Virtual dissection method based on VTK visual framework

Abstract

The invention provides a virtual dissection method based on a VTK visual framework, and relates to the field of virtual dissection processing. The method comprises the following steps: firstly scanning the appearance of a target corpse to obtain an original slice, constructing a body three-dimensional voxel model, then performing angiography to acquire an image, then reconstructing a blood vessel three-dimensional voxel model, splicing the original slice and the blood vessel three-dimensional voxel model into a target corpse structure three-dimensional diagram, and finally performing virtual anatomy treatment and display on the diagram. The invention provides a systematic and comprehensive virtual anatomy solution which covers each link to ensure coherence and integrity, accurately reconstructs a three-dimensional structure by means of advanced imaging and algorithm, serves multiple fields of medical science, realizes efficient interaction and visualization based on a VTK framework, facilitates user operation and observes a result in real time, and improves the user experience. The understanding and analysis capability of the anatomical structure is improved, and related work is assisted to be carried out.

Description

Technical Field

[0001] The present invention mainly relates to the field of virtual anatomy processing, and in particular to a virtual anatomy method based on a VTK visualization framework. Background Art

[0002] With the continuous development of modern medicine, the accurate understanding and research of human anatomy has become increasingly important in the fields of medical education, clinical practice, and medical research. Traditional anatomical research mainly relies on methods such as cadaver dissection, atlas consultation, and physical model observation, but these methods have many limitations.

[0003] As a classic method of anatomical research, cadaver dissection can provide the most intuitive and authentic information about human structure, but it is strictly limited by factors such as ethics, sample scarcity, and preservation conditions. The operation process is complicated, requires special dissection sites and equipment, and is difficult to carry out on a large scale, which cannot meet the growing needs of medical education and research. In addition, the irreversibility of cadaver dissection makes it impossible for students or researchers to conduct multiple observations and operations on the same specimen for different purposes, and the efficiency of learning and research is relatively low.

[0004] Although virtual dissection technology has made some progress, it still faces some challenges. For example, in the process of image acquisition, how to obtain higher resolution and more accurate image data to reduce the impact of artifacts and noise on three-dimensional reconstruction; in the construction of three-dimensional models, how to improve the accuracy and realism of the model to better restore the complex structure and physical properties of human tissue; in the process of interactive operation, how to optimize the user experience, make the operation more natural and smooth, and provide richer and more effective feedback information, etc. Therefore, it is of great practical significance to develop a more efficient, accurate and easy-to-use virtual dissection method. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a virtual dissection method based on the VTK visualization framework in view of the deficiencies in the prior art.

[0006] The technical solution of the present invention to solve the above technical problem is as follows: a virtual dissection method based on the VTK visualization framework, comprising the following steps: S1, scanning the appearance of the target corpse to obtain a plurality of original slices, and constructing a three-dimensional voxel model of the body with the plurality of scanned original slices; S2, performing angiography on the target corpse, acquiring a vascular image of the target corpse, and performing three-dimensional reconstruction of the whole body blood vessels to obtain a three-dimensional voxel model of the blood vessels; S3, combining the body 3D voxel model and the blood vessel 3D voxel model to obtain a 3D image of the target corpse structure; S4. Perform virtual anatomical processing on the three-dimensional image of the target cadaver structure based on the VTK visualization framework and display the anatomical image.

[0007] The beneficial effects of the present invention are: providing a systematic and comprehensive virtual dissection solution, covering all aspects from data collection to final result display, so that the virtual dissection process has consistency and integrity; using advanced imaging technology and image processing algorithms, it can accurately reconstruct the three-dimensional structure of the corpse, providing a highly realistic virtual dissection model for medical research, teaching and clinical practice, and helping to improve work efficiency and accuracy in related fields; based on the VTK visualization framework, efficient interactive operation and visualization display are achieved, and users can easily perform virtual dissection operations and observe the dissection results in real time, enhancing the understanding and analysis capabilities of anatomical structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A schematic flow chart of a virtual dissection method based on a VTK visualization framework provided in an embodiment of the present invention; Figure 2 A schematic flow chart of a process for constructing a three-dimensional voxel model of a body provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0009] The principles and features of the present invention are described below in conjunction with the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0010] Example 1: Figure 1 As shown, the present invention provides a virtual dissection method based on the VTK visualization framework, comprising the following steps: S1, scanning the appearance of the target corpse to obtain a plurality of original slices, and constructing a three-dimensional voxel model of the body with the plurality of scanned original slices; S2, performing angiography on the target corpse, acquiring a vascular image of the target corpse, and performing three-dimensional reconstruction of the whole body blood vessels to obtain a three-dimensional voxel model of the blood vessels; S3, combining the body 3D voxel model and the blood vessel 3D voxel model to obtain a 3D image of the target corpse structure; S4. Perform virtual anatomical processing on the three-dimensional image of the target cadaver structure based on the VTK visualization framework and display the anatomical image.

[0011] This embodiment describes in detail the complete process of implementing virtual dissection based on the VTK visualization framework. First, by scanning the appearance of the target corpse, a plurality of original slices are obtained, which is the basic data source for building a virtual dissection model.

[0012] Next, a series of processing is performed on the collected original slices to improve the image quality and prepare for subsequent 3D reconstruction. In terms of angiography, CTA tomography angiography is used to obtain vascular images, and a specific algorithm is used to reconstruct the vascular 3D. Then the body 3D voxel model and the vascular 3D voxel model are spliced ​​to obtain a complete 3D image of the target corpse structure. Finally, based on the VTK visualization framework, virtual anatomical processing of the 3D image of the target corpse structure is realized, and the anatomical results are displayed in real time, providing users with an intuitive anatomical operation experience.

