Coronary artery maximum pressure gradient calculation method and device based on X-ray contrast image

Abstract

The invention provides a coronary artery maximum pressure gradient calculation method and device based on an X-ray contrast image, and is applied to the technical field of data processing. According to the method, automatic screening of functional angiography frames is carried out based on preprocessed angiography images, difference between each preprocessed image frame and a first background frame is carried out to obtain a difference image, gray variance of the difference image is calculated, and the frame with the maximum variance is selected as a target angiography image of the visual angle for an image sequence of each projection angle; performing three-dimensional blood vessel model reconstruction and coordinate transformation on the target angiography image, and constructing a coronary artery three-dimensional network model; calculating hemodynamic parameters based on the coronary artery three-dimensional network model; a PPG index is constructed based on hemodynamic parameters, lesion classification is carried out, a PPG index formula is derived by integrating the 20mm maximum pressure gradient and the diffuse lesion proportion, a single PPG index value corresponds to a treatment strategy through an inverse proportion weighting algorithm, and classification information of coronary artery lesion is generated.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and apparatus for calculating the maximum pressure gradient of coronary arteries based on X-ray angiography images. Background Technology

[0002] In the clinical assessment of coronary atherosclerotic heart disease, the maximum pressure gradient (PPG) is a core indicator for assessing the hemodynamic impact of coronary artery stenosis, defined as the pressure difference between the two ends of the stenotic segment. Current mainstream methods rely on manually withdrawing an invasive pressure guidewire to collect fractional flow reserve (FFR), combined with manual analysis of the stenosis location using two-dimensional coronary angiography images. The specific procedure is as follows: the physician manually reviews the X-ray angiography image sequence frame by frame, subjectively selecting the "optimal frame" as the assessment basis; the pressure guidewire is advanced into the distal coronary artery via a guiding catheter, manually withdrawn at a uniform speed, and FFR values ​​are recorded at 10-20 mm intervals, generating a curve of FFR changing with vessel length; based on the stenosis location identified by the angiography "optimal frame," the pressure difference corresponding to the FFR curve is extracted, and combined with the vessel diameter stenosis rate, PPG is estimated.

[0003] Current technologies rely on digital subtraction angiography (DSA), pressure guidewires, and hemodynamic monitoring devices, but they have significant drawbacks: manual selection of the optimal frame is inefficient (requiring 5-10 minutes per vessel) and highly subjective (20%-30% of cases are misdiagnosed); uneven manual retraction speed of the pressure guidewire leads to insufficient data accuracy, incomplete capture of diffuse lesions (missed diagnosis rate of 15%-20%), and underestimates the PPG of tandem lesions (deviation of 30%-40%); two-dimensional angiography cannot reflect the three-dimensional characteristics of blood vessels, resulting in 25%-30% of "mild stenosis" being missed, and ignoring the progressive pressure loss of diffuse plaques.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this application is to provide a method and apparatus for calculating the maximum coronary pressure gradient (PPG) based on X-ray angiography images, which at least to some extent overcomes the problems existing in the prior art. It achieves automated PPG calculation through X-ray angiography sequences: acquiring multi-view DICOM sequences and parameters, preprocessing them with Gaussian filtering and downsampling; selecting the optimal frame based on grayscale variance, reconstructing a three-dimensional model through coordinate transformation, angle correction, and least squares method; calculating blood flow parameters using the Navier-Stokes equations, constructing a PPG index-corresponding treatment strategy, solving the problems of manual inefficiency and inaccurate data acquisition, and improving assessment efficiency.

[0006] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0007] According to one aspect of this application, a method for calculating the maximum pressure gradient of coronary arteries based on X-ray angiography images is provided, comprising: acquiring and preprocessing multi-view coronary angiography images to generate preprocessed angiography images; performing automated functional angiography frame selection based on the preprocessed angiography images, subtracting each preprocessed image from the first background frame to obtain a difference image, calculating the grayscale variance of the difference image, and selecting the frame with the largest variance as the target angiography image for each projection angle image sequence; reconstructing and transforming a three-dimensional vascular model of the target angiography image to construct a three-dimensional network model of the coronary arteries; calculating hemodynamic parameters based on the three-dimensional network model of the coronary arteries; constructing a PPG index based on the hemodynamic parameters and classifying lesions, deriving the PPG index formula by integrating the 20mm maximum pressure gradient and the proportion of diffuse lesions, and generating classification information of coronary artery lesions by using an inverse proportional weighting algorithm to make a single PPG index value correspond to a treatment strategy.

