Hepatocellular carcinoma vascular network three-dimensional reconstruction method and system, electronic equipment and medium

Abstract

The invention provides a hepatocellular carcinoma vascular network three-dimensional reconstruction method and system, electronic equipment and a medium, and belongs to the technical field of image processing. The method comprises the following steps: acquiring a plurality of original pathological section images, wherein the original pathological section images are obtained by slicing a hepatocellular carcinoma sample; performing spatial position alignment on the plurality of original pathological section images to obtain a plurality of corresponding first pathological section images; performing blood vessel enhancement, segmentation and filling on the plurality of first pathological section images to obtain a plurality of corresponding second pathological section images; and performing three-dimensional reconstruction according to the plurality of second pathological section images to obtain a three-dimensional model of a hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample, and performing visualization. According to the method, the three-dimensional model of the hepatocellular carcinoma vascular network can be efficiently, accurately and reliably constructed through a computer vision technology.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a method, system, electronic device, and medium for three-dimensional reconstruction of a hepatocellular carcinoma vascular network. Background Art

[0002] In the current auxiliary diagnosis process for hepatocellular carcinoma samples, doctors are typically required to mark suspicious areas in each pathology slide image. Then, by reviewing the pathology slide images at different levels, they independently construct a three-dimensional understanding of the hepatocellular carcinoma sample to assist in evaluating the abnormalities of the vascular structures formed in the hepatocellular carcinoma sample. However, when the pathology slide images are interfered with by complex background colors or the suspected abnormal structures are small in size, even experienced doctors may make deviations in their visual judgment, resulting in the overlooking of certain small but potentially important abnormal structures. This shows that the method of relying on doctors to review individual pathology slide images to form a three-dimensional understanding is not conducive to achieving accurate auxiliary diagnosis. Summary of the Invention

[0003] The main purpose of this application is to propose a three-dimensional reconstruction method, system, electronic equipment and medium for the hepatocellular carcinoma vascular network, aiming to use computer vision technology to efficiently, accurately and reliably construct a three-dimensional model of the hepatocellular carcinoma vascular network.

[0004] To achieve the above objectives, one aspect of the present application provides a method for three-dimensional reconstruction of a hepatocellular carcinoma vascular network, the method comprising:

[0005] Acquire multiple original pathological slice images, which are obtained by slicing a hepatocellular carcinoma sample;

[0006] performing spatial position alignment on the multiple original pathological slice images to obtain corresponding multiple first pathological slice images;

[0007] performing blood vessel enhancement, segmentation, and filling on the plurality of first pathological slice images to obtain a corresponding plurality of second pathological slice images;

[0008] A three-dimensional reconstruction is performed based on the multiple second pathological slice images to obtain a three-dimensional model of the hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample and visualize the model.

[0009] Furthermore, each of the original pathological slice images carries a unique slice serial number; and the spatial alignment of the multiple original pathological slice images to obtain the corresponding multiple first pathological slice images includes:

[0010] The original pathology slice image with the smallest slice number is used as the global reference image, and all the original pathology slice images with non-minimum slice numbers are used as images to be aligned;

[0011] For each of the images to be aligned, selecting an original pathology slice image that satisfies a preset relationship condition with the image to be aligned from the multiple original pathology slice images and using it as a local reference image;

[0012] Calculating a first transformation matrix between the image to be aligned and the global reference image, and calculating a second transformation matrix between the local reference image and the global reference image;

[0013] Performing feature point matching on the image to be aligned and the local reference image to obtain position information of all feature point pairs;

[0014] According to the position information of all the feature point pairs, the first transformation matrix and the second transformation matrix, a rigid transformation is performed on the images to be aligned to obtain a first pathological slice image.

[0015] Furthermore, performing a rigid transformation on the image to be aligned according to the position information of all feature point pairs, the first transformation matrix, and the second transformation matrix to obtain a first pathological slice image includes:

[0016] Calculating a local similarity transformation matrix between the image to be aligned and the local reference image according to the position information of all feature point pairs, the first transformation matrix, and the second transformation matrix;

[0017] Calculating a final transformation matrix according to the first transformation matrix and the local similarity transformation matrix;

[0018] According to the final transformation matrix, a rigid transformation is performed on the images to be aligned to obtain the first pathological slice image.

[0019] Furthermore, the performing blood vessel enhancement, segmentation, and filling on the plurality of first pathological slice images to obtain corresponding plurality of second pathological slice images includes:

[0020] For each of the first pathological slice images, performing color deconvolution on the first pathological slice image to obtain a third pathological slice image;

[0021] calculating all eigenvalues ​​of the Hessan matrix corresponding to the third pathological slice image to construct a blood vessel enhancement filter, and then using the blood vessel enhancement filter to perform multi-scale enhancement and screening on the blood vessel structure contained in the third pathological slice image to obtain a fourth pathological slice image;

[0022] performing coarse segmentation of the vascular structure contained in the fourth pathological slice image using a region growing algorithm to obtain a first mask image, and then performing fine segmentation of the vascular structure contained in the first mask image using an active contour model to obtain a second mask image;

[0023] The second mask image is subjected to blood vessel skeleton recognition and optimization to obtain a third mask image, and then the optimized blood vessel skeleton contained in the third mask image is subjected to flood filling to obtain a second pathological slice image.