[0013] In this embodiment, a systematic and comprehensive virtual dissection solution is provided, covering all aspects from data collection to final result display, so that the virtual dissection process has coherence and integrity. Using advanced imaging technology and image processing algorithms, the three-dimensional structure of the corpse can be accurately reconstructed, providing a highly realistic virtual dissection model for medical research, teaching and clinical practice, which helps to improve the work efficiency and accuracy in related fields. Based on the VTK visualization framework, efficient interactive operation and visualization display are realized. Users can easily perform virtual dissection operations and observe the dissection results in real time, which enhances the understanding and analysis ability of anatomical structures.

[0014] Example 2: Figure 2 As shown, in S1, the appearance of the target corpse is image scanned to obtain a plurality of original slices, and a three-dimensional voxel model of the body is constructed from the plurality of original slices obtained by scanning, specifically: S1.1. Scan the target body based on CT scanning to obtain multiple original slices, establish a unified coordinate system in each original slice, and obtain an original slice sequence of the target body; S1.2, pre-processing the original slice sequence by filtering and grayscale correction; S1.3, performing registration processing from a low-resolution image to a high-resolution image on the preprocessed original slice sequence based on the coordinate system to obtain a high-resolution slice sequence; S1.4, determining the position of the three-dimensional space based on the high-resolution slice sequence, dividing the three-dimensional space into a plurality of cubic grids, and performing value assignment processing on each cubic grid; S1.5. Construct a three-dimensional voxel model of the shape based on each cube grid after the assignment.

[0015] In this embodiment, when performing an image scan of the appearance of the target corpse, based on the principle of CT scanning, a detailed description is given on how to obtain multiple original slices. CT scanning uses X-rays to perform tomographic scanning of the corpse, which can clearly display the structure of hard tissues such as bones; a unified coordinate system is established on each original slice to ensure that the positional relationship between different slices is accurate, thereby obtaining the original slice sequence of the target corpse.

[0016] For the original slice sequence, preprocessing operations of filtering and grayscale correction are performed. The filtering process uses a 5×5 rectangular area as the filtering window, and removes the noise in the image through methods such as median filtering. The specific operation is to input each original slice in the original slice sequence into the filtering window, determine the neighborhood with the current pixel as the center, extract the grayscale values ​​of all pixels in the neighborhood and sort them, select the pixel in the middle position as the new value after filtering, and complete the filtering and denoising after traversing all pixels. Grayscale correction selects any original slice as the reference image, calculates its grayscale histogram, and then determines the total number of pixels with a grayscale level less than or equal to a certain grayscale level, uses the transformation function to map the pixels of the reference image to the new grayscale level, and then uses the grayscale matching formula to match the grayscale of other original slices with the corrected reference image to ensure the grayscale consistency of the entire slice sequence.

[0017] It should be understood that the present embodiment is not limited to imaging by CT scanning, and imaging may also be performed by other imaging methods, for example, by MRI magnetic resonance imaging.

[0018] In this embodiment, the establishment of a unified coordinate system provides an accurate spatial positioning basis for subsequent operations such as image registration and three-dimensional reconstruction, so that different slices can accurately correspond and merge, and the accuracy of three-dimensional model construction is improved. Filter denoising preprocessing effectively removes noise interference in the image, improves image quality, and makes subsequent feature extraction, segmentation and other processing more accurate and reliable. The grayscale correction operation ensures the consistency of the slice sequence in grayscale, reduces the grayscale differences caused by factors such as imaging conditions, is conducive to more accurate identification and analysis of anatomical structures, and improves the accuracy and reliability of the virtual anatomical model.

[0019] Embodiment 3: In S1.2, the original slice sequence is pre-processed by filtering and denoising, specifically: Set the filter window parameters to a 5×5 rectangular area; Inputting each original slice in the original slice sequence into a filtering window respectively; Starting from the upper left corner of the original slice and centering on the current pixel (i, j), a neighborhood having the same size as the selected window is determined, wherein the neighborhood is a 5×5 pixel area centered on the current pixel (i, j); Extracting the grayscale values ​​of all pixels in the neighborhood, sorting all the grayscale values, and obtaining a pixel sequence; In the pixel sequence, a pixel at the middle position is selected as the new value of the current pixel after filtering. In this process, after traversing all pixels of all the original slices, the preprocessing of filtering and denoising of the original slice sequence is completed.

[0020] In this embodiment, the specific steps and operation details in the filtering and denoising process are further elaborated in detail. After setting the filtering window to a 5×5 rectangular area, for each original slice in the original slice sequence, starting from the upper left corner and centered on the current pixel (i, j), a 5×5 pixel neighborhood is determined. In this neighborhood, the grayscale values ​​of all pixels are extracted, and then these grayscale values ​​are sorted to form a pixel sequence. By selecting the pixel value at the middle position in the pixel sequence as the new value of the current pixel after filtering, this median filtering method can effectively remove impulse noise such as salt and pepper noise, while better retaining the edge and detail information of the image. When all pixels in the original slice are traversed in this way, the filtering and denoising processing of the original slice is completed, and this process is repeated until all slices in the original slice sequence are processed.

