Vein blood vessel image processing method and device, equipment and storage medium

Abstract

The invention discloses a vein blood vessel image processing method and device, equipment and a storage medium. The vein blood vessel image processing method comprises the steps of collecting a vein blood vessel image; the collected vein blood vessel image is preprocessed; an improved U-Net segmentation network and an improved Zhang-Suen skeletonization algorithm are adopted to process the preprocessed image, and multi-modal results are fused to obtain a blood vessel segmentation result; and based on a blood vessel segmentation result, performing blood vessel three-dimensional modeling by adopting a multi-stage B-spline mapping method, and quantitatively analyzing blood vessel spatial geometric features and hemodynamic parameters to obtain a blood vessel form-function joint evaluation index. The vein blood vessel image is processed and analyzed, three-dimensional modeling is carried out on the blood vessel, the radiation path of radiotherapy is planned, and a foundation is laid for follow-up treatment of varicosity through in-vitro radiation.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical image processing, and in particular relates to a venous blood vessel image processing method, device, equipment and storage medium. Background Art

[0002] Varicose veins of the lower extremities are a common condition, particularly among those who engage in sustained physical labor or standing work. The main symptoms are dilation, elongation, and tortuosity of the great saphenous vein in the lower extremities, resulting in soreness, fatigue, and heaviness in the affected limbs. Severe cases are often accompanied by complications such as calf ulcers or superficial phlebitis. Sufferers are often athletes, teachers, and other people who stand for long periods of time. When standing, gravity pulls downward, and blood must return to the heart from the farthest point. If venous insufficiency occurs, the veins will dilate and become varicose, leading to venous hypertension in the lower extremities. If varicose veins of the lower extremities are not treated promptly, they may lead to serious consequences such as calf ulcers and venous thrombosis.

[0003] Radiotherapy can close dilated varicose veins. However, effective venous image processing is required during radiotherapy. Existing methods cannot accurately analyze the vessels, resulting in poor radiotherapy accuracy and ineffective treatment results. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention proposes a venous blood vessel image processing method, device, equipment and storage medium.

[0005] In order to achieve the above object, the technical solution of the present invention is as follows:

[0006] In a first aspect, the present invention discloses a method for processing a venous blood vessel image, comprising:

[0007] Step S1: Acquire venous blood vessel images;

[0008] Step S2: pre-processing the acquired venous blood vessel image;

[0009] Step S3: using the improved U-Net segmentation network and the improved Zhang-Suen skeletonization algorithm to process the preprocessed image and obtain the blood vessel segmentation result;

[0010] Step S4: Based on the vascular segmentation results, a multi-level B-spline mapping method is used to perform 3D vascular modeling, quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters, and obtain a vascular morphology-function joint evaluation index.

[0011] On the basis of the above technical solution, the following improvements can be made:

[0012] As a preferred solution, step S2 includes:

[0013] Step S2.1: performing sliding window median filtering on the acquired venous blood vessel image;

[0014] Step S2.2: performing CLAHE contrast enhancement on the image after median filtering;

[0015] Step S2.3: Perform Laplacian sharpening on the image after CLAHE contrast enhancement processing.

[0016] As a preferred solution, step S3 includes:

[0017] Step S3.1: Construct an improved U-Net segmentation network, train it using a composite loss function and a dynamic data augmentation strategy, and use the trained U-Net segmentation network to segment the preprocessed image to obtain the initial blood vessel segmentation result;

[0018] Step S3.2: Perform improved Zhang-Suen skeletonization on the initial blood vessel segmentation result to obtain a refined blood vessel segmentation result.

[0019] As a preferred solution, step S4 includes:

[0020] Step S4.1: converting the blood vessel segmentation result in the image coordinate system to the world coordinate system;

[0021] Step S4.2: Perform three-dimensional reconstruction of the vascular centerline in the world coordinate system to generate a topologically continuous vascular centerline model;

[0022] Step S4.3: Constructing a vascular surface geometric model based on the vascular axis and radius distribution;

[0023] Step S4.4: Based on the vascular surface geometric model, quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters to obtain a vascular morphology-function joint evaluation index.

