Coronary artery inflammation plaque recognition method based on multi-sequence CTA images
By using multi-sequence CTA imaging technology, combined with feature point matching and local thickness gradient correction, the contrast agent retention index and pericoronary fat inflammation index were analyzed, solving the problem that CTA scans could not distinguish whether coronary artery plaques were inflamed. This enabled non-invasive, quantifiable identification and early warning of inflammatory plaques.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- FUWAI HUAZHONG CARDIOVASCULAR HOSPITAL
- Filing Date
- 2026-04-29
- Publication Date
- 2026-07-03
AI Technical Summary
Current CTA scanning technology has difficulty in accurately distinguishing whether coronary artery plaques are inflamed, which makes it impossible to accurately guide clinical treatment and increases the risk of acute myocardial infarction.
By acquiring three-phase CTA images of the coronary arteries (plain, arterial, and delayed phase images), feature point matching and local thickness gradient correction were performed to eliminate motion artifacts. The contrast agent retention index and the inflammation index of the pericoronary adipose tissue were analyzed to identify inflammatory plaques.
It enables non-invasive and quantifiable identification of coronary artery inflammatory plaques, reduces the risk of acute myocardial infarction, guides stable plaques to avoid overtreatment, and timely identifies and intervenes in high-risk inflammatory plaques.
Smart Images
Figure CN122336435A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical imaging technology, specifically to a method for identifying coronary inflammatory plaques based on multi-sequence CTA images. Background Technology
[0002] Coronary inflammatory plaques are high-risk, vulnerable plaques formed on the inner wall of coronary arteries due to lipid deposition and inflammatory cell infiltration, representing an active stage of atherosclerosis. The core characteristics of coronary inflammatory plaques are a thin, fragile fibrous cap, a large lipid core, and an abundance of inflammatory cells such as macrophages and lymphocytes, often accompanied by neovascularization and microbleeds. Coronary inflammatory plaques are highly prone to rupture, triggering platelet aggregation and thrombus formation. When these thrombi block the coronary arteries, they can cause acute myocardial infarction. Furthermore, they accelerate coronary artery stenosis, affecting blood supply to the heart and acting as a major trigger for acute ischemic events, which can be life-threatening in severe cases.
[0003] However, not all coronary artery plaques are inflammatory plaques. Coronary artery plaques include non-inflammatory stable plaques and inflammatory plaques. Non-inflammatory stable plaques have a dense structure and are not easily ruptured; routine interventions are sufficient to slow their progression. Inflamed and non-inflammatory arterial plaques pose vastly different risks, require different treatment plans, and have drastically different prognoses. Therefore, distinguishing the inflammatory state of coronary artery plaques can accurately guide clinical treatment, avoid overtreatment of stable plaques, and enable timely and targeted interventions for inflammatory plaques, reducing the risk of acute events. Differentiating whether a plaque is inflamed is of great significance.
[0004] Computed Tomography Angiography (CTA) is currently the mainstream tool for screening coronary artery plaques. The routine CTA scan for screening coronary artery plaques typically involves a single, single-phase CTA scan of the coronary artery region to obtain static images. This captures the contrast agent's enhancement status within the coronary artery and plaque at a specific time point, clearly showing the external morphological characteristics of the plaque, but it cannot directly distinguish whether it is inflamed, an intrinsic indicator. Summary of the Invention
[0005] To address the above-mentioned technical problems, the present invention aims to provide a method for identifying coronary inflammatory plaques based on multi-sequence CTA images.
[0006] The present invention provides a method for identifying coronary inflammatory plaques based on multi-sequence CTA images, the specific technical solution of which is as follows: Acquire three-phase CTA images of the coronary arteries, including plain scan images, arterial scan images, and delayed scan images; Feature point matching and local thickness gradient correction are performed on the three phase CTA images to obtain the registered three phase CTA images; Based on the arterial phase image, plaque identification is performed, candidate plaques on the arterial phase image are marked, and the candidate plaques are mapped to the registered CTA three-phase image to obtain the corresponding candidate plaques on the plain scan image and the delayed scan image. Based on the registered CTA three-phase images, the changes in CT values within the candidate plaques over time are analyzed to obtain the contrast agent retention index and screen for suspected inflammatory plaques. For the suspected inflammatory plaque, based on the arterial phase image and the registered delayed phase image, the enhancement amplitude and attenuation gradient of the CT value of the adipose tissue region surrounding the plaque are analyzed to obtain the pericoronal fat inflammation index and confirm the inflammatory plaque.
[0007] In some embodiments of the present invention, feature point matching is performed on the CTA three-phase images, including: For the three phases of CTA images, the coronary artery centerline is identified, and the bifurcation points and curvature maxima on the coronary artery centerline are extracted to form feature point sets corresponding to the plain scan image, arterial scan image, and delayed scan image. Using the arterial phase image as a reference, the feature point set of the arterial phase image is matched with the feature point sets corresponding to the plain scan image and the delayed phase image, respectively, to obtain the CTA three-phase image after feature point matching. The CTA three-phase image after feature point matching includes the arterial phase image, the feature point matched plain scan image, and the feature point matched delayed phase image.
[0008] In some embodiments of the present invention, using the arterial phase image as a reference, the feature point set of the arterial phase image is type-matched with the feature point sets corresponding to the plain scan image and the delayed phase image, respectively, to obtain a CTA three-phase image after feature point matching, including: Using the arterial phase image as a reference, the feature point set of the arterial phase image is matched with the feature point sets corresponding to the plain scan image and the delayed phase image respectively through nearest neighbor search. False matches with a distance greater than a preset pixel are eliminated to obtain the effective matching pairs corresponding to the feature point sets of the arterial phase image and the feature point sets corresponding to the plain scan image and the delayed phase image respectively. Based on the effective matching pairs, the rigid transformation parameters are solved using the least squares method to minimize the sum of squared errors of the matching pairs. The transformation parameters are then applied to the flat scan image and the delayed image, and the feature point matching flat scan image and the feature point matching delayed image are obtained by interpolation.
[0009] In some embodiments of the present invention, local thickness gradient correction is performed on the CTA three-phase images, including: Based on the CTA three-phase images after feature point matching, the differences between the thickness gradients of the blood vessel wall at the same sampling location are analyzed to identify abnormal thickness gradient regions. Based on the CTA phase III images after feature point matching, the thickness gradient anomaly region is locally translated and corrected to obtain the registered CTA phase III images.