[0024] Furthermore, performing blood vessel skeleton recognition and optimization on the second mask image to obtain a third mask image includes:

[0025] identifying a blood vessel skeleton from the second mask image using a skeletonization algorithm, wherein the blood vessel skeleton comprises a plurality of branches, and each branch comprises a plurality of skeleton pixels;

[0026] In the second mask image, all target skeleton pixels that meet the preset screening conditions contained in each branch are deleted to obtain each optimized branch, and then every two optimized branches with the closest distance are connected to generate the third mask image.

[0027] Furthermore, all target skeleton pixels that meet the preset screening conditions contained in each branch are deleted, and each optimized branch includes:

[0028] For each skeleton pixel point included in each branch, extracting a neighborhood pixel region corresponding to the skeleton pixel point from the second mask image and traversing the region to determine a crossover number corresponding to the skeleton pixel point, where the crossover number represents the number of times that background pixels transition to foreground pixels within the neighborhood pixel region;

[0029] Taking the skeleton pixel point as the foreground pixel point and the crossover number corresponding to the skeleton pixel point being equal to a preset value as a condition, all candidate skeleton pixel points that meet the condition are screened out from the branch to form a branch path;

[0030] When the length of the branch path does not exceed a preset length threshold, all the candidate skeleton pixel points included in the branch are taken as target skeleton pixel points and deleted to obtain an optimized branch.

[0031] Furthermore, performing three-dimensional reconstruction based on the plurality of second pathological slice images to obtain a three-dimensional model of the hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample and visualizing the model includes:

[0032] interpolating the plurality of second pathological slice images using an optical flow interpolation algorithm to form a set of pathological slice images;

[0033] A ray casting algorithm is used to perform volume rendering on the pathological slice image set to obtain a three-dimensional model of the hepatocellular carcinoma vascular network and visualize it.

[0034] To achieve the above objectives, another aspect of the present application provides a three-dimensional reconstruction system for hepatocellular carcinoma vascular network, the system comprising:

[0035] The first module is used to obtain a plurality of original pathological slice images, which are obtained by slicing a hepatocellular carcinoma sample;

[0036] A second module is configured to perform spatial position alignment on the plurality of original pathological slice images to obtain a corresponding plurality of first pathological slice images;

[0037] A third module is configured to perform blood vessel enhancement, segmentation, and filling on the plurality of first pathological slice images to obtain a corresponding plurality of second pathological slice images;

[0038] The fourth module is configured to perform three-dimensional reconstruction based on the plurality of second pathological slice images to obtain and visualize a three-dimensional model of the hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample.

[0039] To achieve the above-mentioned purpose, another aspect of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the above-mentioned three-dimensional reconstruction method of the hepatocellular carcinoma vascular network when executing the computer program.

[0040] To achieve the above objectives, another aspect of the present application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-mentioned method for three-dimensional reconstruction of a hepatocellular carcinoma vascular network.

[0041] This application includes at least the following beneficial effects: by using computer vision technology to sequentially perform spatial position alignment, vascular structure processing and three-dimensional reconstruction on multiple original pathological slice images associated with hepatocellular carcinoma samples, a three-dimensional model of the hepatocellular carcinoma vascular network can be obtained efficiently, accurately and reliably, so that doctors can subsequently directly obtain useful pathological information from it, which is conducive to the auxiliary diagnosis of hepatocellular carcinoma samples. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a flow chart of a method for three-dimensional reconstruction of a hepatocellular carcinoma vascular network provided in an embodiment of the present application;

[0043] Figure 2 This is a schematic diagram of the composition of a three-dimensional reconstruction system for hepatocellular carcinoma vascular network provided in an embodiment of the present application;

[0044] Figure 3 This is a schematic diagram of the hardware structure of the electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present application. They are merely examples of systems and methods consistent with some aspects of the embodiments of the present application as detailed in the appended claims.

[0046] It will be understood that the terms "first", "second", etc. used in this application may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0047] The terms "at least one", "plurality", "each", "any", etc. used in this application include "at least one", "two" or more, "plurality" or "each", "any" or "any one", "each" or "any one" as used herein.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0049] Before explaining the embodiments of the present application in detail, some of the nouns and terms involved in the embodiments of the present application are first explained. The nouns and terms involved in the embodiments of the present application are subject to the following explanations.

[0050] Color deconvolution algorithm is a technology that decomposes the color information in an image into multiple color channels or base colors.

[0051] The Hessi an matrix of an image is a second-order matrix used to describe the change in grayscale values ​​in a local area of ​​the image. It contains the second-order partial derivative information of the image and can be used to detect features such as corners and edges in the image.

[0052] The region growing algorithm is a process of developing groups of pixels or regions into larger regions. Its basic principle is: starting from a set of seed points, the region growth of these seed points is achieved by merging adjacent pixels with similar properties (such as intensity, grayscale, texture color, etc.) to each seed point into this region.

[0053] The active contour model is a framework for extracting object contours from two-dimensional images that may contain noise. Its basic principle is: first, an initial curve is created in the two-dimensional image. The shape of the initial curve is not restricted but the contour of the target object needs to be included inside. Then, an energy equation is established, which includes an internal energy term based on the shape of the canonical curve and an external energy term based on the degree of proximity between the canonical curve and the contour of the target object. The energy equation is then operated. By minimizing the internal energy term, the curve can be continuously tightened inward and kept smooth. By minimizing the external energy term, the curve can be continuously close to the contour of the target object until consistency is reached.

