module A20 is further configured to:

[0166] obtain the abnormality index of the upstream branch of the target blood vessel segment by using the difference between the abnormality degree of the upstream branch and the overall abnormality degree of all upstream branches as a difference ratio weight, and combining the abnormality degree of the upstream branch; wherein the abnormality degree includes the stenosis degree and the dilation degree;

[0167] determine a first abnormality index and a second abnormality index of the upstream branch of the target blood vessel segment before the contrast agent flows in and after the stenosis is relieved, respectively;

[0168] The first abnormality index and the second abnormality index are used to calculate a blood vessel improvement response degree of the upstream branch of the target blood vessel segment.

[0169] Further, the blood vessel classification and early warning module A30 is further configured to:

[0170] determine a target time length ratio of the target blood vessel segment relative to a maximum relief time length of all blood vessel segments during the target relief time length in the stenosis relief process;

[0171] The target time length ratio, the stenosis relief degree, and the blood vessel improvement response degree of the upstream branch and the downstream branch are used to calculate a target stenosis index of the target blood vessel segment.

[0172] The specific implementation of the cardiovascular angiography image intelligent classification system is basically the same as that of the cardiovascular angiography image intelligent classification method, and will not be repeated here.

[0173] In addition, the present application also provides a computer readable storage medium. The computer readable storage medium of the present application stores a cardiovascular angiography image intelligent classification program, wherein the cardiovascular angiography image intelligent classification program is executed by a processor to realize the steps of the cardiovascular angiography image intelligent classification method as described above.

[0174] The method realized by the cardiovascular angiography image intelligent classification program when executed can refer to each embodiment of the cardiovascular angiography image intelligent classification method of the present application, which will not be repeated here.

[0175] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0176] Each embodiment in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0177] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, a system or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer-readable program code thereon for use by or in connection with an instruction execution system. For the purposes of this description, a computer-usable or computer readable storage medium can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

[0178] The above descriptions are only the preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any equivalent structure / method transformation or direct / indirect application in other related technical fields made according to the content of the present application specification and drawings are included in the protection scope of the present application.

Claims

1. A method of intelligent classification of cardiovascular angiogram images, characterized in that, The method comprises the following steps: Each branch point after the blood vessel skeletonization of the cardiovascular angiogram is used to determine the corresponding blood vessel segment of each branch; The blood vessel diameter deviation in the target blood vessel segment is used to determine the stenosis region and the stenosis degree of the stenosis region, and the initial abnormality index of the target blood vessel segment is determined by using each stenosis degree; The stenosis relief degree of the registration target blood vessel segment is determined by using the change of the initial abnormality index of the registration target blood vessel segment before and after the inflow of the contrast agent; The blood vessel improvement response degree of the upstream branch and the downstream branch of the registration target blood vessel segment after the stenosis relief is respectively determined; The target stenosis index of the registration target blood vessel segment is determined by using the respective blood vessel improvement response degree and the stenosis relief degree, and the registration target blood vessel segment is classified and warned based on the target stenosis index.

2. The cardiovascular angiogram image intelligent classification method of claim 1, wherein, The each branch point after the blood vessel skeletonization of the cardiovascular angiogram is used to determine the corresponding blood vessel segment of each branch, comprising: Each skeleton node after the blood vessel skeletonization of the cardiovascular angiogram is determined, and the cosine similarity between the front blood vessel direction vector and the rear blood vessel direction vector of the skeleton node is determined; The cosine similarity and the preset similarity threshold are compared, the each branch point in the skeleton node is determined, and the corresponding blood vessel segment is obtained by taking the branch point as the boundary of each branch.