[0008] Another aspect of this application discloses a device for calculating the maximum pressure gradient of coronary arteries based on X-ray angiography images, characterized by comprising: an acquisition module for acquiring and preprocessing multi-view coronary angiography images to generate preprocessed angiography images; a processing module for automatically filtering functional angiography frames based on the preprocessed angiography images, subtracting each preprocessed image from the first background frame to obtain a difference image, calculating the grayscale variance of the difference image, and selecting the frame with the largest variance as the target angiography image for each projection angle; reconstructing and transforming a three-dimensional vascular model of the target angiography image to construct a three-dimensional network model of the coronary arteries; calculating hemodynamic parameters based on the three-dimensional network model of the coronary arteries; constructing a PPG index based on the hemodynamic parameters and classifying lesions, deriving the PPG index formula by integrating the 20mm maximum pressure gradient and the proportion of diffuse lesions, and generating classification information of coronary artery lesions by using an inverse proportional weighting algorithm to make a single PPG index value correspond to a treatment strategy.

[0009] According to another aspect of this application, an electronic device is characterized by comprising: a first processor; and a memory for storing executable instructions of the first processor; wherein the first processor is configured to execute the above-described method for calculating the maximum coronary artery pressure gradient based on X-ray angiography images by executing the executable instructions.

[0010] According to another aspect of this application, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a second processor, implements the above-described method for calculating the maximum coronary artery pressure gradient based on X-ray angiography images.

[0011] This application provides a method and apparatus for calculating the maximum pressure gradient (PPG) of coronary arteries based on X-ray angiography images. The method achieves automated and accurate calculation of the PPG using X-ray angiography image sequences. The process is as follows: acquiring multi-view DICOM angiography sequences and equipment parameters; preprocessing with 3×3 kernel Gaussian filtering for noise reduction and 2×2 mean downsampling; automatically selecting the functionally optimal frame with the largest variance through difference image grayscale variance analysis; reconstructing a three-dimensional vascular model through coordinate transformation, angle correction, and least squares fitting; calculating blood flow parameters based on the Navier-Stokes equations; extracting the maximum pressure gradient and functional lesion length within a 20mm window; constructing a PPG index and applying it inversely proportionally to correspond to treatment strategies. This effectively solves the problems of inefficiency in manual screening and inaccurate pressure acquisition, improving assessment efficiency and accuracy.

[0012] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0013] Figure 1 This document illustrates a flowchart of a method for calculating the maximum coronary artery pressure gradient based on X-ray angiography images, provided in an embodiment of this application.

[0014] Figure 2 This illustration shows a schematic diagram of a coronary artery maximum pressure gradient calculation device based on X-ray angiography images, provided in one embodiment of this application. Detailed Implementation

[0015] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0016] The following is combined Figure 1 This application describes a method for calculating the maximum coronary artery pressure gradient based on X-ray angiography images, according to exemplary embodiments thereof. It should be noted that the following application scenarios are shown only to facilitate understanding of the spirit and principles of this application, and the embodiments of this application are not limited in any way. Rather, the embodiments of this application are applicable to any suitable scenario.

[0017] In one embodiment, this application also proposes a method and apparatus for calculating the maximum pressure gradient of the coronary artery based on X-ray angiography images. Figure 1A schematic flowchart illustrating a method for calculating the maximum coronary artery pressure gradient based on X-ray angiography images according to an embodiment of this application is shown.

[0018] S101: Acquire multi-view coronary angiography images and perform preprocessing to generate preprocessed angiography images.

[0019] In one implementation, X-ray angiography sequences in DICOM format are acquired, and equipment parameters are recorded simultaneously. Clinicians use a digital subtraction angiography (DSA) machine to perform coronary angiography on patients suspected of having coronary artery disease. The left anterior descending coronary artery (LAD) is scanned from two projection angles: a 30° left anterior oblique angle and a 20° right anterior oblique angle. The equipment automatically generates a DICOM format image sequence. Thirty frames are acquired at each angle, with a resolution of 512×512 pixels and a grayscale value range of 0-4095 (12-bit depth) for each frame, serving as the angiography sequence information.

[0020] The synchronous recording equipment parameters are as follows: distance from radiation source to detector (SID) is 100cm; projection angle is left anterior oblique β = 30°, α = -15°; right anterior oblique β = 20°, α = 10°; pixel physical spacing is Δx = 0.2mm / pixel, Δy = 0.2mm / pixel; image size is width W = 102.4mm (512×0.2) and height H = 102.4mm.

[0021] A Gaussian filter with a 3×3 kernel and a standard deviation σ = 1.0 was used to denoise the original image. A 2×2 mean downsampling was performed on the 512×512 pixel image to generate a 256×256 pixel image, thus producing the preprocessed angiographic image. The original image was obtained directly from an unprocessed DICOM format X-ray angiography sequence after multi-angle scanning of the coronary arteries using a digital subtraction angiography (DSA) machine. The denoising formula is as follows: Where I(x,y) is the pixel value of the original image at coordinates (x,y), G(x,y) is the pixel value of the filtered image, the 3×3 kernel indicates that the filter's effective range on the image is a 3×3 neighborhood centered at (x,y), and σ=1.0 controls the decay rate of the Gaussian function (the smaller σ is, the weaker the filtering effect and the more image details are preserved; the larger σ is, the stronger the noise reduction effect, but the edges may be blurred).