[0024] In a second aspect, the present invention discloses a venous blood vessel image processing device, comprising:

[0025] An acquisition module, used for acquiring venous blood vessel images;

[0026] A preprocessing module, used for preprocessing the acquired venous blood vessel images;

[0027] The segmentation module is used to process the preprocessed image using the improved U-Net segmentation network and the improved Zhang-Suen skeletonization algorithm to obtain the blood vessel segmentation result;

[0028] The analysis module is used to perform 3D vascular modeling based on the vascular segmentation results using a multi-level B-spline mapping method, quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters, and obtain vascular morphology-function joint evaluation indicators.

[0029] As a preferred solution, the preprocessing module includes:

[0030] A first pre-processing unit is used to perform sliding window median filtering on the acquired venous blood vessel image;

[0031] The second pre-processing unit is used to perform CLAHE contrast enhancement processing on the image after the median filtering processing;

[0032] The third pre-processing unit is used to perform Laplace sharpening processing on the image after the CLAHE contrast enhancement processing.

[0033] As a preferred solution, the segmentation module includes:

[0034] The first segmentation unit is used to construct an improved U-Net segmentation network, which is trained using a composite loss function and a dynamic data enhancement strategy. The trained U-Net segmentation network is used to segment the preprocessed image to obtain the initial blood vessel segmentation result.

[0035] The second segmentation unit is used to perform improved Zhang-Suen skeletonization processing on the initial blood vessel segmentation result to obtain a refined blood vessel segmentation result.

[0036] As a preferred solution, the analysis module includes:

[0037] A first analysis unit is used to convert the blood vessel segmentation result in the image coordinate system into the world coordinate system;

[0038] The second analysis unit is used to perform three-dimensional reconstruction of the blood vessel centerline in the world coordinate system to generate a topologically continuous blood vessel centerline model;

[0039] The third analysis unit is used to construct a blood vessel surface geometric model based on the blood vessel central axis and radius distribution;

[0040] The fourth analysis unit is used to quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters based on the vascular surface geometric model to obtain a vascular morphology-function joint evaluation index.

[0041] In a third aspect, the present invention discloses a computing device, comprising:

[0042] one or more processors;

[0043] Memory;

[0044] and one or more programs, wherein the one or more programs are stored in a memory and configured to be executed by one or more processors, and the one or more programs include instructions of any of the above-mentioned venous vessel image processing methods.

[0045] In a fourth aspect, the present invention discloses a storage medium storing one or more computer-readable programs, wherein the one or more programs include instructions suitable for being loaded by a memory and executing any of the above-mentioned venous vessel image processing methods.

[0046] The present invention discloses a venous blood vessel image processing method, device, equipment and storage medium, which have the following beneficial effects:

[0047] First, the present invention processes and analyzes venous vessel images, performs three-dimensional modeling of the vessels, and plans the irradiation path of radiotherapy, laying the foundation for the subsequent use of external irradiation to treat varicose veins.

[0048] Second, the present invention uses pre-processing such as median filtering, CLAHE contrast enhancement, and Laplace sharpening to improve the blood vessel-background contrast and signal-to-noise ratio.

[0049] Third, the present invention combines the improved U-Net segmentation network and the improved Zhang-Suen skeletonization to achieve blood vessel segmentation with high segmentation accuracy.

[0050] Fourth, the present invention adopts a multi-level B-spline vascular mapping method for vascular modeling and analysis, and the analysis results are more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 This is a flow chart of a venous blood vessel image processing method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0055] The use of ordinal numbers "first," "second," "third," etc. to describe common objects merely indicates that different instances of similar objects are involved and is not intended to imply that the objects so described must have a given order in time, space, ranking, or in any other manner.