[0010] In some embodiments of the present invention, based on the CTA three-phase images after feature point matching, the differences in thickness gradients between the vessel walls corresponding to the same sampling location are analyzed to identify regions with abnormal thickness gradients, including: For the CTA phase III images after feature point matching, the vessel wall thickness corresponding to the same sampling position is obtained, and the difference of the vessel wall thickness at adjacent sampling positions is calculated to obtain the vessel wall thickness gradient at each sampling position on the CTA phase III images after feature point matching. Using the arterial phase image as a reference, calculate the absolute thickness gradient deviation between the plain scan image and the delayed scan image after feature point matching at the same sampling location and the arterial phase image, respectively. Set a preset absolute thickness gradient deviation threshold, mark thickness gradient abnormal points, and identify thickness gradient abnormal regions.
[0011] In some embodiments of the present invention, based on the CTA three-phase image after feature point matching, local translation correction is performed on the thickness gradient anomaly region to obtain the registered CTA three-phase image, including: For the CTA three-phase images after feature point matching, a local window is divided with the thickness gradient anomaly region as the center, and the local window in the flat scan image and the delayed image after feature point matching is translated multiple times with different displacements. Get the updated thickness gradient corresponding to each thickness gradient anomaly point in the translated local window, get the gray value of each pixel in the translated local window, and get the gray value of each pixel in the corresponding local window of the arterial phase image. The difference between each updated thickness gradient and the thickness gradient of the corresponding thickness gradient anomaly point in the arterial phase image is analyzed, and the grayscale difference of each pixel in the local window after translation and the local window corresponding to the arterial phase image is analyzed to obtain the comprehensive score corresponding to each displacement of each local window; Based on the comprehensive score, the optimal displacement is determined to obtain the registered CTA three-phase images, which include the arterial phase image, the registered plain scan image, and the registered delayed scan image.
[0012] In some embodiments of the present invention, plaque identification is performed based on the arterial phase image, and candidate plaques on the arterial phase image are marked, including: In the arterial phase image, along the coronary artery centerline, multiple consecutive sampling points are examined as a unit segment. Segments with a wall thickness exceeding a preset wall thickness and a CT value less than a preset CT value are marked as candidate plaques, and a binary mask is generated for each candidate plaque.
[0013] In some embodiments of the present invention, based on the registered CTA three-phase images, the changes in CT values within the candidate plaques over time are analyzed to obtain the contrast agent retention index, including: For the registered CTA three-phase images, the CT values of the candidate plaques are extracted; The difference in CT values between the candidate plaques in the registration delay image and the registration plain scan image is analyzed to obtain the contrast agent retention amount of the candidate plaques; The difference in CT values between the candidate plaques in the arterial phase image and the registered plain scan image is analyzed to obtain the contrast agent filling amount of the candidate plaques; The contrast agent retention index is obtained based on the contrast agent retention amount and the contrast agent filling amount.
[0014] In some embodiments of the present invention, for the suspected inflammatory plaque, based on the arterial phase image and the registered delayed phase image, the enhancement magnitude of the CT value of the adipose tissue region surrounding the plaque is analyzed, including: Using the suspected inflammatory plaque as the center, a search is performed along the extravascular region, interfering tissues are eliminated, and a non-vascular ring region outside the blood vessel is obtained as the peripheral candidate adipose tissue region of the suspected inflammatory plaque. On the arterial phase image, calculate the first average CT value of all pixels within the peripheral candidate adipose tissue region; On the registered delayed image, calculate the second average CT value of all pixels in the peripheral candidate adipose tissue region; The enhancement magnitude of the CT value in the peripheral candidate adipose tissue region of the suspected inflammatory plaque is obtained based on the first average CT value and the second average CT value.
[0015] In some embodiments of the present invention, analyzing the attenuation gradient of CT values in the adipose tissue region surrounding the plaque includes: The peripheral candidate adipose tissue region is divided into a proximal layer and a distal layer according to its distance from the outer wall of the blood vessel; On the registered delayed image, the average CT value of all pixels in the proximal layer and the distal layer is calculated respectively to obtain the attenuation gradient of the CT value of the peripheral candidate adipose tissue region of the suspected inflammatory plaque.
[0016] Compared with existing technologies, the coronary artery inflammatory plaque identification method based on multi-sequence CTA images provided by this invention has the following beneficial effects: This invention acquires three phase CTA images of the coronary artery (plain scan image, arterial scan image, and delayed scan image), performs feature point matching and local thickness gradient correction on the three phase CTA images, and obtains registered three phase CTA images, eliminating motion artifacts between multiple phase images and ensuring accurate correspondence of plaque regions in the time series.
[0017] This invention identifies candidate plaques based on arterial phase images. Using registered CTA three-phase images, it analyzes the changes in CT values within the candidate plaques over time to obtain the contrast agent retention index and screen for suspected inflammatory plaques. For the suspected inflammatory plaques, based on the arterial phase images and the registered delayed-phase images, it analyzes the enhancement amplitude and attenuation gradient of CT values in the surrounding adipose tissue region to obtain the pericoronary fat inflammation index, confirming the inflammatory plaque. In other words, this invention identifies inflammatory plaques with microvascular retention characteristics by constructing a contrast agent retention index and a pericoronary fat inflammation index, and achieves dual quantitative verification of the spread of inflammation from microvascular leakage within the plaque to the surrounding adipose tissue, significantly improving identification accuracy and enabling early non-invasive warning of coronary artery inflammation.
[0018] This invention provides a non-invasive, quantifiable inflammatory plaque screening tool that helps guide stable plaques to avoid overtreatment, while also enabling timely identification and intervention of high-risk inflammatory plaques to reduce the risk of acute myocardial infarction. Attached Figure Description
[0019] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 This is a schematic diagram of the basic process of a method for identifying coronary inflammatory plaques based on multi-sequence CTA images, provided in one embodiment of the present invention. Figure 2 This is a schematic diagram of a coronary artery inflammatory plaque CTA during the arterial phase, provided as an embodiment of the present invention. Detailed Implementation
[0021] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the method for identifying coronary inflammatory plaques based on multi-sequence CTA images proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0022] 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 this invention pertains. Terms such as “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a circuit structure, article, or device comprising a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the article or device that includes that element. Relational terms such as “first” and “second” are used merely to distinguish one entity or operation from another and do not necessarily require or imply any such actual relationship or order between these entities or operations.
[0023] The following description, in conjunction with the accompanying drawings, details the specific scheme of a method for identifying coronary inflammatory plaques based on multi-sequence CTA images provided by the present invention.