[0054] The basic principle of the skeletonization algorithm is: based on morphological operations and image processing technology, a series of transformations and operations are performed on the image to finally obtain the skeleton of the target object. The skeleton can be regarded as the most detailed representation of the target object, which can maintain the basic structure and topological relationship of the target object while removing unnecessary detail information.

[0055] The optical flow interpolation algorithm is a video interpolation technology based on optical flow estimation. Its basic principle is to predict the intermediate frame by calculating the pixel motion information (i.e., optical flow information) between adjacent frames.

[0056] The basic principle of the ray casting algorithm is as follows: starting from each pixel on the screen, a ray is emitted along the line of sight. When the ray passes through the volume data, samples are taken at equal distances along the direction of the ray, and the color value and opacity of the sampling point are calculated by interpolation. Then, the sampling points on the ray are synthesized in a front-to-back or back-to-front order to calculate the color value of the pixel on the screen corresponding to the ray.

[0057] In current medical practice, doctors still rely mainly on their accumulated professional knowledge, clinical experience and intuition for diagnosis. Although this traditional method is intuitive and has been proven effective in long-term practice, it is inevitably affected by human cognitive limitations (such as attention fatigue, subjective bias, and insufficient perception of subtle abnormalities). This can easily lead doctors to ignore key diagnostic clues, thereby causing medical errors or unnecessary misdiagnosis.

[0058] Specifically, in the existing auxiliary diagnosis process for hepatocellular carcinoma samples, doctors are usually required to mark suspicious areas in each pathological slice image, and then construct a three-dimensional understanding of the hepatocellular carcinoma sample by themselves by viewing the pathological slice images at different levels one by one, so as to assist in evaluating the abnormalities of the vascular structure formed in the hepatocellular carcinoma sample. However, this implementation method requires extremely high capabilities of the doctor, that is, not only a solid professional knowledge background is required, but also excellent three-dimensional spatial imagination and sustained concentration. In addition, when there is complex background color interference in the pathological slice image or the suspicious abnormal structure is small in size, even experienced doctors may have deviations in visual judgment, resulting in some small but potentially important abnormal structures being ignored. It can be seen that the way doctors rely on viewing each pathological slice image to form three-dimensional cognition is not conducive to achieving accurate auxiliary diagnosis.

[0059] In view of this, the embodiments of the present application provide a method, system, electronic device and medium for three-dimensional reconstruction of the hepatocellular carcinoma vascular network. This solution uses computer vision technology to sequentially perform spatial position alignment, vascular structure processing and three-dimensional reconstruction on multiple original pathological slice images associated with hepatocellular carcinoma samples. It can efficiently, accurately and reliably obtain a three-dimensional model of the hepatocellular carcinoma vascular network, allowing doctors to subsequently directly obtain useful pathological information from it, which is conducive to the auxiliary diagnosis of hepatocellular carcinoma samples.

[0060] The embodiment of the present application provides a three-dimensional reconstruction method for the vascular network of hepatocellular carcinoma, which relates to the field of image processing technology and can be applied to a terminal or a server, or can be software running in a terminal or a server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, and a car terminal, etc., but is not limited to this; the server side can be configured as an independent physical server, or as a server cluster or distributed system composed of multiple physical servers, or as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network; the software can be an application that implements the above method, etc., but is not limited to the above forms.

[0061] See also Figure 1 , Figure 1 This is an optional flow chart of a method for three-dimensional reconstruction of a hepatocellular carcinoma vascular network provided in an embodiment of the present application. The method may include, but is not limited to, the following steps S101 to S104:

[0062] Step S101: Acquire multiple original pathological slice images, which are obtained by slicing a hepatocellular carcinoma sample;

[0063] Step S102: spatially aligning the multiple original pathological slice images to obtain corresponding multiple first pathological slice images;

[0064] Step S103: performing blood vessel enhancement, segmentation, and filling on the plurality of first pathological slice images to obtain corresponding plurality of second pathological slice images;

[0065] Step S104 : performing three-dimensional reconstruction based on the plurality of second pathological slice images to obtain a three-dimensional model of the hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample and perform visualization.

[0066] In steps S101 to S104 shown in the embodiment of the present application, a three-dimensional model of the hepatocellular carcinoma vascular network can be constructed efficiently, accurately and reliably by introducing computer vision technology.

[0067] In step S101 of some embodiments, a plurality of continuous original pathological slice images can be obtained by finely slicing the hepatocellular carcinoma sample vertically and performing high-definition imaging of each layer. In this case, each original pathological slice image carries a unique slice serial number, which can be understood as the imaging result of the layer of the hepatocellular carcinoma sample to which the original pathological slice image belongs. Its working principle is similar to that of CT (Computed Tomography). In addition, before calling multiple original pathological slice images for processing, the pixel information of each original pathological slice image can be split into multiple rectangular blocks of fixed size, and then the pixel information of each rectangular block is compressed in the form of three-channel 24-bit data using JPEG (Joint Photographic Experts Group) and stored in sequence. Subsequently, the multiple original pathological slice images after block compression and storage are merged into a TIFF (Tag Image File Format) file for storage.