[0056] In addition, the expression of “comprising” an element is an “open” expression, which merely means that corresponding components or steps exist, and should not be interpreted as excluding additional components or steps.

[0057] In order to achieve the purpose of the present invention, some embodiments of the venous blood vessel image processing method, such as Figure 1 As shown, the venous blood vessel image processing method includes:

[0058] Step S101: Acquire venous blood vessel images;

[0059] Step S102: pre-processing the acquired venous blood vessel image;

[0060] Step S103: using an improved U-Net segmentation network and an improved Zhang-Suen skeletonization algorithm to process the preprocessed image to obtain a blood vessel segmentation result;

[0061] Step S104: Based on the vascular segmentation results, a multi-level B-spline mapping method is used to perform 3D vascular modeling, and the vascular spatial geometric characteristics and hemodynamic parameters are quantitatively analyzed to obtain a vascular morphology-function joint evaluation index.

[0062] The venous blood vessel image processing method of the present invention is used to process the collected venous blood vessel image, perform three-dimensional modeling on the blood vessel, and automatically generate an irradiation path based on this.

[0063] The above steps are described in detail below.

[0064] Step S101 is the image acquisition step. To ensure accurate subsequent radiotherapy, the patient adopts a comfortable, reproducible, and varicose vein treatment position, secured with a body-fixing device such as a thermoplastic body wrap, vacuum pad, or body frame. In this position, a near-infrared light source and infrared-sensitive camera are used to capture venous images.

[0065] Specifically, a near-infrared light source with a wavelength of 750-1100nm is used to illuminate the target skin area. Hemoglobin's absorption rate of light in this wavelength band is 20%-40% higher than that of surrounding tissue, resulting in a light absorption difference between blood vessels and the background. An infrared-sensitive camera (preferably an InGaAs sensor with a quantum efficiency >80%) is configured to receive the reflected light signal. The imaging resolution is set to 640×480 pixels or above, and the frame rate is not less than 30fps. An optical filter is used to filter out interference from the visible light band to obtain the original infrared blood vessel image, with an image grayscale range of 0-255.

[0066] Step S102 is a pre-processing step, which includes:

[0067] Step S102.1: performing sliding window median filtering on the acquired venous blood vessel image;

[0068] Step S102.2: performing CLAHE contrast enhancement on the median filtered image;

[0069] Step S102.3: Perform Laplacian sharpening on the image after the CLAHE contrast enhancement process.

[0070] Step S102.1 performs median filtering and denoising, specifically including:

[0071] 1) Construct a 3×3 or 5×5 two-dimensional convolution kernel, preferably using an odd-sized kernel to maintain image symmetry.

[0072] 2) Perform a sliding window traversal on the original image. Within each window:

[0073] Extract the grayscale values of all pixels in the window;

[0074] After sorting in ascending order, select the median value;

[0075] Replace the center pixel value of the window with the median value.

[0076] 3) Iteratively process the entire image to remove salt and pepper noise while preserving the sharpness of blood vessel edges (edge preservation rate > 90%).

[0077] Step S102.2 performs CLAHE contrast enhancement, specifically including:

[0078] 1) Divide the image into 8×8 local sub-regions, with the size of each sub-region being 1 / 8 of the image width and height.

[0079] 2) Perform histogram equalization on each sub-region independently:

[0080] Calculate the grayscale histogram distribution of the sub-region;

[0081] Set the contrast limit threshold CL = 2.0-3.0 (specifically: 2.5);

[0082] Pixels are redistributed for the portion of the histogram that exceeds the CL.

[0083] 3) Bilinear interpolation is used to eliminate grayscale jumps at the sub-region boundaries and generate a globally balanced image.