[0022] Before filtering, the original image showed significant granular noise at the edges of coronary artery branches, low gray-level contrast between the blood vessels and surrounding tissues, and blurred outlines of small blood vessels (diameter <1mm). After filtering, Gaussian filtering significantly reduced granular noise and improved the smoothness of blood vessel edges. For example, in a 2mm diameter coronary artery segment, the standard deviation of pixel gray levels decreased from 25 to 12 after filtering, and the contrast between the blood vessel and the background increased by 15%, providing clearer input for subsequent blood vessel segmentation.

[0023] The filtered 512×512 pixel image is downsampled by 2×2 mean to generate a preprocessed 256×256 pixel image. During the downsampling process, the original image is divided into 2×2 non-overlapping pixel blocks, and the output pixel value of each block is the mean of the four pixels within the block.

[0024] Before downsampling, the filtered 512×512 image contains a large amount of redundant pixel information, resulting in a large data volume (approximately 512KB per image), which is detrimental to the rapid inference of subsequent deep learning models. After downsampling, a 256×256 pixel image is generated, reducing the data volume to 128KB and decreasing computational resource consumption. Simultaneously, mean downsampling preserves the overall morphological features of the blood vessels, such as the course of the main branches of the coronary arteries and the location of stenotic sites. In the downsampled image, blood vessel segments with an original diameter of 3 pixels can still be clearly identified without any breaks or distortion.

[0025] S102, based on the preprocessed angiography image, perform automatic screening of functional angiography frames, subtract the preprocessed image of each frame from the first background frame to obtain the difference image, calculate the grayscale variance of the difference image, and select the frame with the largest variance from the image sequence of each projection angle as the target angiography image of that viewpoint.

[0026] In one embodiment, the difference between each preprocessed image frame and the first background frame before contrast agent infusion is performed to eliminate interference from extravascular tissues and equipment noise, generating a difference image. The difference between each preprocessed angiography image frame and the first background frame before contrast agent infusion is performed to eliminate interference from extravascular tissues (such as bone and myocardium) and equipment noise, highlighting the vascular areas containing contrast agent. The first background frame contains no contrast agent, and the grayscale difference between the vascular area and surrounding tissue is small; in the angiography frame, the vessels are high grayscale due to contrast agent infusion, and the difference can preserve the significant difference between the vessels and the background. The background frame is an image of the initial stage of coronary angiography (before contrast agent injection), with a size of 256×256 pixels. The grayscale values ​​of the vascular area (such as the left anterior descending artery LAD) and myocardial tissue are close (approximately 150-200), and the noise standard deviation is approximately 30. The image taken 10 seconds after contrast agent injection shows that the grayscale value of the LAD vessel region rises to 350-400 due to the filling of contrast agent, while the grayscale value of the surrounding tissue remains at 150-200. Subtracting the corresponding pixel grayscale values, the difference in the vessel region is 150-250 (350-150=200), while the difference in the surrounding tissue is close to 0 (200-190=10). The resulting difference image shows clear vessel contours without tissue interference.

[0027] The variance of grayscale values ​​is calculated for the difference image. The variance formula is as follows: Among them, D n (x,y) represents the pixel grayscale value at coordinates (x,y) of the nth frame difference image, μ. n W×H represents the mean gray level of the difference image, and W×H represents the image size, such as 256×256. The larger the variance, the more significant the gray level difference between the blood vessel region and the background in the difference image, and the clearer the lesion information (such as contrast agent retention at the stenosis).

[0028] Taking a 256×256 pixel difference image as an example, calculate the grayscale mean μ. n The average gray level of the blood vessel region (approximately 10% of pixels) is 200, and the average gray level of the background region (90% of pixels) is 10. Therefore, μ n =0.1×200 + 0.9×10 = 29. Calculate the variance. n The contribution from the vascular region was 0.1 × (200-29). 2 =2924.1, with a background region contribution of 0.9×(10-29). 2 =324.9, Total Variance n =2924.1+234.9=3249. The variance of this frame is 3249. If the mean of the images in the same sequence is 2000, then 3249÷2000=1.62 (>1.5), which meets the condition of "prioritizing the screening of target angiography images", because high variance indicates that vascular lesions (such as stenosis) are clearly displayed.

[0029] For each projection angle image sequence, the frame with the largest variance is selected as the target angiographic image for that viewpoint. When the variance of a frame is greater than 1.5 times the sequence mean, it is prioritized as the target angiographic image to ensure that it contains the lesion of the target vessel. For each projection angle image sequence (e.g., 30 frames of left anterior oblique 30°), the frame with the largest variance is selected as the target angiographic image; if the variance of a frame is ≥ 1.5 times the sequence mean, it is prioritized. A patient's right coronary artery (RCA) angiography sequence contains 25 frames with a variance range of 1800-4500 and a sequence mean of 2500. The selection process is as follows: the 15th frame has a variance of 4200 (4200 ÷ 2500 = 1.68 > 1.5), which is the maximum value in the sequence; in this frame, the contrast filling defect at the mid-segment stenosis of the RCA is obvious (grayscale difference reaches 300), and the surrounding vessel branches are clear; it is determined to be the target angiographic image for this viewpoint and used for subsequent 3D reconstruction and pressure gradient calculation.