[0024] Please see Figure 1 This illustrates the basic flowchart of a method for identifying coronary inflammatory plaques based on multi-sequence CTA images provided by an embodiment of the present invention.
[0025] like Figure 1 As shown, an embodiment of the present invention provides a method for identifying coronary inflammatory plaques based on multi-sequence CTA images, specifically including: S100: Acquire three-phase CTA images of the coronary arteries, including plain phase, arterial phase, and delayed phase images.
[0026] Acquire three-phase CTA images of the coronary arteries, including plain phase images, arterial phase images, and delayed phase images. The specific implementation method is as follows: Before injecting the contrast agent, a multi-slice spiral CT scanner is used to perform an initial spiral scan of the target coronary artery region (the coronary artery region identified as the target or where the lesion is located) to obtain a baseline density image as a plain scan image of the coronary artery region.
[0027] Contrast agent was injected via the antecubital vein, and a multi-slice spiral CT scanner was used. The scanning range was the same as that of the plain scan, and the scan was triggered by bolus-tracking technology. The images were acquired during mid-diastole to capture the peak of the contrast agent in the vascular lumen, clearly displaying the anatomy of the coronary arteries and obtaining arterial phase images of the coronary artery region.
[0028] After the arterial phase image scan, wait 3-5 minutes, then use a multi-slice spiral CT scanner to scan the same area again to capture characteristic images of contrast agent retention in the microvessels of the plaque and obtain delayed phase images of the coronary artery region.
[0029] In addition, the same tube voltage and reconstruction algorithm must be used for the three phases of CTA scans to ensure that the CT values in the three phases of CTA images (plain scan, arterial scan, and delayed scan) are comparable.
[0030] To facilitate unified analysis, the coronary artery centerline was first extracted using the minimum path method. Then, the three-dimensional longitudinal cross-sectional images generated by straightening the three-dimensional vessels along the coronary artery centerline using CPR technology from the three-phase CTA images were used as the analysis object to obtain the coronary section images of the plain scan phase. Coronal section images during the arterial phase and coronal section images of the delayed period It should be noted that, unless otherwise specified, the plain scan images, arterial phase images, and delayed phase images mentioned in the following descriptions are all assumed to be coronal section images from the plain scan phase. Coronal section images during the arterial phase and coronal section images of the delayed period .
[0031] At this point, three-phase CTA images of the coronary arteries were obtained, including plain scan images of the coronary artery region. Arterial phase images and delayed images .
[0032] In existing technologies, conventional CTA scans for screening coronary artery plaques typically involve a single, single-phase CTA scan of the coronary artery region to obtain static images, capturing the contrast agent's enhancement status within the coronary artery and plaque at a specific time point. Inflammatory coronary artery plaques contain extensive inflammatory cell infiltration, neovascularization, and microbleeds. These pathological structures lead to microvascular retention of the contrast agent within the plaque; that is, the speed at which the contrast agent enters the plaque, the duration of its residence, and the pattern of its regression will differ significantly from those of stable plaques without inflammation. This difference is difficult to accurately quantify using only a single-phase CTA scan. Therefore, this single-phase CTA scan is not conducive to identifying inflammatory coronary artery plaques.
[0033] In summary, this invention combines multi-sequence CTA imaging to quantify the retention characteristics of contrast agents within plaques. For the same coronary artery region, multiple CTA scans are performed at different time stages after contrast agent injection, comprehensively capturing the time-density changes of the contrast agent within the coronary artery plaque. By quantifying the retention characteristics of the contrast agent after entering the plaque and the density changes of the surrounding tissue, inflammatory plaques can be identified through multi-phase CTA. Therefore, the specific implementation mainly includes steps S200 to S500.
[0034] S200: Perform feature point matching and local thickness gradient correction on the CTA phase III images to obtain the registered CTA phase III images.
[0035] To ensure accurate alignment of CTA images across three phases, it is essential to align the images before analyzing the time-density changes of contrast agent within coronary plaques. Since the plain, arterial, and delayed phase images are acquired at different time points, changes in the patient's heart rate, respiration, and slight positional movements can cause non-rigid spatial misalignment between the three phases, potentially leading to errors during image alignment. Therefore, to eliminate motion artifacts and registration errors during acquisition and ensure that all subsequent analyses are based on strictly aligned images, the acquired CTA images across three phases must first be aligned.
[0036] Based on the above analysis, in the embodiments of the present invention, the registered CTA three-phase images are obtained by performing feature point matching and local thickness gradient correction on the CTA three-phase images.
[0037] Of the three phases in a CTA scan, the arterial phase is when the contrast agent concentration in the coronary artery reaches its peak. At this time, the vessel lumen is fully filled, the contrast between the vessel wall and surrounding tissues is highest, and the boundaries of plaques and the contours of the inner and outer walls of the vessel are most clearly distinguishable. Therefore, the arterial phase image is selected as the comparison benchmark, and the plain scan and delayed scan images are initially aligned with the arterial phase image.
[0038] During the initial alignment process, the bifurcation points of blood vessels are the core nodes of the coronary artery tree topology. Their positions are relatively fixed in CTA images of different phases and will not change due to changes in contrast agent concentration or slight movement. In addition, the point of maximum curvature corresponds to the location where the blood vessel bends most significantly, which is also an inherent feature of blood vessel morphology. Therefore, the bifurcation points and points of maximum curvature of the blood vessel centerline are used as feature point sets for initial alignment.
[0039] Based on the above analysis, in some embodiments of the present invention, feature point matching is performed on CTA three-phase images. Further methods include: First, for the three phases of CTA images, the coronary artery centerline is identified, and the bifurcation points and curvature maxima on the coronary artery centerline are extracted to form feature point sets for the plain scan, arterial phase, and delayed phase images. Specifically, on the initial arterial phase image (the non-coronary arterial phase image without CPR processing), the coronary artery centerline is extracted using the minimum path method. Identify the center line of blood vessels The bifurcation points and curvature maxima on the surface constitute the feature point set of the arterial phase image. Similarly, the vessel centerline was extracted from the initial plain scan image and the initial delayed scan image, respectively. and Identify the center line of blood vessels and The bifurcation points and curvature maxima on the image form the feature point sets corresponding to the plain scan image and the delayed scan image, respectively. and .