[0021] In this embodiment, the median filtering method has a good protection effect on the edges and details of the image while removing noise, avoiding the edge blurring problem that may be caused by the traditional linear filtering method. This is crucial for accurately identifying and retaining detailed information such as organ boundaries and vascular structures in virtual anatomy, and helps to improve the quality of the virtual anatomy model, so that subsequent analysis and operations can be performed based on clearer and more accurate image data, thereby improving the accuracy and reliability of virtual anatomy.

[0022] Embodiment 4: In S1.2, the grayscale correction preprocessing is performed on the original slice sequence, specifically: Taking any original slice in the original slice sequence after filtering as a reference image, determining a grayscale histogram H of the reference image, wherein k represents the grayscale level, the value range of k is 0 to 255, and H(k) represents the number of pixels with grayscale level k; According to the grayscale histogram H, calculate the total number of pixels in the reference image whose grayscale level is less than or equal to grayscale level k: , in, is the total number of pixels whose gray level is less than or equal to gray level k, is a pixel whose gray level is less than or equal to gray level k; The pixels of the reference image are mapped from the original grayscale to the new grayscale through the transformation function to obtain the grayscale corrected reference image: ×(L 1), in, is the total number of pixels of the reference image, L is the total number of gray levels, and the value of L is 256; Assume that the mean of the reference image is , the variance is The mean of the original slice without grayscale correction is , the variance is , grayscale matching is performed on each original slice that has not been grayscale corrected with the grayscale corrected reference image using the grayscale matching formula: , in, is the gray value after correction, is the grayscale value of the reference image.

[0023] In this embodiment, the specific calculation process and principle of grayscale correction are described in depth. First, any original slice in the original slice sequence after filtering is selected as a reference image, and its grayscale histogram H is calculated, where k represents the grayscale level, and the value range is 0 to 255. Indicates the number of pixels with gray level k. According to the gray level histogram, calculate the total number of pixels in the reference image whose gray level is less than or equal to a certain gray level k , obtained by accumulating the number of pixels with gray levels less than or equal to k. Then, using the transformation function The pixels of the reference image are mapped from the original grayscale to the new grayscale. In the transformation function, N is the total number of pixels in the reference image, and L is the total number of grayscale levels (value 256). The mapping process is based on the cumulative distribution function, which makes the grayscale distribution of the image more uniform.

[0024] Assume that the mean of the reference image is , the variance is The mean of the original slice without grayscale correction is , the variance is σi, and the grayscale matching formula g (x, y) is used to match the grayscale of each original slice that has not been grayscale corrected with the grayscale corrected reference image. In the formula, g (x, y) is the grayscale value after correction, and f (x, y) is the grayscale value of the reference image. By considering the mean and variance relationship between the reference image and the slice to be corrected, the grayscale correction based on statistical characteristics is realized, so that the grayscale between different slices has better comparability.

[0025] In this embodiment, the grayscale correction process adjusts the grayscale distribution of the image to make the grayscale between different slices more consistent, thereby improving the contrast and clarity of the image and helping to more accurately display the details of the anatomical structure. The correction method based on the grayscale histogram and statistical characteristics can adapt to different image features, effectively reduce the grayscale differences caused by factors such as imaging equipment and scanning conditions, and provide a more stable and reliable grayscale foundation for subsequent image processing and analysis, thereby improving the accuracy and reliability of the construction of the virtual anatomical model, and facilitating the accurate identification and analysis of anatomical structures in medical research and clinical diagnosis.

[0026] Embodiment 5: In S1.3, the preprocessed original slice sequence is subjected to registration processing from a low-resolution image to a high-resolution image based on the coordinate system to obtain a high-resolution slice sequence, specifically: Perspective deformation processing is performed based on the original slices in the preprocessed original slice sequence, including: The SIFT scale-invariant feature transformation algorithm is used to construct different image scale spaces from the preprocessed original slices, and extreme point detection is performed on different image scale spaces; The similarity values ​​of the extreme points corresponding to different original slices are calculated by Euclidean distance, and all similarity values ​​are judged according to the set similarity threshold to obtain the matching feature matching point pairs in different original slices; Substitute all feature matching point pairs into the perspective transformation model through the constructed perspective transformation matrix, perform perspective deformation processing on the original slice, and obtain the affine transformation parameter matrix: , in, is the element of the perspective transformation matrix; (x, y) is the coordinate of the point on the original slice, and the corresponding homogeneous coordinate is expressed as (x, y, 1); ( , ) is the point coordinate after perspective transformation, and the corresponding homogeneous coordinate is expressed as ( , ,ω); The mapped pixel values ​​in each low-resolution image in the original slice image sequence are determined according to the affine transformation parameter matrix, the pixel values ​​are interpolated by a bilinear interpolation method to generate a corresponding high-resolution image, and a high-resolution slice image sequence is obtained according to all the high-resolution images.

[0027] In this embodiment, the process of registering the low-resolution image to the high-resolution image based on the coordinate system for the preprocessed original slice sequence is described in detail. First, the SIFT scale-invariant feature transformation algorithm is used to construct different image scale spaces for the preprocessed original slices, and extreme point detection is performed on different scale spaces. The SIFT algorithm constructs a Gaussian difference pyramid to find extreme points in the image at different scales. These extreme points are scale and rotation invariant and can represent significant features in the image.

[0028] Then, the similarity values ​​of the extreme points corresponding to different original slices are calculated by Euclidean distance, and all similarity values ​​are judged according to the set similarity threshold to obtain the feature matching point pairs matching in different original slices. This feature point-based matching method can find corresponding feature points between images of different resolutions, providing a basis for subsequent image registration.

