A method for comprehensive quantitative assessment of vessel stenosis from contrast images

Abstract

The application discloses a method for comprehensive quantitative evaluation of blood vessel stenosis from angiography images, which comprises image preprocessing, automatic stenosis evaluation and interactive stenosis evaluation modules. The preprocessing module combines TOF, UM, CLAHE and Hessian matrix multi-scale enhancement algorithm to improve image quality and highlight blood vessel structure. The automatic stenosis evaluation module uses a new tracking algorithm to extract blood vessel skeleton and blood vessel segmentation, and uses CV model to extract blood vessel contour, measure diameter and stenosis degree of each part of blood vessel tree and detect stenosis. The interactive stenosis evaluation module selects target blood vessel segment by user, uses the proposed energy function-based tracking algorithm to extract target blood vessel skeleton, uses CV model to extract contour, measures diameter and stenosis degree of each part of blood vessel segment and detects stenosis. The application combines automatic and interactive stenosis evaluation, which can not only automatically detect angiography images in batches but also quantitatively evaluate specific images in detail.

Description

Technical Field

[0001] This invention relates to a comprehensive quantitative assessment method for vascular stenosis in coronary angiography images, including a fully automated assessment method and an interactive assessment method, belonging to the field of medical image processing. Background Technology

[0002] Cardiovascular disease is recognized worldwide as one of the leading causes of death from chronic diseases. In recent years, the incidence and mortality rates of cardiovascular disease have continued to rise, ranking first among all diseases. Coronary angiography is a commonly used and effective method for diagnosing coronary heart disease (CHD), considered the "gold standard" for diagnosis and widely applied in clinical practice. Human arteries are composed of soft tissue and are normally invisible under X-rays. However, by injecting a radiopaque substance into the coronary arteries and simultaneously irradiating the area with X-rays, the arteries can be visualized under X-rays. Doctors can then use these images to identify the location and extent of coronary artery stenosis and determine the appropriate treatment plan.

[0003] Currently, the determination of coronary artery lesion location and the assessment of coronary artery stenosis degree are mostly based on doctors' subjective measurements. This repetitive work not only reduces efficiency but also introduces significant subjectivity, leading to inaccurate results. Therefore, to improve the efficiency and accuracy of coronary artery disease detection, we need to utilize image processing technology to accurately extract the vascular skeleton and contour from angiographic images and measure the diameter and degree of stenosis of various blood vessels.

[0004] Because the vascular structure differs significantly from the background in terms of grayscale and morphology in contrast images, a suitable method can be used to comprehensively consider these features to accurately extract the vascular skeleton and contour. Using the obtained skeleton and contour information, the diameter and degree of stenosis at various points along the vessel can be measured, thereby accurately assessing vascular stenosis.

[0005] However, most current methods for assessing vascular stenosis rely on a single, global detection approach. In practice, healthcare professionals may sometimes require further quantitative analysis of a specific segment within the vascular tree. Furthermore, due to inherent design flaws in various technologies, the accuracy of the results still needs improvement. Summary of the Invention

[0006] This invention addresses the low efficiency and accuracy of manual coronary angiography in practical medical scenarios, as well as the limited and imprecise image processing techniques used. It proposes a comprehensive method for quantitatively assessing vascular stenosis in angiography images, including both fully automated and interactive assessment methods. In practical applications, this method can be used for efficient and automated global vascular stenosis assessment of coronary angiography images, while also allowing medical personnel to interactively select a specific vessel segment for targeted quantitative analysis. The two methods complement each other and possess strong robustness, effectively meeting the clinical needs of cardiovascular disease diagnosis and treatment. This invention mainly comprises three parts: image preprocessing, automated stenosis assessment, and interactive assessment.

[0007] The image preprocessing section achieves image denoising, improves image contrast, enhances tubular structures in the image, significantly increases the difference between the vascular tree and the background, and reduces the interference of non-tubular targets in the background.

[0008] The automatic stenosis assessment section utilizes a CV model to extract the vascular tree contour and proposes a novel tracking algorithm to extract the vascular skeleton (vascular centerline), detect bifurcation points, and segment the blood vessel. Based on this, the diameter of the blood vessel at various points is measured and the degree of stenosis is assessed. This method can assess all stenosis points in the entire vascular tree without human interference throughout the process.