[0040] Then, using the arterial phase image as a reference, the feature point set of the arterial phase image is type-matched with the corresponding feature point sets of the plain scan image and the delayed scan image, respectively, to obtain the feature point-matched CTA three-phase images. These three-phase CTA images include the arterial phase image, the feature point-matched plain scan image, and the feature point-matched delayed scan image. Specifically, the arterial phase image is used as the reference, i.e., the feature point set of the arterial phase image is used as the reference. Based on this, the feature point set of the arterial phase image is... Feature point sets corresponding to the plain scan image and the delayed scan image, respectively and Matching pairs are established through nearest neighbor search, and false matches with a distance greater than a preset pixel (20 pixels) are eliminated to obtain the feature point set of the arterial phase image. Feature point sets corresponding to the plain scan image and the delayed scan image, respectively and The corresponding valid matching pairs; based on the valid matching pairs, the least squares method is used to solve for the rigid transformation parameters, minimizing the sum of squared errors of the matching pairs, and the transformation parameters are then applied to the images during the flat scan period. and delayed images The feature point matching flat scan image was obtained by interpolation. Matching feature points with delayed images This yields three phases of CTA images after feature point matching, including the arterial phase image. Feature point matching during the plain scan image Matching feature points with delayed images .
[0041] After completing rigid registration based on feature points, the CTA phase III images have achieved approximate alignment. However, since rigid transformation can only solve overall translation and rotation, it cannot correct local nonlinear deformations. These deformations usually originate from heartbeats, respiratory movements, or subtle changes in the patient's body position, which may result in subpixel-level local misalignment in the plaque area.
[0042] Therefore, after performing feature point matching on the CTA phase III images and completing rigid alignment, it is necessary to further fine-tune the local non-rigid misalignment caused by respiratory factors.
[0043] During the alignment of images across different time phases, the absolute value of vessel wall thickness is a relatively stable anatomical parameter, and alignment can be achieved through the consistency of vessel wall thickness. However, in images from different time phases, the absolute value of thickness may exhibit certain measurement fluctuations due to the influence of contrast agent concentration, image noise, and partial volume effects. Meanwhile, the thickness gradient can reflect the local trend of thickness change, and this local trend is extremely sensitive to minute misalignments. Even if the absolute values of thickness are not significantly different, as long as the misalignment causes a change in the trend of thickness increase or decrease, the gradient will rapidly deviate from the normal value.
[0044] Based on the above analysis, in the embodiments of the present invention, local thickness gradient correction is performed on the CTA three-phase images to achieve fine-tuning of local non-rigid misalignments caused by respiratory factors. Further aspects include: First, based on the CTA images from three phases after feature point matching, the differences in thickness gradients between the vessel walls at the same sampling location are analyzed to identify regions with abnormal thickness gradients. This involves quantifying changes in thickness gradients to identify local non-rigid misalignments in the registration process. Further steps include: First, for the CTA three-phase images after feature point matching, the vessel wall thickness corresponding to the same sampling location is obtained, and the difference in vessel wall thickness at adjacent sampling locations is calculated to obtain the vessel wall thickness gradient at each sampling location in the CTA three-phase images after feature point matching. Specifically, in the arterial phase image... The Unet model was used to segment the vascular region and the inner and outer walls of the blood vessels, and arterial phase images were obtained. A vessel wall segmentation mask was created; sampling positions were evenly set along the centerline of the coronary artery, with an interval of 0.5 mm between sampling positions. At this location, along the direction perpendicular to the centerline of the blood vessel, the distance between the inner and outer walls of the blood vessel is measured to obtain the arterial phase image. Upsampling position The corresponding blood vessel wall thickness is denoted as Arterial phase images Using the vessel wall segmentation mask as a reference, it is mapped to feature point matching plain scan images. Matching feature points with delayed images This serves as the starting point for thickness measurement, allowing for the acquisition of feature point matching images during the flat scan period. Matching feature points with delayed images Upsampling position The corresponding blood vessel wall thickness is denoted as Arterial phase images Above, calculate adjacent sampling positions. and sampling location Corresponding blood vessel wall thickness and The difference between them is used as the arterial phase image. Upsampling position The thickness gradient of the blood vessel wall is denoted as . ,Right now ,in , This indicates the number of sampling locations; similarly, feature point matching images during the flat scan period are obtained. Matching feature points with delayed images Upsampling position The corresponding thickness gradient of the blood vessel wall is denoted as . and This thickness gradient reflects the local rate of change in vessel wall thickness along the vessel's extension direction.
[0045] Second, using the arterial phase image as a benchmark, the absolute thickness gradient deviation between the plain and delayed phase images (after feature point matching at the same sampling location) and the arterial phase image is calculated. A preset absolute thickness gradient deviation threshold is used to mark thickness gradient anomalies and identify abnormal thickness gradient regions. Specifically, at the sampling location... At this point, feature point matching is calculated for the plain scan image. Corresponding thickness gradient of blood vessel walls With arterial phase images Corresponding thickness gradient of blood vessel walls The absolute deviation between them is ; Calculate feature point matching delay period image Corresponding thickness gradient of blood vessel walls With arterial phase images Corresponding thickness gradient of blood vessel walls The absolute deviation between them is Preset thickness gradient absolute deviation threshold (e.g., 0.2 pixels), if or Then the sampling position These are marked as gradient anomaly points. Adjacent gradient anomaly points are merged to obtain several thickness gradient anomaly regions. Each thickness gradient anomaly region It contains a set of consecutive position indices.
[0046] Then, in order to fine-tune the local non-rigid misalignment caused by breathing factors, local translation correction is needed for the thickness gradient anomaly regions identified after the initial alignment. By probing tiny translations within a small window of the anomaly region and selecting the displacement that optimizes both the thickness gradient and grayscale, local residual misalignment can be eliminated with minimal computational cost.
[0047] Therefore, in some embodiments of the present invention, based on the CTA three-phase images after feature point matching, local translation correction is performed on the thickness gradient anomaly region to obtain the registered CTA three-phase images. Further embodiments include: First, for the three phases of CTA images after feature point matching, local windows are divided with the thickness gradient anomaly region as the center, and multiple local translations with different displacements are performed on the local windows in the flat scan and delayed scan images after feature point matching. Specifically, for each thickness gradient anomaly region... A rectangular window is defined centered on its geometric center, and denoted as the thickness gradient anomaly region. Local window Local window The size needs to cover the entire thickness gradient anomaly region and its adjacent range, such as a local window. The width is 30 pixels and the height is 20 pixels. For the target phase (the flat scan image after feature point matching)... and delayed images In a local window For the inner part, try a set of candidate displacements. The local translation yields the translational displacement. The subsequent local window image ;in, That is, remain stationary, move 1 pixel to the right, move 1 pixel to the left, move 1 pixel down, and move 1 pixel up.