[0030] By eliminating non-vascular interference through interpolation and combining it with variance quantification of image information, objective screening of target angiographic frames can be achieved, avoiding the subjectivity of manual judgment. For example, in cases of calcified stenosis, high-variance frames can clearly show the grayscale difference between the stenotic segment and normal vessels, providing accurate anatomical basis for subsequent pressure gradient calculation, reducing the stenosis localization error from 5-10 mm in manual screening to 1-2 mm.

[0031] S103, the target angiography image is reconstructed into a three-dimensional vascular model and its coordinates are transformed to construct a three-dimensional network model of the coronary arteries.

[0032] In one implementation, pixel coordinates (u,v) are converted to physical coordinates (x,y) using the formulas x = (uW / 2) × Δx and y = (vH / 2) × Δy, where the origin is located at the image center. Converting image pixel coordinates (u,v) to actual physical coordinates (x,y) establishes a scale relationship between the two-dimensional image and three-dimensional space. (u,v) represents the image pixel coordinates (u is the horizontal pixel index, v is the vertical pixel index); W and H represent the image width and height (unit: pixels); Δx and Δy represent the pixel physical spacing (unit: mm / pixel), provided by the DSA device parameters; the origin (0,0) is located at the image center, i.e., the (W / 2,H / 2) pixel position.

[0033] Taking a left anterior oblique angiography image as an example: Image parameters: W = 256 pixels, H = 256 pixels, Δx = Δy = 0.2 mm / pixel. A certain blood vessel point is located in the image at (u = 150, v = 100); x = (150 - 256 / 2) × 0.2 = (150 - 128) × 0.2 = 22 × 0.2 = 4.4 mm; y = (100 - 256 / 2) × 0.2 = (100 - 128) × 0.2 = (-28) × 0.2 = -5.6 mm. The physical coordinates of this blood vessel point are (4.4 mm, -5.6 mm), indicating that it is located 4.4 mm to the right and 5.6 mm below the center of the image.

[0034] The Z-axis coordinate is initialized to the distance *d* from the human body to the detector, generating initial 3D coordinates (x, y, d). The 2D physical coordinates (x, y) are combined with depth information to generate the initial 3D coordinates (x, y, d), where *d* is the distance from the human body to the detector (unit: mm), directly provided by the DSA device (typically 1000-1200 mm). Specifically, with the device parameters set to detector distance *d* = 1000 mm, the initial 3D coordinates of the blood vessel point are (4.4 mm, -5.6 mm, 1000 mm).

[0035] The intermediate coordinates (x', y', z') are generated by rotating around the Y-axis, and the rotation matrix is: The conversion formula is Specifically, with the projection angle at the left front oblique position β = 30°, then cos30° ≈ 0.866, sin30° = 0.5. The rotation calculation process is as follows: x′ = x × cosβ + z × sinβ = 4.4 × 0.866 + 100 × 0.5 = 503.81 mm; y′ = y = -5.6 mm; z′ = -x × sinβ + z × cosβ = -4.4 × 0.5 + 1000 × 0.866 ≡ 863.8 mm. The intermediate coordinates are (503.81 mm, -5.6 mm, 863.8 mm).

[0036] The three-dimensional coordinates (X, Y, Z) in the world coordinate system are generated by rotating around the X-axis. The rotation matrix is: The conversion formula is Specifically, with the projection angle at the left front oblique position α = -15°, then cos(-15°) ≈ 0.966, sin(-15°) ≈ -0.259. The rotation calculation process is as follows: X = x′ = 503.81mm; Y = y′×cosα - z′×sinα = -5.6×0.966 - 863.8×(-0.259) ≈ 218.29mm; Z = y′×sinα + z′×cosα = -5.6×(-0.259) + 863.8×0.966 ≈ 835.85mm. The world coordinate system coordinates are (503.81mm, 218.29mm, 835.85mm).

[0037] The 3D coordinates of the reconstructed vessels from multiple perspectives were determined by grouping and fitting using the least squares method. For the 3D coordinates of the same vessel location under different perspectives, a set of spatial straight lines was constructed by connecting them, providing a geometric basis for subsequent target point fitting. The coordinates of the same vessel location (such as the midpoint of a stenotic segment of the coronary artery) were obtained from at least two perspectives, such as left anterior oblique and right anterior oblique. Each connecting line represents the spatial mapping of that location under different perspectives.

[0038] The multi-view coordinates are as follows: The vessel point coordinates A (500, 200, 800) mm are measured in the left anterior oblique view (LAO 30°); the corresponding vessel point coordinates B (490, 190, 820) mm are measured in the right anterior oblique view (RAO 20°); connecting A and B forms a spatial straight line L1, with direction vector... The coordinates S1 of the radiation source represent the position of the left anterior oblique radiation source (0,0,0) mm.