[0009] The interactive stenosis assessment section also utilizes the CV model to extract the vascular tree contour. After the user interactively selects a segment of blood vessel, a second energy function-based tracking algorithm is proposed to extract the skeleton of that segment, thereby measuring the diameter and assessing the degree of stenosis at various points in the segment and detecting all stenosis points in that segment.

[0010] The advantages of this comprehensive quantitative assessment of vascular stenosis are:

[0011] 1. This invention minimizes manual operation during detection. For the automatic stenosis assessment section, no human intervention is required throughout the entire process; for interactive assessment, the user only needs to manually select a start point and an end point to determine the target vessel segment, and the rest of the operation is fully automatic.

[0012] 2. The two tracking algorithms proposed in this invention have higher accuracy and stronger robustness compared with existing tracking methods such as adaptive geometric tracking, and can extract the vascular tree skeleton or the skeleton of a specific target vascular segment more accurately.

[0013] 3. This invention achieves a good balance between computational complexity and evaluation effect, enabling efficient detection of vascular stenosis and facilitating clinical application;

[0014] 4. The quantitative processing performed in this invention includes diameter measurement and stenosis assessment at various points, which can present the relative stenosis of blood vessels at various points to the user in a very intuitive and accurate manner, and has an unparalleled advantage over manual estimation. Attached Figure Description

[0015] Figure 1 This is a flowchart of the automatic stenosis assessment method proposed in this invention.

[0016] Figure 2 This is a flowchart of the interactive stenosis assessment method proposed in this invention.

[0017] Figure 3 This is a schematic diagram of the initial tracking direction proposed in this invention.

[0018] Figure 4 This is a schematic diagram of the tracking direction proposed in this invention.

[0019] Figure 5 This is a schematic diagram of the center line adjustment for the tracking point proposed in this invention.

[0020] Figure 6 This is a schematic diagram of the bifurcation point detection proposed in this invention.

[0021] Figure 7 This is a schematic diagram of the blood vessel segmentation proposed in this invention.

[0022] Figure 8 This is a schematic diagram of diameter measurement proposed in this invention. Detailed Implementation

[0023] The advantages and spirit of the present invention can be further understood from the following detailed description of the invention and the accompanying drawings.

[0024] Step S101: Read coronary angiography image data. The data can be DICOM sequence slice images or single two-dimensional angiography images.

[0025] Step S102 involves preprocessing the angiography image. First, the ROF algorithm is used for denoising. Then, the UM algorithm is used to generate a new enhanced edge image from the old image. Next, contrast-limited adaptive histogram equalization is used to increase the image contrast. Finally, a multi-scale image enhancement method based on the image Hessian matrix is ​​used to highlight the vascular structure.

[0026] The following are the same preliminary steps as the automatic stenosis assessment and interactive stenosis assessment sections.

[0027] Step S103: Apply the CV model to the preprocessed image to obtain the vascular tree contour.

[0028] Step S104: Detect vascular ridges (local brightness maxima) in the preprocessed image, and identify (x,y) pixels that satisfy the following condition:

[0029] (1).

[0030] The following is the remaining section on automatic narrowness assessment.

[0031] Step S105: Randomly select seed points from the obtained ridge points. Initial tracking is performed, and the initial tracking direction can be obtained from the grayscale information of blood vessels near the seed point. Specifically, using... With center at and radius at, Searching for grayscale maxima on a circle The positive direction of tracking is represented as:

[0032] (2)

[0033] After obtaining the forward tracking angle, the reverse tracking angle can be obtained from the forward direction angle. opposite direction The search scope centered Search for local maxima on the corresponding arc Similarly, we can track in the opposite direction. .

[0034] This step is available. Figure 1 express.

[0035] Step S106, tracing forward from the current point to the next tracking point, is a key part of the tracking algorithm. The current tracking direction starts from the previous tracking point. Up to the current point Direction determines:

[0036] (3)

[0037] Then, in the arc The upper search finds local maxima that satisfy the following formula. :

[0038] (4)

[0039] in, yes grayscale value, It is a given threshold. yes The number of tracking points in the surrounding area It is also a threshold.

[0040] This process is available. Figure 2 express.