[0029] Next, all feature matching point pairs are substituted into the perspective transformation model through the constructed perspective transformation matrix, and the original slice is perspectively deformed to obtain the affine transformation parameter matrix. The perspective transformation matrix is ​​used to describe the perspective deformation relationship between images. By substituting the feature matching point pairs into the model, the elements in the matrix can be solved to determine the transformation relationship between the images. Finally, the mapped pixel values ​​of each low-resolution image in the original slice image sequence are determined according to the affine transformation parameter matrix, and the pixel values ​​are interpolated using the bilinear interpolation method to generate the corresponding high-resolution image, and finally a high-resolution slice image sequence is obtained. The bilinear interpolation method interpolates the pixel values ​​of the mapped non-integer coordinate positions by considering the values ​​of the four nearest neighboring pixels, making the generated high-resolution image smoother and more natural.

[0030] In this embodiment, the application of the SIFT algorithm enables accurate detection and matching of feature points between images of different resolutions and possible deformations, improves the accuracy and robustness of image registration, and can find reliable matching points even when the images undergo certain degrees of rotation, scaling, and translation. The calculation of the perspective transformation and affine transformation parameter matrices can accurately describe the geometric deformation relationship between images. The high-resolution images generated by bilinear interpolation not only improve the resolution of the images, but also maintain the continuity and detail information of the images, providing high-quality image data for subsequent three-dimensional model construction, helping to improve the accuracy and realism of virtual anatomical models, and being able to more accurately display the details and spatial relationships of anatomical structures in medical research and teaching.

[0031] Embodiment 6: In S1.4, the position of the three-dimensional space is determined based on the high-resolution slice sequence, the three-dimensional space is divided into a plurality of cubic grids, and each cubic grid is assigned a value, specifically: Any high-resolution slice of the high-resolution slice sequence is used as the high-resolution slice to be divided, the upper left corner pixel point of the high-resolution slice to be divided is used as the origin of the three-dimensional space coordinate system, and a three-dimensional coordinate axis is established, and the directions of the X, Y and Z axes are determined, and the scale of the new coordinate axis is determined according to the resolution and actual physical size of the target high-resolution slice, and the three-dimensional space is divided into a plurality of cubic grids of equal size along the X, Y and Z axes according to the three-dimensional coordinate axis and the new coordinate axis scale according to the set size, in this way, all the high-resolution slices in the high-resolution slice sequence are divided; Assume that the grayscale value range in the high-resolution slice is [gmin, gmax], and determine the attribute value related to the corresponding organ as the grid assignment according to the proportion of pixel values ​​in the cube grid falling within the grayscale value range. In this way, all cube grids are assigned values.

[0032] In this embodiment, when determining the three-dimensional space position and dividing the cubic grid based on the high-resolution slice sequence, any high-resolution slice in the high-resolution slice sequence is used as the high-resolution slice to be divided, and the upper left corner pixel point is used as the origin of the three-dimensional space coordinate system to establish the three-dimensional coordinate axis (X, Y, Z axis). The scale of the new coordinate axis is determined according to the resolution and actual physical size of the target high-resolution slice. For example, if it is known that the pixel spacing of the slice is dx mm / pixel in the X and Y directions, and the distance between adjacent slices in the Z direction is dz mm, then in the three-dimensional space, the actual physical length of a pixel in the X and Y directions is dx mm, and the actual physical distance between two adjacent cubic grids in the Z axis direction is dz mm.

[0033] The three-dimensional space is divided into multiple cube grids of equal size according to the set size along the X, Y and Z axes, and the same operation is performed on all slices in the high-resolution slice sequence. For each cube grid, the grayscale value range in the high-resolution slice is set to [gmin, gmax]. The attribute value related to the corresponding organ is determined as the grid assignment according to the proportion of pixel values ​​in the cube grid that fall within the grayscale value range. For example, if most of the pixel values ​​in a cube grid fall within the interval representing the grayscale value range of the liver, then an attribute value related to the liver (such as a density estimate, etc.) can be calculated based on the distribution of these pixel values ​​as the grid assignment, and the assignment processing of all cube grids is completed in this way.

[0034] In this embodiment, the three-dimensional coordinate system and cubic grid division established based on the high-resolution slice sequence provide a structured framework for accurately representing the anatomical structure in three-dimensional space, so that each cubic grid can correspond to a specific spatial position and anatomical tissue area. The method of determining the grid assignment according to the grayscale value range can convert the grayscale information in the image into attribute values ​​related to the organ, and give physiological information to the subsequently constructed three-dimensional voxel model, which helps to more intuitively represent and analyze the characteristics of the anatomical structure. This precise three-dimensional space representation and assignment method provides a detailed and accurate model foundation for virtual anatomy, which is conducive to improving the understanding and research of anatomical structures, and has important application value in the fields of medical education and surgical planning.

[0035] Embodiment 7: In S1.5, a three-dimensional voxel model of a shape is constructed based on each cube grid after the assignment, specifically: The grid assignment and basic parameter values ​​corresponding to each cubic grid are stored in the form of a three-dimensional array to obtain voxel data; The VTK visualization framework is called, and a vtkImageData object and a vtkRenderer object are created in the VTK visualization framework. The vtkImageData object is used to store each voxel data, and the vtkRenderer object is used to convert the processing object from its own local coordinate system to the scene global coordinate system according to the voxel data to ensure that different objects are correctly laid out in the same space, and to perform rendering processing to obtain a three-dimensional voxel model of the shape.