3. The method of claim 1, wherein, The blood vessel diameter deviation in the target blood vessel segment is used to determine the stenosis region and the stenosis degree of the stenosis region, and the initial abnormality index of the target blood vessel segment is determined by using each stenosis degree, comprising: The blood vessel diameter and the average diameter of the window at the target sampling point in the target blood vessel segment are determined, and the blood vessel diameter deviation of the blood vessel diameter relative to the average diameter is determined; The blood vessel diameter deviation is compared with the preset deviation threshold, and the target sampling point greater than the preset deviation threshold is taken as the stenosis sampling point; The stenosis region and the stenosis degree are obtained by traversing from the stenosis sampling point to both ends along the blood vessel axis until the blood vessel diameter deviation of both ends is less than or equal to the preset deviation threshold; The difference ratio weight between the stenosis length of the stenosis region and the total blood vessel length of the target blood vessel segment is determined, and the initial abnormality index of the target blood vessel segment is determined by using the difference ratio weight and the stenosis degree.

4. The cardiovascular angiogram image intelligent classification method of claim 3, wherein, The stenosis region and the stenosis degree are obtained, comprising: The minimum blood vessel diameter in the stenosis region is taken as the target pipe diameter of the stenosis region, and the normal pipe diameter mean of the normal blood vessel diameter of the target blood vessel segment is determined; The target pipe diameter and the normal pipe diameter mean are used to calculate the stenosis degree of the stenosis region.

5. The method of claim 1, wherein, The registration target blood vessel segment is obtained by performing feature registration on the multiple cardiovascular angiograms by using the blood vessel skeleton, comprising: The cardiovascular angiogram when the blood vessel is not filled with the contrast agent is taken as the reference frame, and the subsequent cardiovascular angiogram when the blood vessel is filled with the contrast agent is taken as the subsequent frame; A plurality of matched feature point pairs are extracted from the blood vessel skeletons of the reference frame and the subsequent frame, and the feature point pairs are subjected to spatial distance error minimization to obtain the registration target blood vessel segment after rigid registration.

6. The method of intelligent classification of angiogram images according to claim 5, wherein, The registration target blood vessel segment after rigid registration is obtained by performing spatial distance error minimization on the feature point pairs, and then further comprising: The vessel skeleton along the reference frame and the subsequent frame is sampled at a preset distance to obtain a plurality of skeleton sampling points, and a coordinate difference between each skeleton sampling point and the reference frame is determined. A global deformation field is obtained by using a thin plate spline difference function to solve the coordinate difference, and a pixel position of the subsequent frame is corrected by using the global deformation field to obtain an elastically registered registration target vessel segment.

7. The method of claim 1, wherein, The method for determining the stenosis relief degree of the registration target vessel segment by using the initial abnormality index change of the registration target vessel segment before and after the contrast agent flows in, comprises: The initial abnormality index before the contrast agent flows in and the minimum abnormality index after the contrast agent flows in of the registration target vessel segment are determined respectively. The stenosis relief degree of the registration target vessel segment is calculated by using the change difference between the initial abnormality index before the contrast agent flows in and the minimum abnormality index after the contrast agent flows in.

8. The method of claim 1, wherein, The method for determining the respective vessel improvement response degrees of the upstream branch and the downstream branch of the registration target vessel segment after the stenosis is relieved, comprises: An abnormality index of the upstream branch of the registration target vessel segment is obtained by using a difference ratio weight of the abnormality degree of the upstream branch compared with the overall abnormality degree of all upstream branches, and combining the abnormality degree of the upstream branch; wherein the abnormality degree comprises a stenosis degree and a dilatation degree. The first abnormality index and the second abnormality index of the upstream branch of the registration target vessel segment before the contrast agent flows in and after the stenosis is relieved are determined respectively. The vessel improvement response degree of the upstream branch of the registration target vessel segment is calculated by using the first abnormality index and the second abnormality index.

9. The method of claim 1, wherein, The method for determining the target stenosis index of the registration target vessel segment by using the respective vessel improvement response degrees and the stenosis relief degree, comprises: A target time length ratio of a target relief time length of the registration target vessel segment maintained in the stenosis relief process relative to a maximum relief time length of all vessel segments is determined. The target stenosis index of the registration target vessel segment is calculated by using the target time length ratio, the stenosis relief degree, and the respective vessel improvement response degrees of the upstream branch and the downstream branch.

10. A cardiovascular angiogram image intelligent classification system, the system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, The processor executes the computer program to realize the steps of the method of any one of claims 1-9.

Images