[0084] Step S102.3 performs Laplace sharpening processing, which specifically includes:

[0085] 1) Construct sharpening convolution kernel: [0,-1,0

[0086] -1,5,-1

[0087] 0,-1,0]

[0088] 2) Perform convolution operation on the image processed by CLAHE. The calculation formula is:

[0089] G(x,y)=5*I(x,y)-∑[I(x+i,y+j)];

[0090] in:

[0091] x,y are the horizontal and vertical coordinates of the image;

[0092] i, j represents the offset of the horizontal and vertical coordinates of the neighborhood centered on the current pixel in the image, and can take three values: -1, 0, and 1 respectively;

[0093] i,j∈{-1,0,1}.

[0094] 3) The convolution result is fused with the original image at a weight of 1:0.8 to enhance the second-order differential features of the vascular texture and improve the clarity of the vascular edge by 40%-60%.

[0095] The processed image is grayscale normalized and the pixel values are mapped to the standard range of 0-255 to generate the final vascular enhanced image, whose vascular-background contrast is 3-5 times higher than that of the original image and the signal-to-noise ratio (SNR) is ≥25dB.

[0096] Step S103 is a step of boundary extraction and blood vessel segmentation, which includes:

[0097] Step S103.1: Construct an improved U-Net segmentation network, train it using a composite loss function and a dynamic data augmentation strategy, and use the trained U-Net segmentation network to segment the preprocessed image to obtain an initial blood vessel segmentation result;

[0098] Step S103.2: Performing improved Zhang-Suen skeletonization on the initial blood vessel segmentation result to obtain a refined blood vessel segmentation result.

[0099] The improved U-Net segmentation network is as follows:

[0100] The encoder consists of 4 stages, each of which uses a combination of 3×3 convolution (stride 1) + batch normalization + ReLU activation, followed by 2×2 max pooling (stride 2);

[0101] The decoder sets up 4 upsampling stages and uses transposed convolution (kernel size 2×2, stride 2) to restore the resolution;

[0102] The skip connection adopts channel splicing to fuse the feature map of the encoder layer n with the feature map of the decoder layer (4-n) in the channel dimension.

[0103] The final output layer uses 1×1 convolution + Sigmoid activation to generate pixel-level binary segmentation masks.

[0104] Use composite loss function: L = α*Dice Loss + β*BCE Loss;

[0105] For example, α can be 0.6 and β can be 0.4.

[0106] The Dice Loss calculation formula is:

[0107]

[0108] The BCE Loss calculation formula is:

[0109]

[0110] in:

[0111] P represents the network prediction result, G represents the true label, and ε is a small constant to prevent the denominator from dividing by zero.

[0112] At the same time, edge-sensitive weights are introduced, and a boundary weight map is generated through distance transformation. Loss weights are assigned to the n-pixel range of the blood vessel boundary. The weight loss calculation formula is:

[0113]

[0114] The final loss is

[0115] Where: x, y represent the coordinates of the image pixel points, σ represents the variance, D(.) represents the Euclidean distance function, and W(.) represents the final generated distance weight map.

[0116] The training strategy is as follows:

[0117] The input image size is normalized to 512×512 pixels and the batch size is set to 8;

[0118] Adam optimizer is used (initial learning rate 3e-4, β1 = 0.9, β2 = 0.999);

[0119] Dynamic data augmentation is implemented: random rotation (±15°), grayscale jitter (±20%), and elastic deformation (σ=8, α=20).

[0120] The improved Zhang-Suen skeletonization specifically includes the following contents.

[0121] First, iterative refinement process:

[0122] Phase 1: Traverse all foreground pixels, and when:

[0123] ①2≤N(p1)≤6 (N(p1) is the number of foreground points in the 8-neighborhood);

[0124] ②Z0(p1)=1(number of 0→1 mode transitions in 8 neighborhoods);

[0125] ③p2p4p6=0;

[0126] ④p4p6p8=0;

[0127] The point is marked for deletion.

[0128] Phase 2: The traversal conditions are adjusted to:

[0129] ③p2p4p8=0;

[0130] ④p2p6p8=0;

[0131] The remaining conditions are the same as those in stage 1.