[0039] Connect the coordinates of points belonging to the same blood vessel location from different viewpoints, and solve for the target point P such that the sum of its distances to all connecting lines is minimized. The formula is as follows: Where P is the three-dimensional coordinate of the target point to be solved, and S i Let be the coordinates of the radiation source at the i-th viewpoint (determined by DSA equipment parameters). Let be the direction vector of the i-th connection, and let molecule be... The denominator is the magnitude of the cross product of vectors, representing the perpendicular distance from p to the line connecting them; The magnitude of the direction vector is used to standardize the distance.

[0040] Taking two perspectives as an example: Given the coordinates of the radiation source S1(0,0,0), the direction vector of the line L1 is... The coordinates of the radiation source are S2(1000,0,0) (right anterior oblique radiation source), and the direction vector of the line connecting them is L2. Let the target point be P(x,y,z), then cross product distance Similarly, calculate d2, and the total distance sum is d1+d2. Minimize d1+d2 through numerical iteration (such as gradient descent) to obtain the optimal solution P(495,195,810)mm; the sum of the distances from this point to L1 and L2 is minimized, with an error ≤0.5mm, which meets the clinical accuracy requirements.

[0041] All fitted vascular points (e.g., one point fitted every 1 mm) are connected into continuous line segments to form a three-dimensional mesh model, which fully recreates the spatial morphology of the vessel course, branches, and stenosis sites. Fitting was performed on the left anterior descending artery (LAD), resulting in 200 target points. The resulting three-dimensional model accurately reproduces the changes in vessel diameter (the diameter of the stenosis segment narrows from 3 mm to 1.5 mm); precisely captures the branch angles (the angle between the LAD and D1 branches is 75°); and has a spatial position error ≤1 mm, providing a precise anatomical basis for subsequent hemodynamic calculations (such as pressure gradient and flow velocity distribution).

[0042] By eliminating multi-view imaging errors through least-squares fitting, two-dimensional angiographic images are transformed into high-precision three-dimensional vascular models, solving problems such as vascular overlap and spatial distortion in traditional two-dimensional assessment. For example, for tandem stenosis lesions, the three-dimensional model can clearly distinguish the relative positions of the two stenosis lesions (15mm apart), providing reliable geometric parameters for calculating the pressure superposition effect caused by "lesion crosstalk," reducing the calculation error of MaxPPG20mm from 30%-40% in traditional methods to less than 5%, and completing the construction of a three-dimensional mesh model of the coronary arteries.

[0043] S104, calculates hemodynamic parameters based on a three-dimensional coronary artery network model.

[0044] In one implementation, for each cross-section of the 3D network model, the vessel contour is extracted using an edge detection algorithm, and the cross-sectional area is calculated. For each cross-section of the 3D vessel model, the vessel contour is extracted using an edge detection algorithm (Canny operator), and the cross-sectional area is calculated based on the contour, using the following formula: Where D is the diameter of the blood vessel cross-section (obtained directly from the 3D reconstruction results), and A is the cross-sectional area (unit: mm). 2 ).

[0045] In a 3D model, the diameter of a narrow section of the cross-section is D = 2.4 mm. What is the cross-sectional area? This parameter is used for subsequent blood flow and velocity calculations, reflecting the physical restriction of blood flow on vascular stenosis (e.g., the smaller the area of ​​the stenotic segment, the greater the blood flow resistance).

[0046] Blood flow was calculated based on contrast agent filling time and vessel volume, and the velocity distribution under laminar flow conditions was also calculated. Under laminar flow conditions, the velocity distribution within the vessel follows a parabolic law, where the velocity formula is v(r) = v max (1-r 2 / R 2 ),in, V(r) is the flow velocity at the radial coordinate r (unit: mm / s); V max R is the maximum flow velocity at the cross-section; R is the radius of the blood vessel, and r is the distance from a point to the center of the blood vessel (0 ≤ r ≤ R). Blood flow rate Q = 100 mm.3 / s, vessel radius R = 1.2mm, cross-sectional area A ≈ 4.52mm² 2 Maximum flow rate At the center of the blood vessel (r=0), the flow velocity V(0)=22.1×(1-0)=22.1mm / s; at the vessel wall (r=R=1.2mm), the flow velocity V(1.2)=22.1×(1-1)=0mm / s (meets the no-slip boundary condition).

[0047] The NS equations are discretized using the finite element method, and boundary conditions are set. The pressure distribution across the entire section is obtained through iterative solution. The NS equations are discretized using the finite element method as follows: in, The continuity equation indicates that blood is incompressible, u is the velocity vector, ρ is the blood density, μ is the dynamic viscosity coefficient, and f is the volume force; the boundary condition is set at the inlet: pressure p. in =100mmHg; Outlet: Flow rate Blood vessel wall: No slip boundary condition (u = 0). A 50mm long coronary artery segment was simulated, with a stenotic segment located in the middle 10mm region (diameter narrowed from 3mm to 1.5mm). After iterative solution, the proximal pressure of the stenotic segment was 95mmHg, the distal pressure was 70mmHg, and the pressure gradient was... The results reflect the significant pressure loss caused by the narrowing, providing a basis for subsequent PPG calculations.