[0036] In this embodiment, the grid assignment and basic parameter values ​​corresponding to each cube grid are stored in the form of a three-dimensional array to obtain voxel data, which fully describes the attribute information of each position in the three-dimensional space. The VTK visualization framework is called to create a vtkImageData object and a vtkRenderer object. The vtkImageData object is used to store each voxel data. It provides an efficient data structure to manage and access the three-dimensional voxel array, ensuring the reasonable storage and fast retrieval of voxel data in memory.

[0037] The vtkRenderer object plays a key role in the virtual dissection process. It converts the processed object from its own local coordinate system to the scene global coordinate system based on the voxel data to ensure that different objects (such as different organs, tissues, etc.) can be correctly laid out in the same space. During the rendering process, the vtkRenderer object is responsible for processing the settings of rendering parameters such as lighting, materials, and colors, and converting voxel data into visual image outputs. For example, through rendering algorithms such as ray casting and texture mapping, the three-dimensional voxel model is rendered into a realistic image, thereby obtaining a visualization result of the three-dimensional voxel model of the body, allowing users to intuitively observe and analyze the constructed three-dimensional model.

[0038] In this embodiment, the method of storing voxel data in a three-dimensional array facilitates data management and processing, improves data storage efficiency and access speed, and provides convenience for subsequent visualization and analysis. The collaborative work of vtkImageData and vtkRenderer objects in the VTK visualization framework realizes efficient three-dimensional model rendering and visualization, and can convert complex voxel data into intuitive and realistic image displays, greatly enhancing the user's understanding and perception of anatomical structures. Through precise coordinate system conversion and rendering processing, the accuracy and authenticity of the three-dimensional model in spatial layout and visual effects are guaranteed, providing high-quality visualization support for virtual anatomy, which is helpful for the display, analysis and communication of anatomical structures in medical research, teaching and clinical practice.

[0039] Embodiment 8: In S2, angiography is performed on the target corpse to acquire a vascular image of the target corpse, and three-dimensional reconstruction of the whole body blood vessels is performed to obtain a three-dimensional voxel model of the blood vessels, specifically: Performing angiography on the target corpse based on CTA tomography angiography, acquiring images according to a standardized scanning protocol, and obtaining a vascular image of the target corpse; The adaptive threshold segmentation algorithm based on region growing selects several seed points in the vascular image, and calculates the adaptive growth threshold according to the grayscale mean and standard deviation of the area around the seed points, so that the region growing process is carried out along the vascular boundary until the entire vascular area is segmented; The vascular region is converted into a vascular voxel representation, and a vascular 3D voxel model is constructed based on the vascular voxel representation.

[0040] In this embodiment, when performing angiography on the target corpse using the CTA tomographic angiography method, image acquisition is performed according to a standardized scanning protocol to ensure that high-quality and consistent vascular images are obtained. The standardized scanning protocol includes precise control of scanning parameters (such as scanning layer thickness, layer spacing, scanning speed, contrast agent injection dose and rate, etc.) to ensure that the morphology, structure and blood flow of the blood vessels can be clearly displayed.

[0041] After acquiring the vascular image, an adaptive threshold segmentation algorithm based on region growing is used to segment the vascular region. First, several seed points are selected in the vascular image. The selection of these seed points can be based on prior knowledge or automatic detection algorithms. For example, the area on the center line of the vessel or with obvious vascular features is selected as the seed point. Then, the adaptive growth threshold is calculated based on the grayscale mean and standard deviation of the area around the seed point. The region growing process gradually expands from the seed point based on this threshold, and pixels similar to the seed point (within the grayscale value range) are merged into the growth area until the entire vascular area is segmented. This adaptive threshold segmentation algorithm can automatically adjust the threshold according to the grayscale characteristics of the local area of ​​the image, adapt to the grayscale changes in different vascular areas, and improve the accuracy of vascular segmentation.

[0042] Finally, the segmented vascular area is converted into a vascular voxel representation, and a vascular three-dimensional voxel model is constructed based on the vascular voxel representation. That is, each pixel in the vascular area is mapped to a voxel in three-dimensional space, and the voxel is assigned a value according to the relevant properties of the blood vessel (such as contrast agent concentration, vascular wall thickness, etc.), thereby constructing a voxel model that can accurately reflect the three-dimensional structure of the blood vessel.

[0043] It should be understood that the present embodiment is not limited to imaging by CTA tomography angiography, and other imaging methods may also be used, for example, MRA magnetic resonance angiography. MRA mainly uses the blood flow effect in the principle of MRI magnetic resonance imaging for imaging.

[0044] In this embodiment, the CTA tomographic angiography method combined with the standardized scanning protocol can obtain high-resolution and clear vascular images, providing a high-quality data basis for accurate vascular segmentation and three-dimensional reconstruction. The adaptive threshold segmentation algorithm based on region growing effectively overcomes the limitations of the traditional fixed threshold segmentation method in processing vascular images, can more accurately extract vascular regions, reduce mis-segmentation and missed segmentation, and improve the accuracy of vascular three-dimensional voxel model construction. Accurate vascular three-dimensional voxel models are of great significance for the study of vascular diseases, surgical planning (such as vascular interventional surgery, cardiovascular surgery, etc.), and vascular-related medical education. It can help doctors better understand vascular structure and pathological conditions and formulate more reasonable treatment plans.