[0068] In some embodiments, step S102 may include, but is not limited to, the following steps S201 to S202:

[0069] Step S201: defining the original pathological slice image with the smallest slice number as the global reference image, and directly using the original pathological slice image with the smallest slice number as the first pathological slice image corresponding thereto, and defining all original pathological slice images with non-smallest slice numbers as images to be aligned;

[0070] Step S202: For each image to be aligned, first, an original pathological slice image that satisfies a preset relationship condition with the image to be aligned is screened out from multiple original pathological slice images and defined as a local reference image, wherein the preset relationship condition is used to limit the slice sequence number between the local reference image and the image to be aligned to be adjacent, and the slice sequence number carried by the local reference image is smaller than the slice sequence number carried by the image to be aligned; then, a first transformation matrix between the image to be aligned and the global reference image is calculated, and a second transformation matrix between the local reference image and the global reference image is calculated, wherein these two transformation matrices can be understood as affine transformation matrices; and feature point matching is performed on the image to be aligned and the local reference image to obtain position information of all feature point pairs; finally, a rigid transformation is performed on the image to be aligned based on the position information of all feature point pairs, the first transformation matrix, and the second transformation matrix to obtain the corresponding first pathological slice image, and the image is longitudinally aligned with the global reference image.

[0071] As an optional implementation, before executing the above step S201, each original pathological slice image can be normalized and preprocessed, that is, each original pathological slice image presented in RGB form is converted into the CAM16-UCS color space and then converted back to the RGB space, and finally a grayscale image with a dark background and a bright tissue is obtained, thereby improving the accuracy of subsequent image spatial position alignment.

[0072] In the above step S202, regarding the step of performing feature point matching on the image to be aligned and the local reference image to obtain position information of all feature point pairs, the corresponding implementation method may include the following:

[0073] First, a BRISK (Binary Robust Invariant Scalable Keypoints) detector is used to extract a number of first key feature points from the image to be aligned, and a number of second key feature points are extracted from the local reference image using the BRISK detector. Then, a SIFT (Scale Invariant Feature Transform) descriptor is used to match the first key feature points with the second key feature points to obtain a plurality of candidate feature point pairs. Finally, a brute force matching is performed on the plurality of candidate feature point pairs. Subsequently, the brute force matching results are filtered using a RANSAC (Random Sample Consensus) algorithm, and all feature point pairs that meet a preset position condition are further screened out. Each feature point pair includes a first feature point and a second feature point that matches it. Then, the position information of the first feature point in the image to be aligned and the position information of the second feature point in the local reference image are obtained. The preset position condition is used to limit the first feature point to fall within the foreground contour line of the image to be aligned and the second feature point to fall within the foreground contour line of the local reference image.

[0074] In the above step S202, regarding the step of performing a rigid transformation on the image to be aligned based on the position information of all feature point pairs, the first transformation matrix, and the second transformation matrix to obtain the corresponding first pathological slice image, the corresponding implementation process may include, but is not limited to, the following steps S301 to S303:

[0075] Step S301: Calculate a local similarity transformation matrix between the image to be aligned and the local reference image based on the position information of all feature point pairs, the first transformation matrix, and the second transformation matrix;

[0076] Specifically, based on the position information of all feature point pairs, the first transformation matrix, and the second transformation matrix, an objective function for minimizing the characterization error is constructed as follows:

[0077]

[0078] Where f is the objective function, K is the number of all feature point pairs, T i is the first transformation matrix, T i-1 is the second transformation matrix, M i,i-1 is the local similarity transformation matrix, each feature point pair includes a first feature point and a second feature point that matches it, the first feature point belongs to the image to be aligned, and the second feature point belongs to the local reference image, P i,kis the position information of the first feature point contained in the kth feature point pair, P i-1,k is the position information of the second feature point contained in the k-th feature point pair;

[0079] By solving the objective function, the optimal local similarity transformation matrix can be obtained.

[0080] Step S302: Calculate the final transformation matrix based on the first transformation matrix and the local similarity transformation matrix, which can be implemented using the following expression:

[0081] M′ i,i-1 =T i M i,i-1

[0082] Where M′ i,i-1 is the final transformation matrix.

[0083] Step S303: Perform a rigid transformation on the image to be aligned according to the final transformation matrix to obtain the corresponding first pathological slice image.

[0084] In an embodiment of the present application, taking into account the problems that may arise during the preparation of pathological slices, such as movement of hepatocellular carcinoma samples and precision limitations of slicing equipment, which lead to inevitable differences in spatial positions between the different original pathological slice images obtained, the spatial positions of other original pathological slice images are aligned based on the original pathological slice image with the smallest slice number, which is conducive to the subsequent reliable three-dimensional reconstruction.

[0085] In step S103 of some embodiments, taking any first pathological slice image as an example, regarding the step of performing blood vessel enhancement, segmentation, and filling on the first pathological slice image to obtain a corresponding second pathological slice image, the corresponding implementation process may include, but is not limited to, the following steps S401 to S404:

[0086] Step S401: Perform color deconvolution on the first pathological section image to obtain a third pathological section image. It can be understood that for the CD34 immunohistochemically stained section, the vascular structure is colored brown by DAB (3,3'-diaminobenzidine), while the vascular endothelial cell nucleus is counterstained blue by hematoxylin. The DAB channel component is extracted from the first pathological section image using the color deconvolution algorithm, so that only the vascular structure is retained in the third pathological section image to filter out other irrelevant feature information.