[0048] A 20mm window is slid along the vessel axis, and the pressure gradient within the window is calculated. The maximum value is taken as the pressure loss intensity of the focal stenosis. If the proximal pressure within a certain window is 80 mmHg and the distal pressure is 55 mmHg, then the pressure gradient is... This value serves as an indicator of the intensity of pressure loss in focal stenosis.

[0049] The total length of continuous vascular segments with a pressure gradient slope ≥ 0.0015 mmHg / mm is defined as the length of the functional lesion. For example, if a vessel is 100 mm long and a 30 mm segment meets the slope condition, then the length of the functional lesion is 30 mm.

[0050] By using fluid dynamics modeling to accurately calculate coronary artery pressure distribution, the error problem of traditional empirical formulas can be solved. For example, for tandem stenosis lesions, the pressure superposition effect caused by "lesion crosstalk" can be captured, making the calculated MaxPPG20mm value 0.08FFRunits higher than that of traditional methods, which is closer to the clinical measurement results and provides quantitative evidence for the accurate diagnosis of coronary heart disease.

[0051] S105 constructs a PPG index based on hemodynamic parameters and classifies lesions. It integrates the 20mm maximum pressure gradient and the proportion of diffuse lesions to derive the PPG index formula. Through an inverse proportional weighting algorithm, it assigns a single PPG index value to a treatment strategy, generating classification information for coronary artery lesions.

[0052] In one implementation, a PPG index is constructed based on hemodynamic parameters and lesion classification is performed. The PPG index formula is derived by integrating the 20mm maximum pressure gradient and the proportion of diffuse lesions. The formula is as follows: Among them, MaxPPG 20mm The maximum pressure gradient (unit: FFRunits) is calculated by sliding a 20mm window along the vessel axis, reflecting the pressure loss intensity of focal stenosis; DiffuseRatio is the proportion of diffuse lesions (range 0-1), that is, the proportion of lesion length with a pressure gradient slope ≥0.0015FFRunits / mm to the total vessel length; through the inverse proportional structure of "focal intensity × (1-diffuse proportion)" and "diffuse proportion + correction term (0.1)", the nonlinear coupling between focal and diffuse lesions is realized.

[0053] Taking a patient with a left anterior descending coronary artery (LAD) lesion as an example, with a total vessel length of 120 mm, the following data were obtained after three-dimensional reconstruction and hemodynamic calculation: Maximum pressure gradient (MaxPPG) at a 20 mm sliding window. 20mm = 0.25FFRunits (corresponding to the pressure difference in the most severe stenosis area); the continuous lesion length with a pressure gradient slope ≥ 0.0015FFRunits / mm is 48mm (i.e., Length). diffuse =48mm).

[0054] Calculate MaxPPG 20mm The pressure gradient within each window is calculated by sliding the device along the blood vessel axis in 20mm increments. Length window =20mm.

[0055] The proximal pressure P of a certain 20mm window proximal = 0.90 FFR units, distal pressure P distal =0.65FFR units, therefore the pressure gradient of this window is 0.90 - 0.65 = 0.25FFR units, which is the maximum value among all windows. Calculate DiffuseRatio and substitute into the formula. Where DiffuseRatio represents the proportion of diffuse lesions, and Length represents the proportion of diffuse lesions. Diffuse The total length of a lesion that continuously meets the slope condition (pressure gradient varies with length ≥ 0.0015FFRunits / mm) reflects the extent of diffuse lesions.

[0056] Calculate PPG index Substitute into the formula

[0057] When PPG index When the value is <0.4, it is considered a diffuse lesion, and drug therapy or coronary artery bypass grafting (CABG) is the preferred treatment.

[0058] The patient's lesions were predominantly diffuse (DiffuseRatio = 0.4) and the PPG index was low, suggesting that unnecessary PCI (percutaneous coronary intervention) should be avoided, which could reduce the length of the stent implantation and lower the risk of perioperative myocardial infarction.

[0059] This application achieves automated and accurate calculation of maximum pressure gradient (PPG) using X-ray angiography image sequences. It improves the acquisition of multi-view DICOM format angiography sequences, records equipment parameters, and preprocesses them with Gaussian filtering for noise reduction and 2×2 mean downsampling. It automatically selects functionally optimal frames by generating a difference image from the first and background frames, calculating the grayscale variance, and selecting the frame with the largest variance. The target frame is reconstructed into a three-dimensional vascular model, which is then fitted using coordinate transformation and angle correction, combined with least squares fitting. Based on the three-dimensional model, hemodynamic parameters are calculated using the Navier-Stokes equations, and the maximum pressure gradient and functional lesion length within a 20mm window are extracted. A PPG index is constructed, fusing the above two parameters, and a corresponding treatment strategy is determined using an inverse proportional weighting algorithm. This solves the problems of low efficiency and inaccurate pressure data acquisition associated with manual screening, improving efficiency and accuracy, providing support for clinical decision-making, and is applicable to the assessment of functional lesions in coronary artery disease.