[0045] Embodiment 9: In the above S3, the body three-dimensional voxel model and the blood vessel three-dimensional voxel model are combined to obtain a three-dimensional image of the target corpse structure, specifically: In the body three-dimensional voxel model and the blood vessel three-dimensional voxel model, the vertex of the target corpse head is used as the origin to determine a unified coordinate system; According to the unified coordinate system, the voxel values ​​in the three-dimensional voxel model of the blood vessel are superimposed on the corresponding voxel positions of the three-dimensional voxel model of the body; The Gaussian filtering method is used to smooth the interface between the vascular 3D voxel model and the body 3D voxel model to obtain a 3D image of the target corpse structure.

[0046] In this embodiment, when the body 3D voxel model and the blood vessel 3D voxel model are combined, the vertices of the head of the target corpse are used as the origin to determine a unified coordinate system. This choice provides a common reference frame for model combination, so that the two models can be accurately aligned in space. According to the unified coordinate system, the voxel values ​​in the blood vessel 3D voxel model are superimposed on the voxel positions corresponding to the body 3D voxel model, realizing the fusion of the two models in space, so that the blood vessel structure can be accurately distributed inside the body structure represented by the body model, reflecting the real anatomical relationship.

[0047] In order to deal with the discontinuity or unnatural transition problems that may occur at the junction of the spliced ​​models, the Gaussian filtering method is used to smooth the junction of the vascular 3D voxel model and the body 3D voxel model. Gaussian filtering calculates the weighted average of the voxel values ​​around the junction and determines the weight according to the Gaussian function, making the voxel value transition at the junction smoother, reducing the traces of model splicing, and improving the overall quality and realism of the 3D image of the target corpse structure after splicing.

[0048] In this embodiment, the determination of a unified coordinate system ensures the accurate fusion of the body and blood vessel models in space, avoids problems such as model misalignment caused by inconsistent coordinate systems, and enables the assembled model to accurately reflect the real anatomical structure of the corpse, providing a reliable basis for subsequent virtual anatomical processing. The smoothing of the junction by Gaussian filtering significantly improves the quality of the assembled model, making the model more natural and continuous visually, and enhancing the realism and usability of the virtual anatomical model. This high-quality model assembly helps to more accurately display and analyze the relationship between blood vessels and surrounding tissues in the human anatomical structure in medical research, teaching, and clinical practice, and provides a more accurate and intuitive tool for work in related fields.

[0049] Embodiment 10: In S4, performing virtual anatomical processing on the three-dimensional image of the target cadaver structure based on the VTK visualization framework and displaying the anatomical image includes: Call the VTK visualization framework and create a vtkRenderWindow object in the VTK visualization framework. The vtkRenderWindow object is used to set the window size for displaying the three-dimensional image of the target cadaver structure and receive anatomical operation instructions from an external device. The three-dimensional image of the target cadaver structure is dissected according to the anatomical operation instructions, and the three-dimensional image of the target cadaver structure after dissection is sent to the vtkRenderer object. The vtkRenderer object is used to obtain voxel data from the three-dimensional image of the target cadaver structure after dissection, perform rendering processing according to the voxel data, and send the rendered anatomical image to the vtkRenderWindow object for display.

[0050] In this embodiment, after the VTK visualization framework is called to create a vtkRenderWindow object, the object can not only set the window size for displaying the three-dimensional image of the target cadaver structure, but also handle a variety of display-related parameters. For example, the background color of the window can be adjusted to provide a more appropriate visual contrast in different anatomical scenes and enhance the visibility of the anatomical structure. At the same time, it supports setting the display mode of the window, such as full-screen mode, window mode, and different resolution options to meet the user's usage needs in different working environments and devices.

[0051] In terms of receiving anatomical operation instructions from external devices, its functions are more diverse. For mouse operation, in addition to the basic click and drag for selecting anatomical parts and performing cutting operations, the mouse wheel can also be used to zoom in and out of the three-dimensional image, which is convenient for users to observe the anatomical structures at different levels and parts in detail. Right-clicking the mouse will pop up a shortcut menu, providing functional options such as saving the current view, switching different anatomical tools, and adjusting display parameters (such as brightness, contrast, transparency, etc.). For keyboard operation, in addition to the conventional tool switching and view adjustment shortcuts, hot keys can also be set to quickly perform specific anatomical operations, such as one-click display of specific organ systems (such as cardiovascular system, nervous system, etc.) and quick switching of different anatomical perspectives (such as front, side, cross-section, etc.).

[0052] When performing dissection operations on the three-dimensional image of the target corpse structure according to the dissection operation instructions, the system performs complex and precise processing internally. Taking the cutting operation as an example, when the user specifies the cutting path with the mouse, the system first determines the intersection area of ​​the cutting plane and the model based on the voxel data of the three-dimensional image, and then processes the voxels in the intersection area. According to the settings of the cutting tool (such as scalpels, electric saws and other simulation tools), the physical responses of different tissues during the cutting process are calculated, such as simulating the elastic deformation and cutting resistance of soft tissues, and the fracture mode and cross-sectional morphology of hard tissues (such as bones). During the cutting process, the voxel data is updated in real time, not only changing the voxel value of the cutting site to show the effect after dissection, but also adjusting the voxel information of the surrounding tissues accordingly, such as simulating bleeding effects (for vascular tissues) or tissue fluid exudation (for other soft tissues), making the dissection process more realistic.