[0087] Step S402: Calculate all eigenvalues ​​of the Hessian matrix corresponding to the third pathological slice image to construct a specific blood vessel enhancement filter, and then use the blood vessel enhancement filter to perform multi-scale enhancement and screening on the blood vessel structure contained in the third pathological slice image to obtain a fourth pathological slice image. The blood vessel enhancement filter can be implemented using the following expression:

[0088]

[0089] Where V1 is the output of the blood vessel enhancement filter, σ min is the preset minimum scale, σ max is the preset maximum scale, σ is the current scale, V0 is the vascular similarity response function, * can be understood as a convolution operation, G(x, y, σ) is a Gaussian function, (x, y) is the position information of the pixel points contained in the third pathological slice image, R B and S are reference parameters set for the convenience of description, where R B =|λ1| / |λ2|, λ1 and λ2 are both eigenvalues ​​of the Hessian matrix corresponding to the third pathological slice image, β and c are both control parameters, β is used to control the reference parameter R B The sensitivity of c is used to control the sensitivity of the reference parameter S.

[0090] Step S403: performing coarse segmentation of the vascular structure contained in the fourth pathological slice image using a region growing algorithm to obtain a first mask image, and then performing fine segmentation of the vascular structure contained in the first mask image using an active contour model to obtain a second mask image;

[0091] Specifically, all the vascular structure feature points contained in the fourth pathological slice image are used as initial seed points. Then, based on the initial seed points, the fourth pathological slice image is subjected to a rough vascular segmentation process using a region growing algorithm to obtain a first mask image. Considering that the first mask image may contain some background pixels, the boundary of the vascular structure is extracted from the first mask image and used as the initial active contour curve. Then, with the goal of minimizing an energy function set within the active contour model, the initial active contour curve contained in the first mask image is optimized to generate the second mask image. The energy function can be expressed as follows:

[0092]

[0093] Where E(C) is the energy function, C is the initial curve of the active contour, I is the first mask image, is the gradient of the first mask image, inside(C) is used to cover all first pixel points contained in the first mask image that fall inside the initial curve of the active contour, μ in is the average grayscale value of all first pixels, outside(C) is used to cover all second pixels contained in the first mask image that fall outside the initial curve of the active contour, μ out is the average grayscale value of all second pixels, α, β, and γ are all weight parameters. α, as an edge term, can be used to attract the contour to the target through image gradient information. β and γ, as regional terms, can be used to suppress background interference by minimizing the grayscale difference inside and outside the contour.

[0094] Step S404: perform vascular skeleton recognition and optimization on the second mask image to obtain a third mask image, and then perform flood filling on the optimized vascular skeleton contained in the third mask image to obtain a corresponding second pathological slice image.

[0095] In an embodiment of the present application, considering that there may be blood vessels of different thicknesses in pathological sections, by introducing a multi-scale enhancement and screening mechanism into the blood vessel enhancement filter for image processing, the linear structural features of the blood vessels of interest can be better extracted from the image and noise can be filtered out; by introducing a region growing algorithm and an active contour model, reliable segmentation of the blood vessel structure contained in the image at different levels can be achieved; considering that there are broken blood vessels in the pathological sections due to staining, by optimizing and flood filling the blood vessel skeleton contained in the image, a more complete blood vessel structure can be presented in the final image.

[0096] In step S404 of some embodiments, regarding the step of performing blood vessel skeleton recognition and optimization on the second mask image to obtain a third mask image, the corresponding implementation process may include, but is not limited to, the following steps S501 to S502:

[0097] Step S501: Using a skeletonization algorithm to identify a vascular skeleton from the second mask image, the vascular skeleton includes multiple branches, and each branch includes multiple skeleton pixels; wherein the skeletonization algorithm is based on two-dimensional thinning, converting the target area in the binary image into a skeleton line with a single pixel width.

[0098] Step S502: In the second mask image, all target skeleton pixels that meet the preset screening conditions in each branch are deleted to obtain each optimized branch, and then every two optimized branches with the closest distance are connected to generate the third mask image.

[0099] In the above step S502, taking any branch included in the blood vessel skeleton as an example, all target skeleton pixels included in the branch that meet the preset screening conditions are deleted to obtain an optimized branch. The corresponding implementation process may include, but is not limited to, the following steps S601 to S603:

[0100] Step S601: For each skeleton pixel point included in the branch, extract the neighborhood pixel area corresponding to the skeleton pixel point from the second mask image and traverse it to determine the crossover number corresponding to the skeleton pixel point. The crossover number is used to represent the number of times that the background pixel point transitions to the foreground pixel point in the neighborhood pixel point. The pixel value of the background pixel point can be 0, and the pixel value of the foreground pixel point can be 1.

[0101] Specifically, the neighborhood pixel area corresponding to the skeleton pixel point is actually an eight-neighborhood pixel area. Each pixel point contained in the eight-neighborhood pixel area belongs to a background pixel point or a foreground pixel point. Each pixel point contained in the eight-neighborhood pixel area is traversed and judged in a clockwise order to count the number of times that background pixels are converted to foreground pixels in the eight-neighborhood pixel area.

[0102] For example, the first pixel point, the second pixel point, the fifth pixel point and the seventh pixel point contained in the eight-neighborhood pixel area corresponding to the skeleton pixel point are all background pixel points, and the third pixel point, the fourth pixel point, the sixth pixel point and the eighth pixel point contained in the eight-neighborhood pixel area are all foreground pixel points. It can be seen that a distribution state conversion occurs between the second pixel point and the third pixel point, between the fifth pixel point and the sixth pixel point, and between the seventh pixel point and the eighth pixel point in the eight-neighborhood pixel area, respectively, and the statistical crossover number corresponding to the skeleton pixel point is 3.