[0048] Second, obtain the updated thickness gradient corresponding to each thickness gradient anomaly point within the translated local window, and obtain the grayscale value of each pixel within the translated local window, as well as the grayscale value of each pixel within the corresponding local window of the arterial phase image. Specifically, for each candidate displacement... The corresponding translated local window image Recalculate the local window image after translation. Inner gradient outliers The thickness gradient is used to obtain the translational displacement. Thickness gradient anomalies within the subsequent local window Corresponding update thickness gradient For each candidate displacement The corresponding translated local window image Obtain the local window image after translation. Inner pixel The grayscale value is denoted as ; and to obtain the pixels within the corresponding local window of the arterial phase image. The grayscale value is denoted as .
[0049] Third, the differences between each updated thickness gradient and the thickness gradient of the corresponding thickness gradient anomaly point in the arterial phase image are analyzed, and the gray-level differences of each pixel in the translated local window and the corresponding local window in the arterial phase image are analyzed to obtain the comprehensive score corresponding to each displacement of each local window. Specifically, candidate displacements are defined. Comprehensive scoring function The displacement is evaluated using a comprehensive scoring function. It needs to meet the following requirement: During the movement, the target phase (the flat scan image after feature point matching) must be maintained. and delayed images The thickness gradient difference between the image and the baseline arterial phase image should be as small as possible, and the similarity between the two should be high, i.e., the grayscale difference should be as small as possible. The specific function design is as follows: In the formula, Indicates a local window Displacement The comprehensive score function; Indicates translational displacement The subsequent local window Internal thickness gradient anomalies The corresponding updated thickness gradient; Indicates abnormal thickness gradient points within arterial phase images The corresponding thickness gradient; Indicates translational displacement The subsequent local window Inner pixel grayscale value; Represents pixels within an arterial phase image grayscale value; Indicates anomaly regions in thickness gradient The corresponding local window; The dimension conversion coefficient can be set based on experience (e.g.) ); Indicates taking the absolute value; This represents a linear normalization function, such as a max-min normalization function, used to normalize the sum of the absolute values of thickness gradient deviations and the sum of the absolute values of grayscale deviations to [values to be specified in the original text]. Within the range, the maximum value in the maximum-minimum normalization function is the preset upper limit of the range, which is set according to the maximum values of thickness gradient deviation and grayscale deviation that appear in historical data. Represented by natural constant An exponential function with base 0.
[0050] For local windows The sum of the absolute values of the thickness gradient deviations at all thickness gradient anomaly points within the area; For local windows The sum of the absolute values of the grayscale deviations of all pixels within a given area reflects the similarity between images. This evaluation function satisfies the condition that a higher overall score will result in a smaller local window. The smaller the difference in the gradient of blood vessel wall thickness change between the translated image and the baseline image (arterial phase image), and the smaller the difference in grayscale between the two images, the closer the two images are to true alignment.
[0051] Fourth, based on the comprehensive score, the optimal displacement is determined to obtain the registered CTA three-phase images, which include the arterial phase image, the registration plain scan image, and the registration delayed phase image. Specifically, the optimal displacement is selected to... Maximum displacement As a local window for this correction The optimal displacement is determined and applied to the local window. On the image within, update the target phase image (the plain scan image after feature point matching). and delayed images In this local window The pixel values of the region. When updating pixel values, the local window... Gaussian smoothing or linear weighted blending is applied to the edge regions to ensure a smooth transition between the corrected and uncorrected regions. This process is repeated until the local window is reached. The maximum thickness gradient deviation change of all thickness gradient anomalies within the region is less than 10% of the thickness gradient of the corresponding thickness gradient anomaly in the arterial phase image, or reaches the preset maximum number of iterations, such as 50. After processing all thickness gradient anomaly regions, the registered plain scan image is obtained. and registration delay images .
[0052] S300: Based on arterial phase images, plaque identification is performed, candidate plaques on the arterial phase images are marked, and the candidate plaques are mapped to the registered CTA three-phase images to obtain the corresponding candidate plaques on the plain scan images and delayed scan images.
[0053] Among the three phase images obtained by CTA scan, the arterial phase image corresponds to the moment when the concentration of contrast agent in the coronary artery reaches its peak. At this time, the lumen of the vessel is fully filled, the contrast between the vessel wall and the surrounding tissue is the highest, and the boundaries of the plaque and the contours of the inner and outer walls of the vessel are most clearly distinguishable.
[0054] Therefore, in embodiments of the present invention, plaque identification is performed based on arterial phase images, and candidate plaques on the arterial phase images are marked. Further steps include: in the arterial phase images, along the coronary artery centerline, examining multiple consecutive sampling points as a unit segment, marking segments with wall thickness exceeding a preset wall thickness and having CT values less than a preset CT value as candidate plaques, and generating a binary mask for each candidate plaque. Specifically, in the arterial phase images... First, the Unet model was used to segment the coronary artery vascular region and its inner and outer walls, and the coronary artery centerline was extracted using the minimum path method. Then, a sliding window approach was used to obtain unit segments; that is, with a step size of one sampling point, 10 consecutive sampling points were extracted along the coronary artery centerline as a unit segment for examination. Segments with a wall thickness exceeding 0.5 mm and containing an average CT value less than 60 HU (low-density region) were marked as candidate plaques, and a binary mask was generated for each candidate plaque. .
[0055] The candidate plaques are then mapped onto the registered CTA three-phase images, that is, the binary mask of each candidate plaque in the arterial phase image. Mapped to the registration plain scan image and registration delay images This yields binary masks of candidate patches on the plain scan image and the delayed scan image. Precisely corresponds to the location.
[0056] S400: Based on the registered CTA three-phase images, analyze the changes in CT values within candidate plaques over time, obtain the contrast agent retention index, and screen for suspected inflammatory plaques.
[0057] The CT value changes of inflammatory and stable coronary artery plaques differ significantly across different phases. The pathological essence of inflammatory coronary artery plaques lies in the active inflammatory response within them, specifically manifested as the formation of numerous new microvessels and significantly increased vascular permeability. When iodine-containing contrast agents are injected intravenously, they enter the coronary circulation with the bloodstream. In stable plaques, due to the lack of active microcirculation, the contrast agent can only passively diffuse through the surface of the vessel lumen, resulting in limited enhancement and rapid washout during the arterial phase. In inflammatory plaques, however, the abundant new microvessels allow for substantial infiltration of the contrast agent into the plaque interstitium, and the increased permeability leads to slow contrast agent clearance, forming a typical delayed retention phenomenon. The more pronounced the retention phenomenon and the greater the amount of retention, the higher the probability of plaque inflammation.