[0132] Secondly, topology correction:

[0133] 1) Detect and repair skeleton breaks: If the distance between adjacent endpoints is ≤ 3 pixels, use cubic B-spline interpolation to connect them;

[0134] 2) Remove burr branches: Calculate the ratio of each branch length to the trunk length and delete short branches with a ratio < 0.12.

[0135] Finally, the skeleton is smoothed:

[0136] The moving least squares (MLS) method was applied to fit the curve, with a control point spacing of 5 pixels;

[0137] Curvature constraint: forces the angle between adjacent line segments to be ≥30° to eliminate unnatural bends.

[0138] The Zhang-Suen skeleton can refine the unet segmentation results and make the blood vessel segmentation results better.

[0139] Step S104 is a three-dimensional modeling and blood vessel analysis step, including:

[0140] Step S104.1: converting the blood vessel segmentation result in the image coordinate system to the world coordinate system;

[0141] Step S104.2: Perform three-dimensional reconstruction of the vascular centerline in the world coordinate system to generate a topologically continuous vascular centerline model;

[0142] Step S104.3: Constructing a vascular surface geometric model based on the vascular centerline and radius distribution;

[0143] Step S104.4: Based on the vascular surface geometric model, quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters to obtain a vascular morphology-function joint evaluation index.

[0144] Step S104.1 is coordinate system conversion, which specifically includes the following contents.

[0145] Establish the mapping relationship between the world coordinate system OxyzOxyz and the image coordinate system OuvOuv:

[0146]

[0147] in:

[0148] u, v are the pixel coordinates in the image coordinate system;

[0149] Δ x ,Δ y are the physical sizes of pixels respectively;

[0150] α k is the depth weight coefficient;

[0151] d k (s) is the displacement function of the sequence image.

[0152] Step S104.2 performs three-dimensional reconstruction of the central axis, which specifically includes the following contents.

[0153] First, the skeleton point sequence is parameterized:

[0154] Input 2D skeleton point set {pi = (ui, vi, ri)}, where ri is the local vessel radius;

[0155] Use cubic uniform B-spline curve to fit the spatial path:

[0156]

[0157] in:

[0158] B j,3 (t) is the basis function, the control point P j Determined by least squares optimization.

[0159] Secondly, perform spatial posture correction:

[0160] Calculate the curvature of three adjacent points:

[0161]

[0162] When κ>0.15mm -1 New control points are inserted when the curve is drawn to ensure curvature continuity.

[0163] Step S104.3 performs blood vessel surface modeling, which specifically includes the following contents.

[0164] First, radius adaptive lofting:

[0165] Generate a normal section circle along the central axis, with the radius gradually changing according to r(s) = r0 + β·s (β is the radius change rate, ranging from -0.05 to 0.03 / mm);

[0166] Adjacent sections are joined using Coons surfaces, and the parametric equation is:

[0167] S(u,v)=(1-u)C0(v)+uC1(v)+(1-v)D0(u)+vD1(u)-(1-u)(1-v)P00-..

[0168] Secondly, topology optimization:

[0169] Bifurcation point processing:

[0170] a) Detection of radius mutation point (Δr / r>40%);

[0171] b) Establish a NURBS transition surface connecting the mother and child vessels with a curvature continuity of G2;

[0172] Mesh refinement: Local subdivision is performed in areas with curvature > 0.1 mm-1, with a mesh size of < 0.05 mm.

[0173] Step S104.4 performs quantitative analysis of blood vessel characteristics, specifically including the following contents.

[0174] First, spatial geometric feature extraction:

[0175] Direction vector calculation:

[0176] Main direction:

[0177]

[0178] Local Direction:

[0179] Field point set.

[0180] Radius change rate analysis:

[0181] Compute the radial gradient along the centerline:

[0182]

[0183] Marking narrow areas

[0184] Secondly, hemodynamic parameters are calculated:

[0185] Poiseuille flow approximation calculation:

[0186] Volume flow rate:

[0187]

[0188] Where: μ is blood viscosity;

[0189] Q is the volume flow rate (volume of liquid flowing per unit time);

[0190] r is the inner radius of the pipe;

[0191] Δp is the pressure difference between the two ends of the pipe;

[0192] L is the length of the pipeline.