[0053] After the 3D image of the target cadaver structure after dissection is sent to the vtkRenderer object, the vtkRenderer object performs a series of rendering operations. In terms of lighting effect calculation, it considers the combined effects of multiple light source types (such as point light source, parallel light, ambient light, etc.), and calculates the reflection, refraction and shadow effects of each voxel under lighting according to the position, intensity and color of the light source, as well as the material properties and surface normal direction of the model, so that the anatomical structure presents a three-dimensional and layered sense. For the update of material properties, the color, glossiness, transparency and other parameters of the material are adjusted according to the tissue changes after the dissection operation. For example, the surface of the cut tissue may change color and glossiness due to the adhesion of blood or tissue fluid, and the system will update the material properties accordingly to reflect this change. In terms of color mapping, according to different tissue types and anatomical states, the color mapping scheme is dynamically adjusted so that normal tissue, diseased tissue, dissected structures, etc. can be clearly distinguished by color. For example, normal liver tissue may appear brown-red, while the diseased area may appear in different warning colors depending on the type of lesion (such as tumor tissue appearing as highlighted yellow). The interior of a dissected blood vessel may appear red (simulating blood), while the blood vessel wall may appear in different shades of white or gray depending on its thickness and condition.

[0054] The rendered anatomical image is sent to the vtkRenderWindow object for display again. The display effect at this time can not only accurately reflect the results of the anatomical operation, but also has a high sense of reality and visual impact. Users can observe the morphological changes of tissues, the adjustment of spatial relationships, and the simulation of various physiological effects in the display window in real time during the dissection process, providing an intuitive, accurate and vivid anatomical information display platform for medical research, teaching and clinical practice.

[0055] In this embodiment, the user can easily perform various anatomical operations on the three-dimensional image of the target cadaver structure by receiving various operation instructions from external devices (such as a mouse, keyboard, etc.) through the vtkRenderWindow object. For example, the use of mouse clicks, drags, scroll wheel operations, and keyboard shortcuts makes operations such as selecting anatomical parts, performing cuts, zooming, switching tools and viewing angles simple and intuitive. This highly interactive operation method is similar to the manual operation in the actual dissection process, which greatly reduces the user's learning cost and improves the operation efficiency, allowing medical professionals and students to quickly get started and skillfully use the virtual dissection system for learning and research.

[0056] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0057] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0058] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A virtual dissection method based on VTK visualization framework, characterized in that: The steps include: S1, performing image scanning on the appearance of the target corpse to obtain a plurality of original slices, and constructing a three-dimensional voxel model of the body with the plurality of scanned original slices; S2, performing angiography on the target corpse, acquiring a vascular image of the target corpse, and performing three-dimensional reconstruction of the whole body blood vessels to obtain a three-dimensional voxel model of the blood vessels; S3, combining the body 3D voxel model and the blood vessel 3D voxel model to obtain a 3D image of the target corpse structure; S4. Perform virtual anatomical processing on the three-dimensional image of the target cadaver structure based on the VTK visualization framework and display the anatomical image.

2. The virtual dissection method according to claim 1, characterized in that: In S1, the appearance of the target corpse is scanned to obtain a plurality of original slices, and a three-dimensional voxel model of the body is constructed from the plurality of scanned original slices, specifically: S1.

1. Scan the target body based on CT scanning to obtain multiple original slices, establish a unified coordinate system in each original slice, and obtain an original slice sequence of the target body; S1.2, pre-processing the original slice sequence by filtering and grayscale correction; S1.3, performing registration processing from a low-resolution image to a high-resolution image on the preprocessed original slice sequence based on the coordinate system to obtain a high-resolution slice sequence; S1.4, determining the position of the three-dimensional space based on the high-resolution slice sequence, dividing the three-dimensional space into a plurality of cubic grids, and performing value assignment processing on each cubic grid; S1.

5. Construct a three-dimensional voxel model of the shape based on each cube grid after the assignment.

3. The virtual dissection method according to claim 2, characterized in that: In S1.2, the original slice sequence is pre-processed by filtering and denoising, specifically: Set the filter window parameters to a 5×5 rectangular area; Inputting each original slice in the original slice sequence into a filtering window respectively; Starting from the upper left corner of the original slice and centering on the current pixel (i, j), a neighborhood having the same size as the selected window is determined, wherein the neighborhood is a 5×5 pixel area centered on the current pixel (i, j); Extracting the grayscale values ​​of all pixels in the neighborhood, sorting all the grayscale values, and obtaining a pixel sequence; In the pixel sequence, a pixel at the middle position is selected as the new value of the current pixel after filtering. In this process, after traversing all pixels of all the original slices, the preprocessing of filtering and denoising of the original slice sequence is completed.

4. The virtual dissection method according to claim 3, characterized in that: In S1.2, the grayscale correction preprocessing is performed on the original slice sequence, specifically: Taking any original slice in the original slice sequence after filtering as a reference image, determining a grayscale histogram H of the reference image, wherein k represents the grayscale level, the value range of k is 0 to 255, and H(k) represents the number of pixels with grayscale level k; According to the grayscale histogram H, calculate the total number of pixels in the reference image whose grayscale level is less than or equal to grayscale level k: ， in, is the total number of pixels whose gray level is less than or equal to gray level k, is a pixel whose gray level is less than or equal to gray level k; The pixels of the reference image are mapped from the original grayscale to the new grayscale through the transformation function to obtain the grayscale corrected reference image: ×(L 1)， in, is the total number of pixels of the reference image, L is the total number of gray levels, and the value of L is 256; Assume that the mean of the reference image is , the variance is The mean of the original slice without grayscale correction is , the variance is , grayscale matching is performed on each original slice that has not been grayscale corrected with the grayscale corrected reference image using the grayscale matching formula: ， in, is the gray value after correction, is the grayscale value of the reference image.