[0103] Step S602: Taking the skeleton pixel point as the foreground pixel point and the cross number corresponding to the skeleton pixel point being equal to the preset value as a condition, all candidate skeleton pixel points that meet the condition are filtered out from the branch and a branch path is formed; wherein, the preset value is preferably set to 1, and when the cross number corresponding to the skeleton pixel point is equal to 1, it indicates that the skeleton pixel point has only one foreground neighborhood connected to it.

[0104] Step S603: When the length of the branch path does not exceed the preset length threshold, it is determined that the branch path is an unnecessary small branch caused by noise interference when the skeletonization algorithm is applied for image recognition. At this time, all candidate skeleton pixel points contained in the branch are used as target skeleton pixel points and deleted to obtain the optimized branch.

[0105] It should be noted that when the length of the branch path exceeds the preset length threshold, the branch path is judged to belong to the main branch of the vascular skeleton, and all candidate skeleton pixels contained in the branch should not be deleted. At this time, the branch is directly used as the optimized branch.

[0106] In the above step S502 , the step of connecting every two optimized branches with the shortest distance between them can be implemented by using a shortest path connection algorithm.

[0107] In some embodiments, step S104 may include, but is not limited to, the following steps S701 to S702:

[0108] Step S701: interpolating a plurality of second pathological slice images using an optical flow interpolation algorithm to form a set of pathological slice images;

[0109] Specifically, for every two second pathological slice images with adjacent slice numbers, the optical flow interpolation algorithm is used to generate three intermediate pathological slice images between the two second pathological slice images, and then the multiple second pathological slice images and the three intermediate pathological slice images generated between every two second pathological slice images are aggregated to form the pathological slice image set.

[0110] Furthermore, the step of generating three intermediate pathological slice images between the two second pathological slice images using the optical flow interpolation algorithm can be understood as first calculating the optical flow field generated between the two second pathological slice images, then performing spatial deformation displacement on the two second pathological slice images according to the optical flow field to obtain two corresponding intermediate pathological slice images, and then performing weighted fusion on the two intermediate pathological slice images to obtain a third intermediate pathological slice image; corresponding implementation methods may include the following:

[0111] The second pathological slice image with a smaller slice number is recorded as the first image, and the other second pathological slice image is recorded as the second image. The first image and the second image are locally modeled using quadratic polynomials as follows:

[0112]

[0113] Wherein, f1(x) refers to the first image, f2(x) refers to the second image, M1 and M2 are both symmetric matrices, n1 and n2 are both vectors, and k1 and k2 are both scalars;

[0114] Assume that the first image is obtained by shifting the second image, and the corresponding shift is recorded as d(x). The expression for the shift d(x) is as follows:

[0115]

[0116] The weight function α(x) is introduced for weighted averaging, as follows:

[0117]

[0118] Apply the above weighted average formula to each pixel in the first image and each pixel in the second image to calculate the optical flow field between the first image and the second image as D(f1,f2)={d(x)|x∈C}, and then introduce the deformation operator in combination with the preset scale factor σ It is used to spatially deform the input image V according to the method of σ·D(f1,f2), and in this process, it is limited that σ≠0 to avoid that the input image V does not shift;

[0119] Set the time factor to t = k / n. When the multiple slice numbers carried by the multiple second pathological slice images are gradually increased from 1, n is the maximum slice number carried by the multiple second pathological slice images, and k is the slice number carried by the first image. At this time, the first intermediate pathological slice image generated according to the first image is And a second intermediate pathological slice image is generated according to the second image: Then the two intermediate pathological slice images are weighted fused to obtain the third intermediate pathological slice image F t =tf 1t +(1-t)f 2t .

[0120] Step S702: performing volume rendering on the pathological slice image set using a ray casting algorithm to obtain a three-dimensional model of the hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample and perform visualization;

[0121] Specifically, the pathological slice image set is first reconstructed in three dimensions using a stereo matching algorithm to obtain the original volume data. The original volume data is then processed using a ray casting algorithm to generate the final required three-dimensional model of the hepatocellular carcinoma vascular network. Finally, the three-dimensional model of the hepatocellular carcinoma vascular network is visualized after three-dimensional Gaussian filtering.

[0122] The specific process of the ray casting algorithm is described as follows:

[0123] (1) Volume data value classification: All voxels contained in the original volume data are classified into voxels of different colors and voxels of different opacities according to the grayscale value of each voxel contained in the original volume data;

[0124] (2) Color and opacity assignment: By using a preset transfer function, the voxel value is mapped to a specific RGB color value and Alpha transparency value to convert the numerical value of the abstract data into a more intuitive visual representation;

[0125] (3) Resampling and interpolation: Construct a more continuous three-dimensional representation through trilinear interpolation, that is, ensure that the middle point is the weighted average of the values ​​of the eight surrounding voxels;

[0126] (4) Light integration and mixed image: All sampling points collected along the light are synthesized backward to obtain the light intensity generated in the direction from the sampling point to the observer, and finally generate the corresponding pixel value.

[0127] Among them, Gaussian filtering is a linear smoothing filter based on the Gaussian function, which is used for noise suppression, edge smoothing and data preprocessing. The three-dimensional Gaussian filter is based on the two-dimensional Gaussian filter and is extended to the three-dimensional space to be applicable to three-dimensional volume data. It is used to perform weighted averaging on each voxel contained in the volume data, while suppressing the high-frequency noise generated by local under-segmentation or over-segmentation while retaining the main structural features. It can reduce local discontinuities and enhance the overall smoothness of the three-dimensional model. The core of the three-dimensional Gaussian filter lies in the three-dimensional Gaussian function. In the case of isotropy, the three-dimensional Gaussian function is defined as:

[0128]

[0129] Where σ is the standard deviation of the same size in the X, Y, and Z directions.