[0193] Wall shear stress:

[0194]

[0195] Varicose vein patients often experience mid-tibial hyperpigmentation, superficial phlebitis, venous ulcers, and venous rupture and bleeding in the lower extremities. By analyzing acquired venous images and creating three-dimensional vascular models, varicose veins can be treated with external beam radiotherapy. Targeted irradiation using low-energy electron beams / microfocus X-rays selectively destroys venous endothelial cells, inducing vascular fibrosis and closure.

[0196] Specifically, in some other embodiments, the present invention further includes:

[0197] Step S105: using the three-dimensional blood vessel model to plan the irradiation path, specifically including the following contents.

[0198] First, the 3D vascular model is discretized into a voxel grid.

[0199] The resolution is set to 0.2×0.2×0.2mm 3 , voxel attributes include:

[0200] Vascular curvature κ (0-0.5 mm -1 );

[0201] Distance to the danger zone D (0-10mm);

[0202] Energy deposition efficiency η (0.5-1.2);

[0203] Secondly, establish an octree spatial index.

[0204] The maximum depth is 18 layers, and the nodes include the minimum bounding box and path weight;

[0205] Weight calculation:

[0206] W=1+3e -2D +5|κ-0.25|

[0207] Then, design the path cost function.

[0208] The node evaluation function is:

[0209] f(n)=g(n)+h(n)+λ·p(n)

[0210] in:

[0211] g(n): actual cost (cumulative path length × energy attenuation coefficient);

[0212] h(n): heuristic cost (Euclidean distance to the end point);

[0213] p(n): Penalty term (curvature penalty + safety distance penalty).

[0214] Dynamic weight adjustment mechanism:

[0215] Curvature penalty:

[0216]

[0217] Safety distance penalty:

[0218] p d =50·e -0.5D

[0219] Then, the path search process is as follows.

[0220] Initialization parameters:

[0221] Laser parameters: spot diameter Φ = 0.6 mm, energy density E = 15 J / mm 2 ;

[0222] The ablation overlap rate must be ≥30% (corresponding to a step length of ≤0.42 mm);

[0223] The improved A* algorithm is used to perform the following steps:

[0224] a) Open list priority queue is sorted by f value;

[0225] b) 26-neighborhood expansion strategy, calculate the reachability in each direction:

[0226] Axial movement cost coefficient is 1.0;

[0227] Diagonal movement cost coefficient √3;

[0228] c) Real-time energy accumulation detection:

[0229]

[0230] Where: A_{overlap} is the area of the light spot overlap region.

[0231] d) Termination condition: reaching the end point and E_{acc}≥preset treatment threshold. Finally, the path is smoothly optimized.

[0232] B-spline curve fitting:

[0233] The control point spacing is 0.5 mm, and the order k = 3;

[0234] Minimize the rate of change of curvature:

[0235]

[0236] Energy density verification:

[0237] The energy distribution is calculated at every 0.1mm sampling point along the path:

[0238]

[0239] Where: σ=Φ / 2√2, N is the number of light spots.

[0240] Ensure that the target area E ≥ 40J / mm 3 , normal tissue E≤5J / mm 3 .

[0241] Furthermore, the planned path can be projected onto the patient's skin in the form of visible light through a projection device to assist the doctor in positioning and operation.

[0242] The present invention utilizes an infrared camera to automatically extract the boundaries of varicose veins and plan the irradiation route.

[0243] In some other embodiments, the present invention discloses a venous blood vessel image processing device, comprising:

[0244] An acquisition module, used for acquiring venous blood vessel images;

[0245] A preprocessing module, used for preprocessing the acquired venous blood vessel images;

[0246] The segmentation module is used to process the preprocessed image using the improved U-Net segmentation network and the improved Zhang-Suen skeletonization algorithm to obtain the blood vessel segmentation result;

[0247] The analysis module is used to perform 3D vascular modeling based on the vascular segmentation results using a multi-level B-spline mapping method, quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters, and obtain vascular morphology-function joint evaluation indicators.