[0060] In one implementation, such as Figure 2 As shown, this application also provides a device for calculating the maximum coronary artery pressure gradient based on X-ray angiography images, comprising:

[0061] The acquisition module 201 is used to acquire multi-view coronary angiography images and perform preprocessing to generate preprocessed angiography images;

[0062] The processing module 202 is used for automated selection of functional angiography frames based on preprocessed angiography images. It calculates the difference image by subtracting each preprocessed image from the first background image, calculates the grayscale variance of the difference image, and selects the frame with the largest variance from the image sequence of each projection angle as the target angiography image for that viewpoint. It then performs three-dimensional vascular model reconstruction and coordinate transformation on the target angiography image to construct a three-dimensional coronary artery network model. Based on the three-dimensional coronary artery network model, it calculates hemodynamic parameters. Based on the hemodynamic parameters, it constructs the PPG index and classifies lesions. It integrates the 20mm maximum pressure gradient and the proportion of diffuse lesions to derive the PPG index formula. Through an inverse proportional weighting algorithm, it assigns a single PPG index value to a treatment strategy, generating classification information for coronary artery lesions.

[0063] The computer-readable storage medium provided in the above embodiments of this application and the method for calculating the maximum coronary artery pressure gradient based on X-ray angiography images provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods used, run or implemented by the applications stored therein.

[0064] The various embodiments in this application are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for evaluating the method, electronic device, electronic device, and readable storage medium for calculating the maximum coronary artery pressure gradient based on X-ray angiography images are basically similar to the embodiments of the method for calculating the maximum coronary artery pressure gradient based on X-ray angiography images described above, and are therefore described simply. Relevant parts can be referred to in the descriptions of the embodiments of the method for calculating the maximum coronary artery pressure gradient based on X-ray angiography images described above.

Claims

1. A method for calculating the maximum pressure gradient of coronary arteries based on X-ray angiography images, characterized in that, include: Acquire multi-view coronary angiography images and perform preprocessing to generate preprocessed angiography images; Based on the preprocessed angiography images, the functional angiography frames are automatically screened. The difference image is obtained by subtracting the preprocessed image from the first background image. The gray scale variance is calculated on the difference image. For each projection angle image sequence, the frame with the largest variance is selected as the target angiography image for that viewpoint. The target angiographic image is reconstructed into a three-dimensional vascular model and its coordinates are transformed to construct a three-dimensional network model of the coronary arteries. Hemodynamic parameters were calculated based on a three-dimensional coronary artery network model. The PPG index is constructed based on hemodynamic parameters and lesion classification is performed. The PPG index formula is derived by integrating the 20mm maximum pressure gradient and the proportion of diffuse lesions. The treatment strategy is assigned to a single PPG index value through an inverse proportional weighting algorithm, thereby generating classification information for coronary artery lesions.

2. The method as described in claim 1, characterized in that, Acquire multi-view coronary angiography images and perform preprocessing to generate preprocessed angiography images, including: Acquire X-ray contrast sequences in DICOM format and record equipment parameters simultaneously; The original image was denoised using a Gaussian filter with a 3×3 kernel and a standard deviation σ = 1.

0. The 512×512 pixel image was downsampled by 2×2 mean to generate a 256×256 pixel image, which was then used to generate the preprocessed angiography image. The original image was an unprocessed DICOM format X-ray angiography sequence directly obtained after multi-angle scanning of the coronary arteries by a digital subtraction angiography machine. The noise reduction formula is Where I(x,y) is the gray value of the original image pixel, G(x,y) is the filtered image, and σ is the standard deviation of the Gaussian filter.

3. The method as described in claim 2, characterized in that, Automated selection of functional angiography frames is performed based on preprocessed angiography images. The difference between each preprocessed image and the first background image is calculated to obtain a difference image. The grayscale variance of the difference image is calculated, and the frame with the largest variance is selected as the target angiography image for each projection angle, including: The difference between each preprocessed image frame and the first frame of the background frame where the contrast agent is not fully filled is calculated to eliminate noise interference from extravascular tissues and equipment, and a difference image is generated. The variance of grayscale values ​​is calculated for the difference image. The variance formula is as follows: Among them, Variance n represents the grayscale variance of the nth frame difference image, used to quantify the grayscale difference between the blood vessel region and the background in the difference image; W×H represents the image size, i.e., the product of the image width W and height H; D n (x,y) represents the pixel grayscale value at coordinates (x,y) of the nth frame difference image; μ n The mean gray level of the difference image; For each projection angle, the frame with the largest variance in the image sequence is selected as the target angiography image for that viewpoint. When the variance of a frame is more than 1.5 times the mean of the sequence, it is preferentially selected as the target angiography image so that it can display the lesion of the target blood vessel.