[0058] Based on the above analysis, in an embodiment of the present invention, based on the registered CTA three-phase images, the changes in CT values within candidate plaques over time are analyzed to obtain the contrast agent retention index and screen for suspected inflammatory plaques. Further steps include: First, CT values of candidate plaques are extracted from the registered CTA three-phase images. Specifically, in the arterial phase images... In the process, extract the binary mask of the candidate patch. Calculate the CT value of each pixel within the range, and calculate the first... Binary mask for candidate patches The third average CT value of all pixels within the area Images during registration plain scan In the process, extract the binary mask of the candidate patch. Calculate the CT value of each pixel within the range, and calculate the first... Binary mask for candidate patches The fourth average CT value of all pixels within the range Images during the registration delay period In the process, extract the binary mask of the candidate patch. Calculate the CT value of each pixel within the range, and calculate the first... Binary mask for candidate patches The fifth average CT value of all pixels within the range .
[0059] Then, the difference in CT values between the corresponding candidate patches in the registration delay image and the registration plain scan image is analyzed to obtain the contrast agent retention amount of the candidate patches. Specifically, based on the... Taking a candidate patch as an example, calculate the image corresponding to the registration delay period during the contrast agent clearance period. The Middle The fifth average CT value of the binary mask ROI for each candidate patch Registered plain scan images with contrast agent not filled The Middle The fourth average CT value of the candidate patch binary mask ROI The difference is the result of the first... The contrast agent retention amount for each candidate plaque is: In the formula, Indicates the first Contrast agent retention in each candidate plaque; Image representing the registration delay period The Middle The fifth average CT value of the binary mask ROI for each candidate plaque; Indicates the registration plain scan image The Middle The fourth average CT value of the candidate plaque binary mask ROI.
[0060] Simultaneously, the differences in CT values between corresponding candidate plaques in the arterial phase image and the registered plain scan image were analyzed to obtain the contrast agent filling amount of the candidate plaques. Specifically, based on the first... Taking a candidate plaque as an example, calculate the arterial phase image corresponding to the contrast agent filling period. The Middle Third average CT value of candidate patch binary mask ROI Registered plain scan images with contrast agent not filled The Middle The fourth average CT value of the candidate patch binary mask ROI The difference is the result of the first... The contrast agent filling volume for each candidate plaque is: In the formula, Indicates the first Contrast agent filling amount for each candidate plaque; Representing arterial phase images The Middle The third average CT value of the binary mask ROI for each candidate patch; Image representing the plain scan period The Middle The fourth average CT value of the candidate plaque binary mask ROI.
[0061] Furthermore, based on the contrast agent retention volume and the contrast agent filling volume, the contrast agent retention index is obtained. Specifically, the first... The retention index of contrast agent within each candidate plaque is: In the formula, Indicates the first The retention index of contrast agent within candidate plaques; Indicates the first Contrast agent retention in each candidate plaque; Indicates the first Contrast agent filling amount for each candidate plaque; This represents the denominator correction parameter, which has the same dimensions as the denominator and takes the smallest value greater than 0. This is to prevent the denominator from being 0; for example, it can be set... This can prevent the denominator from being 0 without significantly interfering with the calculation of normal values. This represents a linear normalization function, such as a max-min normalization function, used to normalize the retention exponent to... Within the range, the maximum value in the maximum-minimum normalization function is the preset upper limit of the range, which is set based on the maximum relative amount of contrast agent retention in historical data.
[0062] Retention Index This reflects the degree of contrast agent retention. The closer the value is to 1, the greater the retention of the contrast agent, indicating a higher probability that the candidate plaque binary mask ROI is an inflammatory plaque.
[0063] Similarly, the retention index of contrast agent within all candidate plaques was obtained.
[0064] Finally, based on the retention index, suspected inflammatory plaques are screened. Specifically, a preset retention index threshold is set (the recommended value is 0.6, which can be adjusted according to actual conditions). When the retention index threshold is reached... Retention index of contrast agent within candidate plaques Greater than the retention index threshold, i.e. Then the number One candidate plaque was marked as a suspected inflammatory plaque. By iterating through all candidate plaques, all suspected inflammatory plaques were identified. .
[0065] S500: For suspected inflammatory plaques, based on arterial phase images and registered delayed phase images, the enhancement amplitude and attenuation gradient of CT values in the adipose tissue region surrounding the plaque are analyzed to obtain the pericoronal fat inflammation index and confirm the inflammatory plaque.
[0066] To validate suspected inflammatory plaques, analyzing the density changes of the surrounding adipose tissue during the delayed phase can further confirm whether the inflammation has spread to the perivascular microenvironment, thereby improving the accuracy and reliability of inflammation identification.
[0067] The inflammatory response in coronary artery plaques is not limited to the vessel wall but also spreads through it to surrounding adipose tissue, causing pathological changes in the pericoronary fat. Inflammatory mediators can stimulate adipocyte edema, fibrosis, or angiogenesis, leading to increased uptake of iodine contrast agents by adipose tissue. This is manifested in elevated CT values of pericoronary fat in delayed-phase images, with the effect being more pronounced closer to the plaque, forming a proximal-distal density gradient. Therefore, by analyzing the enhancement amplitude and attenuation gradient of pericoronary fat in the delayed phase, evidence of inflammation spreading to the perivascular area can be indirectly captured, verifying whether the plaque is in an active inflammatory state.
[0068] Based on the above analysis, in the embodiments of the present invention, for suspected inflammatory plaques, the enhancement amplitude and attenuation gradient of the CT values in the adipose tissue region surrounding the plaque are analyzed based on arterial phase images and registered delayed phase images to obtain the pericoronal fat inflammation index, thus confirming the inflammatory plaque. Wherein: For suspected inflammatory plaques, the enhancement magnitude of CT values in the surrounding adipose tissue region was analyzed based on arterial phase images and registered delayed phase images, including: First, using the suspected inflammatory plaque as the center, a search is conducted along the extravascular region, interfering tissue is removed, and a non-vascular ring region outside the blood vessel is obtained as the peripheral candidate adipose tissue region for the suspected inflammatory plaque. Specifically, for the first... A suspected inflammatory plaque , with the first A suspected inflammatory plaque The geometric center of the region is a circle. A search is performed along the extravascular area with a radius of 5mm to obtain a non-vascular annular region. Considering that in actual anatomical structures, the 5mm radius around a blood vessel may contain myocardium, epicardium, even small veins or calcification artifacts, it is necessary to remove interfering tissues. Specifically, a preset fat CT value range is used, and pixels within this annular region are threshold-filtered, retaining only pixels with CT values within the preset fat CT value range (e.g., -190HU to -30HU). These are the effective fat tissue pixels within the annular region, and are used as the first... A suspected inflammatory plaque Peripheral candidate adipose tissue region .