[0248] Furthermore, the preprocessing module includes:

[0249] A first pre-processing unit is used to perform sliding window median filtering on the acquired venous blood vessel image;

[0250] The second pre-processing unit is used to perform CLAHE contrast enhancement processing on the image after the median filtering processing;

[0251] The third pre-processing unit is used to perform Laplace sharpening processing on the image after the CLAHE contrast enhancement processing.

[0252] Furthermore, the segmentation module includes:

[0253] The first segmentation unit is used to construct an improved U-Net segmentation network, which is trained using a composite loss function and a dynamic data enhancement strategy. The trained U-Net segmentation network is used to segment the preprocessed image to obtain the initial blood vessel segmentation result.

[0254] The second segmentation unit is used to perform improved Zhang-Suen skeletonization processing on the initial blood vessel segmentation result to obtain a refined blood vessel segmentation result.

[0255] Furthermore, the analysis module includes:

[0256] A first analysis unit is used to convert the blood vessel segmentation result in the image coordinate system into the world coordinate system;

[0257] The second analysis unit is used to perform three-dimensional reconstruction of the blood vessel centerline in the world coordinate system to generate a topologically continuous blood vessel centerline model;

[0258] The third analysis unit is used to construct a blood vessel surface geometric model based on the blood vessel central axis and radius distribution;

[0259] The fourth analysis unit is used to quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters based on the vascular surface geometric model to obtain a vascular morphology-function joint evaluation index.

[0260] In some other embodiments, the present invention discloses a computing device comprising:

[0261] one or more processors;

[0262] Memory;

[0263] and one or more programs, wherein the one or more programs are stored in a memory and configured to be executed by one or more processors, and the one or more programs include instructions of any of the above-mentioned venous vessel image processing methods.

[0264] In some other embodiments, the present invention discloses a storage medium storing one or more computer-readable programs, wherein the one or more programs include instructions suitable for being loaded by a memory and executing any of the above-mentioned venous vessel image processing methods.

[0265] Furthermore, it should be noted that: when the venous blood vessel image processing device provided in the above embodiment determines the dominant vertex, the division of the above-mentioned functional modules is only used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the venous blood vessel image processing device can be divided into different functional modules to complete all or part of the functions described above.

[0266] In addition, the venous blood vessel image processing device and the venous blood vessel image processing method provided in the above embodiments belong to the same concept, and their specific implementation processes are detailed in the method embodiments and will not be repeated here.

[0267] The present invention discloses a venous blood vessel image processing method, device, equipment and storage medium, which have the following beneficial effects:

[0268] First, the present invention processes and analyzes venous vessel images, performs three-dimensional modeling of the vessels, and plans the irradiation path of radiotherapy, laying the foundation for the subsequent use of external irradiation to treat varicose veins.

[0269] Second, the present invention uses pre-processing such as median filtering, CLAHE contrast enhancement, and Laplace sharpening to improve the blood vessel-background contrast and signal-to-noise ratio.

[0270] Third, the present invention combines the improved U-Net segmentation network and the improved Zhang-Suen skeletonization to achieve blood vessel segmentation with high segmentation accuracy.

[0271] Fourth, the present invention adopts a multi-level B-spline vascular mapping method for vascular modeling and analysis, and the analysis results are more accurate.

[0272] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which shall fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for processing a venous blood vessel image, characterized in that: include: Step S1: Acquire venous blood vessel images; Step S2: pre-processing the acquired venous blood vessel image; Step S3: using the improved U-Net segmentation network and the improved Zhang-Suen skeletonization algorithm to process the preprocessed image and obtain the blood vessel segmentation result; Step S4: Based on the vascular segmentation results, a multi-level B-spline mapping method is used to perform 3D vascular modeling, quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters, and obtain a vascular morphology-function joint evaluation index.