[0130] In an embodiment of the present application, considering that the spatial resolutions of the multiple second pathological slice images have significant differences in different directions, the multiple second pathological slice images are interpolated by an optical flow interpolation algorithm to reduce the impact of anisotropy on subsequent three-dimensional reconstruction.

[0131] As an optional implementation, after executing step S702, the three-dimensional model of the hepatocellular carcinoma vascular network can be exported to OBJ format and saved for subsequent viewing and analysis. The OBJ format is a standard format widely supported in three-dimensional modeling and computer graphics. The OBJ format is actually represented in ASCII format, which has clear content and is easy to edit and view. It is mainly used to represent information such as vertices, faces, and texture coordinates of the three-dimensional model, and can store information such as color and texture.

[0132] The three-dimensional reconstruction method of the hepatocellular carcinoma vascular network provided in the embodiment of the present application uses computer vision technology to sequentially perform spatial position alignment, vascular structure processing, and three-dimensional reconstruction on multiple original pathological slice images associated with hepatocellular carcinoma samples. It can analyze the spatial characteristics of tissues in pathological slices and can efficiently, accurately, and reliably obtain a three-dimensional model of the hepatocellular carcinoma vascular network, allowing doctors to subsequently directly obtain useful pathological information from it, which is conducive to the auxiliary diagnosis of hepatocellular carcinoma samples. In addition, quantitative analysis can be continued based on the three-dimensional model of the hepatocellular carcinoma vascular network, such as vascular branch diameter, curvature, microvascular density, etc., and by combining it with the professional experience of doctors, it can simplify the diagnostic process and reduce the possibility of misdiagnosis.

[0133] See also Figure 2 , Figure 2 This is a schematic diagram of an optional component of a system for 3D reconstruction of a hepatocellular carcinoma vascular network provided in an embodiment of the present application, which can implement the above-mentioned 3D reconstruction method of a hepatocellular carcinoma vascular network. The system may include, but is not limited to, the following:

[0134] The first module 801 is used to obtain a plurality of original pathological slice images, which are obtained by slicing a hepatocellular carcinoma sample;

[0135] The second module 802 is configured to perform spatial position alignment on the multiple original pathological slice images to obtain corresponding multiple first pathological slice images;

[0136] The third module 803 is configured to perform blood vessel enhancement, segmentation, and filling on the plurality of first pathological slice images to obtain a corresponding plurality of second pathological slice images;

[0137] The fourth module 804 is configured to perform three-dimensional reconstruction based on the plurality of second pathological slice images to obtain and visualize a three-dimensional model of the hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample.

[0138] It can be understood that the contents of the above method embodiments are all applicable to the present system embodiments, the functions specifically implemented by the present system embodiments are the same as the functions specifically implemented by the above method embodiments, and the beneficial effects achieved by the present system embodiments are also the same as the beneficial effects achieved by the above method embodiments.

[0139] The present application also provides an electronic device comprising a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned method for 3D reconstruction of a hepatocellular carcinoma vascular network. The electronic device may include any intelligent terminal, such as a tablet computer or an in-vehicle computer.

[0140] It can be understood that the contents of the above method embodiments are all applicable to the embodiments of the present device, the functions specifically implemented by the embodiments of the present device are the same as the functions specifically implemented by the above method embodiments, and the beneficial effects achieved by the embodiments of the present device are also the same as the beneficial effects achieved by the above method embodiments.

[0141] See also Figure 3 , Figure 3 The hardware structure of an electronic device according to another embodiment is shown. The electronic device includes:

[0142] The processor 901 can be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application.

[0143] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solutions provided by the embodiments of the present application are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called by the processor 901 to execute the technical solutions provided by the embodiments of the present application.

[0144] Input / output interface 903, used to implement information input and output;

[0145] Communication interface 904, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.);

[0146] Bus 905 , which transmits information between various components of the device (e.g., processor 901 , memory 902 , input / output interface 903 , and communication interface 904 );

[0147] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0148] An embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-mentioned method for three-dimensional reconstruction of a hepatocellular carcinoma vascular network.

[0149] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiment, the functions specifically implemented by the present storage medium embodiment are the same as the functions specifically implemented by the above method embodiments, and the beneficial effects achieved by the present storage medium embodiment are also the same as the beneficial effects achieved by the above method embodiments.

[0150] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0151] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0152] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0153] The system embodiment described above is merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0154] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0155] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0156] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0157] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of the above units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or units, which can be electrical, mechanical or other forms.

[0158] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0159] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0160] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes multiple instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: various media that can store programs, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0161] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A method for three-dimensional reconstruction of hepatocellular carcinoma vascular network, characterized in that: The method comprises: Acquire multiple original pathological slice images, which are obtained by slicing a hepatocellular carcinoma sample; performing spatial position alignment on the multiple original pathological slice images to obtain corresponding multiple first pathological slice images; performing blood vessel enhancement, segmentation, and filling on the plurality of first pathological slice images to obtain a corresponding plurality of second pathological slice images; A three-dimensional reconstruction is performed based on the multiple second pathological slice images to obtain a three-dimensional model of the hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample and visualize the model.