4. The method as described in claim 1, characterized in that, The target angiographic image is subjected to 3D vascular model reconstruction and coordinate transformation to construct a 3D coronary artery network model, including: The pixel coordinates (u,v) are converted into physical coordinates (x,y) using the formulas x=(uW / 2)×Δx and y=(vH / 2)×Δy, where the origin is located at the center of the image. The Z-axis coordinate is initialized to the distance d from the human body to the detector, generating the initial three-dimensional coordinates (x, y, d); The intermediate coordinates (x', y', z') are generated by rotating around the Y-axis, and the rotation matrix is: The conversion formula is The three-dimensional coordinates (X, Y, Z) in the world coordinate system are generated by rotating around the X-axis. The rotation matrix is: The conversion formula is The three-dimensional coordinates of blood vessels were determined by least squares group fitting for multi-view reconstruction. Connect the coordinates of points belonging to the same blood vessel location from different viewpoints, and solve for the target point P such that the sum of its distances to all connecting lines is minimized. The formula is as follows: Where P is the target point. The coordinates of the radiation source, Using the direction vectors of the connecting lines, a three-dimensional mesh model of the coronary arteries is constructed.

5. The method as described in claim 4, characterized in that, Hemodynamic parameters were calculated based on a three-dimensional coronary artery network model, including: For each cross section of the 3D network model, the blood vessel contour is extracted using an edge detection algorithm, and the cross-sectional area is calculated. Blood flow rate is calculated based on contrast agent filling time and vessel volume, and the velocity distribution under laminar flow conditions is also calculated. The velocity formula is v(r) = v max (1-r 2 / R 2 ), where R is the radius of the blood vessel and r is the radial coordinate; The NS equations are discretized using the finite element method, and boundary conditions are set. The pressure distribution across the entire section is obtained through iterative solution. The NS equations are discretized using the finite element method as follows: Where u is the flow velocity vector, ρ is the blood density, μ is the dynamic viscosity coefficient, and f is the volume force; boundary conditions are set at the inlet: pressure p. in =100mmHg; Outlet: Flow rate Cross-sectional average velocity Blood vessel wall: No slip boundary condition (u = 0); Slide a 20mm window along the vessel axis, calculate the pressure gradient within the window, and take the maximum value as the pressure loss intensity of the focal stenosis; The total length of continuous vascular segments with a pressure gradient slope ≥ 0.0015 mmHg / mm is defined as the length of functional lesions.

6. The method as described in claim 5, characterized in that, A PPG index is constructed based on hemodynamic parameters for lesion classification. The PPG index formula is derived by integrating the 20mm maximum pressure gradient and the proportion of diffuse lesions. An inverse proportional weighting algorithm is used to assign a single PPG index value to a corresponding treatment strategy, generating classification information for coronary artery lesions, including: The PPG index was constructed based on hemodynamic parameters and lesion classification was performed. The PPG index formula was derived by integrating the 20mm maximum pressure gradient and the proportion of diffuse lesions. The formula is as follows: Length window =20mm, where MaxPPG 20mm for 20mm maximum pressure gradient, p proimal and p distal The length represents the proximal and distal pressure values ​​within the window. window To calculate the length of the window used, PPG index This is the final calculated PPG index; Where DiffuseRatio represents the proportion of diffuse lesions, and Length represents the proportion of diffuse lesions. Diffuse The total length of lesions that continuously meet the slope condition reflects the extent of diffuse lesions; pass The inverse proportional weighting algorithm assigns a single PPG index value to a treatment strategy, generating classification information for coronary artery lesions.

7. A device for calculating the maximum pressure gradient of coronary arteries based on X-ray angiography images, characterized in that, The device includes: The acquisition module is used to acquire multi-view coronary angiography images and perform preprocessing to generate preprocessed angiography images; The processing module is used for automated selection of functional angiography frames based on preprocessed angiography images. It calculates the difference image by subtracting each preprocessed image from the first background frame, calculates the grayscale variance of the difference image, and selects the frame with the largest variance from each projection angle as the target angiography image for that viewpoint. The target angiography image is then used for 3D vascular model reconstruction and coordinate transformation to construct a 3D coronary artery network model. Hemodynamic parameters are calculated based on the 3D coronary artery network model. A PPG index is constructed based on the hemodynamic parameters, and lesion classification is performed. The PPG index formula is derived by integrating the 20mm maximum pressure gradient and the proportion of diffuse lesions. An inverse proportional weighting algorithm is used to assign a single PPG index value to a treatment strategy, generating classification information for coronary artery lesions.

8. An electronic device, characterized in that, include: First processor; and memory for storing executable instructions of the first processor; The first processor is configured to execute the coronary artery maximum pressure gradient calculation method based on X-ray angiography images according to any one of claims 1 to 6 by executing the executable instructions.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the second processor, it implements the method for calculating the maximum coronary artery pressure gradient based on X-ray angiography images as described in any one of claims 1 to 6.

Images