2. The venous blood vessel image processing method according to claim 1, characterized in that: The step S2 comprises: Step S2.1: performing sliding window median filtering on the acquired venous blood vessel image; Step S2.2: performing CLAHE contrast enhancement on the image after median filtering; Step S2.3: Perform Laplacian sharpening on the image after CLAHE contrast enhancement processing.

3. The venous blood vessel image processing method according to claim 1, characterized in that: The step S3 comprises: Step S3.1: Construct an improved U-Net segmentation network, train it using a composite loss function and a dynamic data augmentation strategy, and use the trained U-Net segmentation network to segment the preprocessed image to obtain the initial blood vessel segmentation result; Step S3.2: Perform improved Zhang-Suen skeletonization on the initial blood vessel segmentation result to obtain a refined blood vessel segmentation result.

4. The venous blood vessel image processing method according to claim 1, characterized in that: The step S4 comprises: Step S4.1: converting the blood vessel segmentation result in the image coordinate system to the world coordinate system; Step S4.2: Perform three-dimensional reconstruction of the vascular centerline in the world coordinate system to generate a topologically continuous vascular centerline model; Step S4.3: Constructing a vascular surface geometric model based on the vascular axis and radius distribution; Step S4.4: Based on the vascular surface geometric model, quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters to obtain a vascular morphology-function joint evaluation index.

5. A venous blood vessel image processing device, characterized in that: include: An acquisition module, used for acquiring venous blood vessel images; A preprocessing module, used for preprocessing the acquired venous blood vessel images; The segmentation module is used to process the preprocessed image using the improved U-Net segmentation network and the improved Zhang-Suen skeletonization algorithm to obtain the blood vessel segmentation result; The analysis module is used to perform 3D vascular modeling based on the vascular segmentation results using a multi-level B-spline mapping method, quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters, and obtain vascular morphology-function joint evaluation indicators.

6. The venous blood vessel image processing device according to claim 5, characterized in that: The pre-processing module comprises: A first pre-processing unit is used to perform sliding window median filtering on the acquired venous blood vessel image; The second pre-processing unit is used to perform CLAHE contrast enhancement processing on the image after the median filtering processing; The third pre-processing unit is used to perform Laplace sharpening processing on the image after the CLAHE contrast enhancement processing.

7. The venous blood vessel image processing device according to claim 5, characterized in that: The segmentation module includes: The first segmentation unit is used to construct an improved U-Net segmentation network, which is trained using a composite loss function and a dynamic data enhancement strategy. The trained U-Net segmentation network is used to segment the preprocessed image to obtain the initial blood vessel segmentation result. The second segmentation unit is used to perform improved Zhang-Suen skeletonization processing on the initial blood vessel segmentation result to obtain a refined blood vessel segmentation result.

8. The venous blood vessel image processing device according to claim 5, characterized in that: The analysis module includes: A first analysis unit is used to convert the blood vessel segmentation result in the image coordinate system into the world coordinate system; The second analysis unit is used to perform three-dimensional reconstruction of the blood vessel centerline in the world coordinate system to generate a topologically continuous blood vessel centerline model; The third analysis unit is used to construct a blood vessel surface geometric model based on the blood vessel central axis and radius distribution; The fourth analysis unit is used to quantitatively analyze the vascular spatial geometric characteristics and hemodynamic parameters based on the vascular surface geometric model to obtain a vascular morphology-function joint evaluation index.

9. A computing device, characterized in that include: one or more processors; Memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, and the one or more programs include instructions of the venous blood vessel image processing method according to any one of claims 1 to 4.

10. A storage medium, characterized in that The storage medium stores one or more computer-readable programs, and the one or more programs include instructions, and the instructions are suitable for being loaded by the memory and executing the venous blood vessel image processing method according to any one of claims 1 to 4.

Images