[0041] During the tracking process, the tracking points may deviate from the center of the blood vessel. Therefore, it is necessary to use the blood vessel contour obtained in step S103 to adjust each tracking point to the center line of the blood vessel. Specifically, this involves calculating the normal of the current tracking direction and finding the intersection point between the normal and the blood vessel contour. The tracking point can be adjusted. for:

[0042] (5)

[0043] Accordingly, the tracking direction is adjusted as follows:

[0044] (6)

[0045] This process can be used Figure 3 express.

[0046] Step S107: Bifurcation point detection is required at each tracking point. Specifically, this involves detecting the bifurcation point at each radian. With radius Search for ridge points within the fan-shaped annular region. For each ridge point, it is a bifurcation point if the following condition is met:

[0047] (7)

[0048] This process can be used Figure 4 express.

[0049] Because the diameters of different vascular segments (trunk and branches) in the vascular tree vary greatly, vascular segmentation is necessary to facilitate subsequent stenosis assessment. The vascular segmentation method of this invention is entirely based on the tracking point sequence information recorded during the tracking process. Any two adjacent cutoff points (bifurcation point and segment tracking termination point) in the tracking point sequence constitute a vascular segment.

[0050] This process can be used Figure 5 express.

[0051] Step S108: Based on the adjusted vascular skeleton information and the extracted vascular contour, the diameter and degree of stenosis of the target vascular tracking point can be measured. The measurement method of this invention is as follows: for a given tracking point... In other words, we utilize And fit a straight line to the three nearest tracking points (or all tracking points if there are fewer than three). Then, the corresponding passing point can be obtained. normal The distance between the two intersection points of the normal and the blood vessel contour is taken as the diameter of the blood vessel at that point.

[0052] This process can be used Figure 6 express.

[0053] For a given blood vessel segment, traverse all tracking points along the centerline of the current blood vessel segment from the starting point and calculate the diameter corresponding to each tracking point. and the average diameter of all tracking points in the blood vessel segment Tracking point The degree of stenosis of the blood vessel is as follows:

[0054] (8).

[0055] Step S109: Measure the diameter at various points on the vascular tree to mark stenosis. The procedure is as follows:

[0056] 1. Preprocess the obtained segmentation results to remove vascular segments that are too short to be suitable for evaluation;

[0057] 2. Traverse all blood vessel segments and calculate the diameter and degree of stenosis at each tracking point on each segment;

[0058] 3. Given a threshold T, mark the tracking points where the narrowness is less than T; these are the narrow points of the segment.

[0059] The following is the remaining interactive narrow assessment section.

[0060] In step S110, the user manually selects the start and end points to determine the target blood vessel segment.

[0061] Step S111: Find the ridge points adjacent to the start and end points. Based on our step S104, we obtained the initial set of ridge points (RP). Therefore, the start and end points used for tracing can be represented as:

[0062] (9).

[0063] Step S112, Tracking. Start tracking from the seed point. Select the point on the search arc that maximizes the energy function value below as the next tracking point:

[0064] (10)

[0065] Then, determine the tracking point and The tracking process checks if the distance between the two points is less than a given threshold. If the distance is within a threshold, we consider the two points close enough to stop tracking; otherwise, we continue.

[0066] Step S113: Select the correct tracking route. Execute step S112 in both the forward and reverse tracking directions to obtain two routes, one forward and one reverse. Select the route closer to the termination tracking point. The route.

[0067] Step S114: Calculate the diameter and degree of stenosis at each tracking point on the blood vessel segment. Given a threshold T, mark the tracking points where the degree of stenosis is less than T. These are the stenotic points of the blood vessel segment.