[0069] Then, on the arterial phase images, the first average CT value of all pixels within the peripheral candidate adipose tissue region is calculated. Specifically, on the arterial phase images... Above, calculate the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The first average CT value of all pixels within the range is denoted as .
[0070] Furthermore, on the registered delayed-phase image, the second average CT value of all pixels within the peripheral candidate adipose tissue region is calculated. Specifically, on the registered delayed-phase image... Above, calculate the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The second average CT value of all pixels within the range is denoted as .
[0071] Finally, based on the first and second average CT values, the enhancement magnitude of the CT values in the peripheral candidate adipose tissue region of the suspected inflammatory plaque was obtained. Specifically, the enhancement magnitude of the CT values in the peripheral candidate adipose tissue region was calculated. A suspected inflammatory plaque Arterial phase images and delayed images The difference in average CT values yields the first... A suspected inflammatory plaque Peripheral candidate adipose tissue region The enhancement magnitude of the internal CT value is: In the formula, Indicates the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The magnitude of increase in internal CT values; Indicates the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The first average CT value of all pixels within the area; Indicates the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The second average CT value of all pixels within the area; This represents a linear normalization function, such as a max-min normalization function, used to normalize the enhancement magnitude to... Within the range, the maximum value in the maximum-minimum normalization function is the preset upper limit of the range, which is set based on the maximum enhancement of CT values appearing in historical data.
[0072] To verify whether the inflammation indeed originates from the current plaque and spreads outwards, the attenuation gradient of CT values in the surrounding adipose tissue region was further analyzed. Further analysis included: First, the peripheral candidate adipose tissue region was divided into proximal and distal layers based on its distance from the vessel wall. Specifically, to eliminate interference from non-adipose tissue, only pixels with CT values in the range of -190 HU to -30 HU were retained from the peripheral candidate adipose tissue region. These retained pixels were then further divided into proximal and distal layers based on their distance from the vessel wall; that is, the layer 1-2 mm from the vessel wall was designated as the proximal layer. The area 4-5 mm from the outer wall of the blood vessel is divided into distal layers. .
[0073] Then, on the registered delayed-phase image, the average CT value of all pixels in the proximal and distal layers is calculated to obtain the attenuation gradient of the CT value of the peripheral candidate adipose tissue region of the suspected inflammatory plaque. Specifically, on the registered delayed-phase image... Calculate the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The sixth average CT value of all pixels in the proximal layer , and calculate the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The seventh average CT value of all pixels in the far layer Then, the average CT difference between the proximal and distal layers is calculated to obtain the first... A suspected inflammatory plaque Peripheral candidate adipose tissue region The attenuation gradient of the internal CT value is: In the formula, Indicates the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The attenuation gradient of the internal CT value; Indicates the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The sixth average CT value of all pixels in the proximal layer; Indicates the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The seventh average CT value of all pixels in the far layer; This represents a linear normalization function, such as a max-min normalization function, used to normalize the decaying gradient to... Within the range, the maximum value in the maximum-minimum normalization function is the preset upper limit of the range, which is set according to the maximum attenuation of the CT value that appears in historical data.
[0074] Obtain the pericoronal fat inflammation index to confirm inflammatory plaques. Specifically, the extent of enhancement will be determined. and decay gradient The inflammatory index of pericoronary fat was integrated to quantify the overall intensity and attenuation characteristics of inflammation on adipose tissue, and to construct the first... A suspected inflammatory plaque Peripheral candidate adipose tissue region The formula for calculating the inflammation index is: In the formula, Indicates the first A suspected inflammatory plaque Peripheral candidate adipose tissue region Inflammation index; Indicates the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The magnitude of increase in internal CT values; Indicates the first A suspected inflammatory plaque Peripheral candidate adipose tissue region The attenuation gradient of the internal CT value; Inflammation index The value range is [0,1]. If the fat density in the delayed phase is significantly increased and there is a significant difference in fat density between the proximal and distal ends, the corresponding value is... The closer the value is to 1, the more pronounced the inflammatory response of the suspected inflammatory plaque, and the higher the probability that it is a true inflammatory plaque.
[0075] Set an inflammation index threshold, such as 0.7, to classify suspected inflammatory plaques. Peripheral candidate adipose tissue region Inflammation index Suspected inflammatory plaques with a value greater than 0.7 were confirmed as actual inflammatory plaques. ,like Figure 2 As shown, perform the final output.
[0076] After completing the identification and verification of inflammatory plaques, the final actual inflammatory plaques will be... The results are output in an intuitive and diagnostic format. Specifically, the final actual inflammatory plaques will be presented. The data is mapped back to the original 3D CTA data space, onto a 3D coronary tree, and the inflammatory plaque areas are marked with a bright color and the surrounding myocardial and vascular structures are made semi-transparent to generate a 3D visualization model.
[0077] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0078] The various embodiments in this specification are described in a progressive manner. For the same or similar parts between the various embodiments, please refer to each other. Each embodiment focuses on describing the differences from other embodiments.
Claims
1. A method for identifying coronary artery inflammatory plaques based on multi-sequence CTA images, characterized in that, The method includes: Acquire three-phase CTA images of the coronary arteries, including plain scan images, arterial scan images, and delayed scan images; Feature point matching and local thickness gradient correction are performed on the three phase CTA images to obtain the registered three phase CTA images; Based on the arterial phase image, plaque identification is performed, candidate plaques on the arterial phase image are marked, and the candidate plaques are mapped to the registered CTA three-phase image to obtain the corresponding candidate plaques on the plain scan image and the delayed scan image. Based on the registered CTA three-phase images, the changes in CT values within the candidate plaques over time are analyzed to obtain the contrast agent retention index and screen for suspected inflammatory plaques. For the suspected inflammatory plaque, based on the arterial phase image and the registered delayed phase image, the enhancement amplitude and attenuation gradient of the CT value of the adipose tissue region surrounding the plaque are analyzed to obtain the pericoronal fat inflammation index and confirm the inflammatory plaque.