2. The method for three-dimensional reconstruction of hepatocellular carcinoma vascular network according to claim 1, characterized in that: Each of the original pathological slice images carries a unique slice serial number; and spatially aligning the plurality of original pathological slice images to obtain a plurality of corresponding first pathological slice images comprises: The original pathology slice image with the smallest slice number is used as the global reference image, and all the original pathology slice images with non-minimum slice numbers are used as images to be aligned; For each of the images to be aligned, selecting an original pathology slice image that satisfies a preset relationship condition with the image to be aligned from the multiple original pathology slice images and using it as a local reference image; Calculating a first transformation matrix between the image to be aligned and the global reference image, and calculating a second transformation matrix between the local reference image and the global reference image; Performing feature point matching on the image to be aligned and the local reference image to obtain position information of all feature point pairs; According to the position information of all the feature point pairs, the first transformation matrix and the second transformation matrix, a rigid transformation is performed on the images to be aligned to obtain a first pathological slice image.

3. The method for three-dimensional reconstruction of hepatocellular carcinoma vascular network according to claim 2, characterized in that: The performing a rigid transformation on the image to be aligned according to the position information of all feature point pairs, the first transformation matrix, and the second transformation matrix to obtain a first pathological slice image includes: Calculating a local similarity transformation matrix between the image to be aligned and the local reference image according to the position information of all feature point pairs, the first transformation matrix, and the second transformation matrix; Calculating a final transformation matrix according to the first transformation matrix and the local similarity transformation matrix; According to the final transformation matrix, a rigid transformation is performed on the images to be aligned to obtain the first pathological slice image.

4. The method for three-dimensional reconstruction of hepatocellular carcinoma vascular network according to claim 1, characterized in that: The performing blood vessel enhancement, segmentation, and filling on the plurality of first pathological slice images to obtain corresponding plurality of second pathological slice images includes: For each of the first pathological slice images, performing color deconvolution on the first pathological slice image to obtain a third pathological slice image; calculating all eigenvalues ​​of the Hessan matrix corresponding to the third pathological slice image to construct a blood vessel enhancement filter, and then using the blood vessel enhancement filter to perform multi-scale enhancement and screening on the blood vessel structure contained in the third pathological slice image to obtain a fourth pathological slice image; performing coarse segmentation of the vascular structure contained in the fourth pathological slice image using a region growing algorithm to obtain a first mask image, and then performing fine segmentation of the vascular structure contained in the first mask image using an active contour model to obtain a second mask image; The second mask image is subjected to blood vessel skeleton recognition and optimization to obtain a third mask image, and then the optimized blood vessel skeleton contained in the third mask image is subjected to flood filling to obtain a second pathological slice image.

5. The method for three-dimensional reconstruction of hepatocellular carcinoma vascular network according to claim 4, characterized in that: The performing blood vessel skeleton recognition and optimization on the second mask image to obtain a third mask image includes: identifying a blood vessel skeleton from the second mask image using a skeletonization algorithm, wherein the blood vessel skeleton comprises a plurality of branches, and each branch comprises a plurality of skeleton pixels; In the second mask image, all target skeleton pixels that meet the preset screening conditions contained in each branch are deleted to obtain each optimized branch, and then every two optimized branches with the closest distance are connected to generate the third mask image.

6. The method for three-dimensional reconstruction of hepatocellular carcinoma vascular network according to claim 5, characterized in that: All target skeleton pixels that meet the preset screening conditions contained in each branch are deleted, and each optimized branch includes: For each skeleton pixel point included in each branch, extracting a neighborhood pixel region corresponding to the skeleton pixel point from the second mask image and traversing the region to determine a crossover number corresponding to the skeleton pixel point, where the crossover number represents the number of times that background pixels transition to foreground pixels within the neighborhood pixel region; Taking the skeleton pixel point as the foreground pixel point and the intersection number corresponding to the skeleton pixel point being equal to a preset value as a condition, all candidate skeleton pixel points that meet the condition are screened out from the branch to form a branch path; When the length of the branch path does not exceed a preset length threshold, all the candidate skeleton pixel points included in the branch are taken as target skeleton pixel points and deleted to obtain an optimized branch.

7. The method for three-dimensional reconstruction of hepatocellular carcinoma vascular network according to claim 1, characterized in that: The performing three-dimensional reconstruction based on the plurality of second pathological slice images to obtain a three-dimensional model of the hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample and visualizing the model includes: interpolating the plurality of second pathological slice images using an optical flow interpolation algorithm to form a set of pathological slice images; A ray casting algorithm is used to perform volume rendering on the pathological slice image set to obtain a three-dimensional model of the hepatocellular carcinoma vascular network and visualize it.

8. A three-dimensional reconstruction system for hepatocellular carcinoma vascular network, characterized in that: The system comprises: The first module is used to obtain a plurality of original pathological slice images, which are obtained by slicing a hepatocellular carcinoma sample; A second module is configured to perform spatial position alignment on the plurality of original pathological slice images to obtain a corresponding plurality of first pathological slice images; A third module is configured to perform blood vessel enhancement, segmentation, and filling on the plurality of first pathological slice images to obtain a corresponding plurality of second pathological slice images; The fourth module is configured to perform three-dimensional reconstruction based on the plurality of second pathological slice images to obtain and visualize a three-dimensional model of the hepatocellular carcinoma vascular network formed in the hepatocellular carcinoma sample.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the three-dimensional reconstruction method of the hepatocellular carcinoma vascular network according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for three-dimensional reconstruction of a hepatocellular carcinoma vascular network according to any one of claims 1 to 7 is implemented.

Images