[0068] Although the present invention has been described with reference to preferred embodiments, the examples described above do not constitute a limitation on the scope of protection of the present invention. Any modifications, equivalent substitutions and improvements within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A method for comprehensive quantitative evaluation of vascular stenosis from angiography images, consisting of three key modules, comprising: (1) image preprocessing module performing steps S101-S102: Step S101, reading coronary angiography image data, the data is DICOM sequence slice image or single two-dimensional angiography image; Step S102, performing image preprocessing on the angiography image, first using ROF algorithm for denoising, then using UM algorithm to generate a new enhanced edge image from the old image, and then using contrast limited adaptive histogram equalization to increase the contrast of the image, and finally using a multi-scale image enhancement method based on image Hessian matrix to highlight the vascular structure; Step S103, applying CV model to the preprocessed image to obtain the vascular tree contour; Step S104, detecting the vascular ridge point, i.e. local brightness maximum point, in the preprocessed image; (2) automatic stenosis evaluation module performing steps S103-S104, further comprising the following steps: Step S105, a seed point P0 is randomly selected from the acquired ridge points for initialization tracking, and the initial tracking direction is obtained from the blood vessel grayscale information near the seed point; specifically, the maximum grayscale point P + is searched on the circle with P0 as the center and a radius d, and the positive direction of tracking is represented as: After obtaining the forward tracking angle, the reverse tracking angle is at the positive direction angle θ. + The opposite direction (2π-θ) + The search range centered on (2π-θ) + -Δθ,2π-θ + Search for the local maximum point P on the arc corresponding to +Δθ). - Similarly, we can obtain the tracking direction u(P) in the opposite direction. - ); Step S106, forward tracking from the current point to the next tracking point, the current tracking direction is determined by the direction from the last tracking point P k-1 to the current point P k . on an arc (θ k - Δθ, θ k + Δθ) that satisfies the following equation k+1 : where I(P k+1 ) is the gray value of P k+1 , τ is a given threshold, aroundP(P k+1 ) is the number of tracking points around P k+1 , and limit_num_P is also a threshold. Step S107, bifurcation point detection is required at each tracking point, specifically, search for ridge points in the sector area between the radian (θ k - Δθ,θ k + Δθ) and the radius (r1, r2), and for each ridge point, if the following conditions are met, it is a bifurcation point: Step S108, according to the adjusted blood vessel skeleton information and the extracted blood vessel contour, the diameter and the stenosis degree of the target blood vessel tracking point can be measured; for a tracking point P k In this regard, we use P k and the three tracking points adjacent thereto to fit a straight line l1, and then obtain the normal line l2 corresponding to the point P k , and take the distance between the two intersection points of the normal line and the blood vessel contour as the diameter of the blood vessel at the point. For a certain vessel segment, from the start point, traverse all the tracking points on the current vessel segment centerline, calculate the diameter D corresponding to each tracking point k and the average diameter of all tracking points of the vessel segment Tracking point P k The degree of stenosis of the vessel at the tracking point P is: Step S109, measuring the diameter of the vascular tree at each point to mark the stenosis, the operation being as follows: preprocessing the obtained segmentation result to remove the short blood vessel segments that are not suitable for evaluation; traversing all blood vessel segments to obtain the diameter and stenosis degree at each tracking point on each blood vessel segment; given a threshold T, marking the tracking points with stenosis degree less than T, which are the stenosis positions of the segment; (3) interactive stenosis evaluation module performing steps S103-S104, further comprising the following steps: Step S110, manually selecting start and end points to determine the target blood vessel segment; Step S111, finding the ridge points adjacent to the start and end points; Step S112, tracking based on energy function; Step S113, select the correct tracking route, execute step S112 in forward and reverse tracking direction respectively, get two routes, select the route with the termination tracking point closer to P end ​ Step S114, obtaining the diameter and stenosis degree at each tracking point on the blood vessel segment, and given a threshold T, marking the tracking points with stenosis degree less than T, which are the stenosis positions of the blood vessel segment.

2. The method for comprehensive quantitative evaluation of vascular stenosis according to claim 1, wherein the automatic stenosis evaluation module can automatically extract the vascular tree contour, obtain a large number of seed points located on the vascular centerline, extract the vascular tree skeleton from a certain seed point, identify the vascular bifurcation point, perform blood vessel segmentation, and automatically measure the diameter and evaluate the stenosis degree at each position of each segment of the vascular tree, thereby detecting the vascular stenosis position.

3. The method for comprehensive quantitative evaluation of vascular stenosis according to claim 1, wherein the interactive stenosis evaluation module can extract the contour of the target blood vessel segment and the centerline, and automatically measure the diameter and evaluate the stenosis degree at each position of the blood vessel segment, thereby detecting the stenosis position of the target blood vessel segment.

Images