5. The virtual dissection method according to claim 1, characterized in that: In S1.3, the preprocessed original slice sequence is subjected to registration processing from a low-resolution image to a high-resolution image based on the coordinate system to obtain a high-resolution slice sequence, specifically: Perspective deformation processing is performed based on the original slices in the preprocessed original slice sequence, including: The SIFT scale-invariant feature transformation algorithm is used to construct different image scale spaces from the preprocessed original slices, and extreme point detection is performed on different image scale spaces; The similarity values ​​of the extreme points corresponding to different original slices are calculated by Euclidean distance, and all similarity values ​​are judged according to the set similarity threshold to obtain the matching feature matching point pairs in different original slices; Substitute all feature matching point pairs into the perspective transformation model through the constructed perspective transformation matrix, perform perspective deformation processing on the original slice, and obtain the affine transformation parameter matrix: ， in, is the element of the perspective transformation matrix; (x, y) is the coordinate of the point on the original slice, and the corresponding homogeneous coordinate is expressed as (x, y, 1); ( , ) is the point coordinate after perspective transformation, and the corresponding homogeneous coordinate is expressed as ( , ,ω); The mapped pixel values ​​in each low-resolution image in the original slice image sequence are determined according to the affine transformation parameter matrix, the pixel values ​​are interpolated by a bilinear interpolation method to generate a corresponding high-resolution image, and a high-resolution slice image sequence is obtained according to all the high-resolution images.

6. The virtual dissection method according to claim 1, characterized in that: In S1.4, the position of the three-dimensional space is determined based on the high-resolution slice sequence, the three-dimensional space is divided into a plurality of cubic grids, and each cubic grid is assigned a value, specifically: Any high-resolution slice of the high-resolution slice sequence is used as the high-resolution slice to be divided, the upper left corner pixel point of the high-resolution slice to be divided is used as the origin of the three-dimensional space coordinate system, and a three-dimensional coordinate axis is established, and the directions of the X, Y and Z axes are determined, and the scale of the new coordinate axis is determined according to the resolution and actual physical size of the target high-resolution slice, and the three-dimensional space is divided into a plurality of cubic grids of equal size along the X, Y and Z axes according to the three-dimensional coordinate axis and the new coordinate axis scale according to the set size, in this way, all the high-resolution slices in the high-resolution slice sequence are divided; Assume that the grayscale value range in the high-resolution slice is [gmin, gmax], and determine the attribute value related to the corresponding organ as the grid assignment according to the proportion of pixel values ​​in the cube grid falling within the grayscale value range. In this way, all cube grids are assigned values.

7. The virtual dissection method according to claim 1, characterized in that: In S1.5, a three-dimensional voxel model of the shape is constructed based on each cube grid after the assignment, specifically: The grid assignment and basic parameter values ​​corresponding to each cubic grid are stored in the form of a three-dimensional array to obtain voxel data; The VTK visualization framework is called, and a vtkImageData object and a vtkRenderer object are created in the VTK visualization framework. The vtkImageData object is used to store each voxel data, and the vtkRenderer object is used to convert the processing object from its own local coordinate system to the scene global coordinate system according to the voxel data to ensure that different objects are correctly laid out in the same space, and to perform rendering processing to obtain a three-dimensional voxel model of the shape.

8. The virtual dissection method according to any one of claims 1 to 7, characterized in that: In S2, angiography is performed on the target corpse to acquire a vascular image of the target corpse, and three-dimensional reconstruction of the whole body blood vessels is performed to obtain a three-dimensional voxel model of the blood vessels, specifically: Performing angiography on the target corpse based on CTA tomography angiography, acquiring images according to a standardized scanning protocol, and obtaining a vascular image of the target corpse; The adaptive threshold segmentation algorithm based on region growing selects several seed points in the vascular image, and calculates the adaptive growth threshold according to the grayscale mean and standard deviation of the area around the seed points, so that the region growing process is carried out along the vascular boundary until the entire vascular area is segmented; The vascular region is converted into a vascular voxel representation, and a vascular 3D voxel model is constructed based on the vascular voxel representation.

9. The virtual dissection method according to claim 8, characterized in that: In S3, the body three-dimensional voxel model and the blood vessel three-dimensional voxel model are combined to obtain a three-dimensional image of the target corpse structure, specifically: In the body three-dimensional voxel model and the blood vessel three-dimensional voxel model, the vertex of the target corpse head is used as the origin to determine a unified coordinate system; According to the unified coordinate system, the voxel values ​​in the 3D voxel model of the blood vessel are superimposed on the corresponding voxel positions of the 3D voxel model of the body; The Gaussian filtering method is used to smooth the interface between the vascular 3D voxel model and the body 3D voxel model to obtain a 3D image of the target corpse structure.

10. The virtual dissection method according to claim 9, characterized in that: In S4, virtual anatomical processing is performed on the three-dimensional image of the target cadaver structure based on the VTK visualization framework, and the anatomical image is displayed, including: Call the VTK visualization framework and create a vtkRenderWindow object in the VTK visualization framework. The vtkRenderWindow object is used to set the window size for displaying the three-dimensional image of the target cadaver structure and receive anatomical operation instructions from an external device. The three-dimensional image of the target cadaver structure is dissected according to the anatomical operation instructions, and the three-dimensional image of the target cadaver structure after dissection is sent to the vtkRenderer object. The vtkRenderer object is used to obtain voxel data from the three-dimensional image of the target cadaver structure after dissection, perform rendering processing according to the voxel data, and send the rendered anatomical image to the vtkRenderWindow object for display.

Images