2. The method for identifying coronary inflammatory plaques based on multi-sequence CTA images according to claim 1, characterized in that, Feature point matching is performed on the three phases of the CTA images, including: For the three phases of CTA images, the coronary artery centerline is identified, and the bifurcation points and curvature maxima on the coronary artery centerline are extracted to form feature point sets corresponding to the plain scan image, arterial scan image, and delayed scan image. Using the arterial phase image as a reference, the feature point set of the arterial phase image is matched with the feature point sets corresponding to the plain scan image and the delayed phase image, respectively, to obtain the CTA three-phase image after feature point matching. The CTA three-phase image after feature point matching includes the arterial phase image, the feature point matched plain scan image, and the feature point matched delayed phase image.
3. The method for identifying coronary artery inflammatory plaques based on multi-sequence CTA images according to claim 2, characterized in that, Using the arterial phase image as a reference, the feature point set of the arterial phase image is type-matched with the feature point sets corresponding to the plain scan image and the delayed phase image, respectively, to obtain a CTA three-phase image after feature point matching, including: Using the arterial phase image as a reference, the feature point set of the arterial phase image is matched with the feature point sets corresponding to the plain scan image and the delayed phase image respectively through nearest neighbor search. False matches with a distance greater than a preset pixel are eliminated to obtain the effective matching pairs corresponding to the feature point sets of the arterial phase image and the feature point sets corresponding to the plain scan image and the delayed phase image respectively. Based on the effective matching pairs, the rigid transformation parameters are solved using the least squares method to minimize the sum of squared errors of the matching pairs. The transformation parameters are then applied to the flat scan image and the delayed image, and the feature point matching flat scan image and the feature point matching delayed image are obtained by interpolation.
4. The method for identifying coronary artery inflammatory plaques based on multi-sequence CTA images according to claim 2, characterized in that, Local thickness gradient correction is performed on the three phases of the CTA images, including: Based on the CTA three-phase images after feature point matching, the differences between the thickness gradients of the blood vessel wall at the same sampling location are analyzed to identify abnormal thickness gradient regions. Based on the CTA phase III images after feature point matching, the thickness gradient anomaly region is locally translated and corrected to obtain the registered CTA phase III images.
5. The method for identifying coronary artery inflammatory plaques based on multi-sequence CTA images according to claim 4, characterized in that, Based on the CTA three-phase images after feature point matching, the differences in thickness gradients of the vessel wall at the same sampling location are analyzed to identify regions with abnormal thickness gradients, including: For the CTA phase III images after feature point matching, the vessel wall thickness corresponding to the same sampling position is obtained, and the difference of the vessel wall thickness at adjacent sampling positions is calculated to obtain the vessel wall thickness gradient at each sampling position on the CTA phase III images after feature point matching. Using the arterial phase image as a reference, calculate the absolute thickness gradient deviation between the plain scan image and the delayed scan image after feature point matching at the same sampling location and the arterial phase image, respectively. Set a preset absolute thickness gradient deviation threshold, mark thickness gradient abnormal points, and identify thickness gradient abnormal regions.
6. The method for identifying coronary artery inflammatory plaques based on multi-sequence CTA images according to claim 5, characterized in that, Based on the CTA phase III images after feature point matching, local translation correction is performed on the thickness gradient anomaly region to obtain the registered CTA phase III images, including: For the CTA three-phase images after feature point matching, a local window is divided with the thickness gradient anomaly region as the center, and the local window in the flat scan image and the delayed image after feature point matching is translated multiple times with different displacements. Get the updated thickness gradient corresponding to each thickness gradient anomaly point in the translated local window, get the gray value of each pixel in the translated local window, and get the gray value of each pixel in the corresponding local window of the arterial phase image. The difference between each updated thickness gradient and the thickness gradient of the corresponding thickness gradient anomaly point in the arterial phase image is analyzed, and the grayscale difference of each pixel in the local window after translation and the local window corresponding to the arterial phase image is analyzed to obtain the comprehensive score corresponding to each displacement of each local window; Based on the comprehensive score, the optimal displacement is determined to obtain the registered CTA three-phase images, which include the arterial phase image, the registered plain scan image, and the registered delayed scan image.
7. The method for identifying coronary inflammatory plaques based on multi-sequence CTA images according to claim 1, characterized in that, Based on the arterial phase image, plaque identification is performed, and candidate plaques on the arterial phase image are marked, including: In the arterial phase image, along the coronary artery centerline, multiple consecutive sampling points are examined as a unit segment. Segments with a wall thickness exceeding a preset wall thickness and a CT value less than a preset CT value are marked as candidate plaques, and a binary mask is generated for each candidate plaque.
8. The method for identifying coronary artery inflammatory plaques based on multi-sequence CTA images according to claim 6, characterized in that, Based on the registered CTA images from three phases, the changes in CT values within the candidate plaques over time are analyzed to obtain the contrast agent retention index, including: For the registered CTA three-phase images, the CT values of the candidate plaques are extracted; The difference in CT values between the candidate plaques in the registration delay image and the registration plain scan image is analyzed to obtain the contrast agent retention amount of the candidate plaques; The difference in CT values between the candidate plaques in the arterial phase image and the registered plain scan image is analyzed to obtain the contrast agent filling amount of the candidate plaques; The contrast agent retention index is obtained based on the contrast agent retention amount and the contrast agent filling amount.
9. The method for identifying coronary inflammatory plaques based on multi-sequence CTA images according to claim 1, characterized in that, For the suspected inflammatory plaque, based on the arterial phase image and the registered delayed phase image, the enhancement magnitude of the CT value of the adipose tissue region surrounding the plaque is analyzed, including: Using the suspected inflammatory plaque as the center, a search is performed along the extravascular region, interfering tissues are eliminated, and a non-vascular ring region outside the blood vessel is obtained as the peripheral candidate adipose tissue region of the suspected inflammatory plaque. On the arterial phase image, calculate the first average CT value of all pixels within the peripheral candidate adipose tissue region; On the registered delayed image, calculate the second average CT value of all pixels in the peripheral candidate adipose tissue region; The enhancement magnitude of the CT value in the peripheral candidate adipose tissue region of the suspected inflammatory plaque is obtained based on the first average CT value and the second average CT value.
10. The method for identifying coronary inflammatory plaques based on multi-sequence CTA images according to claim 9, characterized in that, Analysis of the attenuation gradient of CT values in the adipose tissue region surrounding the plaque, including: The peripheral candidate adipose tissue region is divided into a proximal layer and a distal layer according to its distance from the outer wall of the blood vessel; On the registered delayed image, the average CT value of all pixels in the proximal layer and the distal layer is calculated respectively to obtain the attenuation gradient of the CT value of the peripheral candidate adipose tissue region of the suspected inflammatory plaque.