High-efficiency three-dimensional visual medical image processing software system

By designing an efficient three-dimensional visual medical image processing software system, using convolutional neural network and graphics rendering technology, the existing system's insufficient processing capabilities and limited complex pathological analysis capabilities are solved, and efficient and accurate three-dimensional medical image processing and Diagnostic support.

CN118967926APending Publication Date: 2024-11-15XINGXIANG PRECISION MEDICAL TECHNOLOGY (CHENGDU) CO LTD
View PDF 0 Cites 9 Cited by

Patent Information

Application Number
CN202411019972.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-29
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing three-dimensional visual medical image processing system lacks response speed and processing capabilities when processing large-scale data. Relying on manual operations is easy to introduce errors, and its analysis ability of complex pathological structures is limited, making it difficult to meet clinical needs.

Method used

A high-efficiency three-dimensional visual medical image processing software system is designed, including data acquisition module, preprocessing module, three-dimensional reconstruction module and intelligent analysis module. The convolutional neural network algorithm is used to automatically identify and segment tumor areas, and generate them in combination with graphics rendering technology. Visualize three-dimensional images and provide a user-friendly operation interface to support multi-dimensional viewing and interactive operations.

Benefits of technology

Significantly improves the efficiency and accuracy of medical imaging processing, reduces the work burden of doctors, provides more intuitive and accurate diagnostic information, improves the flexibility and user experience of the system, can automatically identify and segment tumor areas, and support a variety of Input and automatic resolution of medical image formats.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118967926A_ABST
    Figure CN118967926A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of medical image processing, in particular to a high-efficiency three-dimensional visual medical image processing software system which comprises a data acquisition module, a preprocessing module, a three-dimensional reconstruction module, an intelligent analysis module and a display interaction module. The data acquisition module is used for collecting and acquiring medical image data; the preprocessing module is used for preprocessing the medical image, including data denoising, image enhancement and image registration; through mutual cooperation of the data acquisition module, the preprocessing module, the three-dimensional reconstruction module, the intelligent analysis module and the display interaction module, high-efficiency three-dimensional visual medical image processing is realized, the medical image processing efficiency and precision are improved, more visual information support is provided for medical diagnosis, and the medical diagnosis experience is improved. And the data acquisition module supports the input of a plurality of medical image formats, so that powerful technical support is provided for medical diagnosis.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of medical image processing, and in particular is a highly efficient three-dimensional visualized medical image processing software system. Background Art

[0002] With the rapid development of medical technology, medical imaging plays a vital role in the diagnosis, treatment and prognosis of diseases. Although traditional two-dimensional medical imaging provides doctors with rich diagnostic information, its limitations are gradually revealed in complex cases. The rise of three-dimensional medical imaging technology has greatly improved the accuracy and efficiency of diagnosis, allowing doctors to more intuitively observe the morphology, location and relationship of lesions with surrounding tissues. However, the processing and analysis of three-dimensional medical imaging data still faces many challenges.

[0003] In the existing 3D visualization medical image processing systems, although basic image processing and 3D reconstruction functions have been realized, many systems still rely on manual operations for image preprocessing, 3D reconstruction and subsequent analysis, which is not only time-consuming and labor-intensive, but also prone to human errors. When processing large-scale 3D medical image data, the response speed and processing capabilities of existing systems are often difficult to meet clinical needs, causing doctors to wait for a long time to obtain processing results. Although some systems have begun to try to introduce intelligent analysis technology, most of them are limited to simple image feature extraction and recognition, and lack in-depth analysis and understanding of complex pathological structures. In this regard, we propose an efficient 3D visualization medical image processing software system. Summary of the invention

[0004] In view of the deficiencies of the prior art, the present invention provides a highly efficient three-dimensional visualization medical image processing software system to solve the above technical problems.

[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: an efficient three-dimensional visualization medical image processing software system, the image processing software system comprising: a data acquisition module, a preprocessing module, a three-dimensional reconstruction module, an intelligent analysis module and a display interaction module;

[0006] The data acquisition module is used to collect and acquire medical imaging data;

[0007] The preprocessing module is used to preprocess medical images, including data denoising, image enhancement and image registration;

[0008] The 3D reconstruction module is used to perform 3D reconstruction on the pre-processed data to generate 3D medical images;

[0009] The intelligent analysis module is used to analyze the generated three-dimensional medical images through the convolutional neural network algorithm to automatically identify and segment the tumor area;

[0010] The display interaction module is used to provide a user-friendly operation interface, supporting multi-dimensional viewing, dynamic adjustment and interactive operation.

[0011] Preferably, the data acquisition module supports input of multiple medical image formats, including but not limited to CT, MRI, and PET, and can automatically identify and parse image data in different formats.

[0012] Preferably, the preprocessing module performs preprocessing by:

[0013] A1. Use median filtering method to denoise the traversal volume data image;

[0014] A2. Using contrast method to perform image enhancement processing on the traversal volume data image;

[0015] A3. Use feature point matching method to perform image registration processing on the traversal volume data image.

[0016] Preferably, the three-dimensional reconstruction module generates three-dimensional medical images in the following specific steps:

[0017] B1, the preprocessing module transmits the processed data to the 3D reconstruction module;

[0018] B2. Traverse each cube in the volume data, set the density and intensity of the tumor as thresholds, determine the isosurface in each cube, determine the geometric shape of the isosurface inside the cube through a lookup table according to the state of the cube vertices, use a triangulation algorithm to divide the isosurface into triangular facets, connect the triangular facets in adjacent cubes, and form a continuous three-dimensional surface model;

[0019] B3. Use a smoothing algorithm to remove jagged edges and noise from the generated 3D surface model, and repair topological errors in the model, including holes and cracks. By mapping the texture information in the original image to the 3D model, the realism of the model is increased, and the 3D model is post-processed and optimized.

[0020] B4. The optimized three-dimensional surface model is rendered and visualized using graphic rendering technology to generate a three-dimensional medical image that can be observed and diagnosed by doctors.

[0021] Preferably, the detailed steps of the triangulation algorithm in step B2 include:

[0022] C1. Construct a super triangle;

[0023] C2. Take out points one by one from the original point set, find the triangle containing the new point, delete the common edges of the affected triangle, connect the new point with all the vertices of the affected triangle, and form a new triangle;

[0024] C3. Check whether the circumscribed circles of adjacent triangles contain other points and swap the diagonals.

[0025] C4. In the final segmentation result, delete the edges and triangles related to the super triangle.

[0026] Preferably, the specific implementation method of the intelligent analysis module generating the convolutional neural network algorithm model is:

[0027] D1, passing the 3D medical image generated by the 3D reconstruction module to the display interaction module;

[0028] D2. Professional medical imaging experts annotate 3D medical images to clarify the location and boundaries of the tumor area, provide label data for the training of convolutional neural networks, and increase the diversity of training data through rotation, scaling, and translation operations to improve the generalization ability of convolutional neural networks;

[0029] D3. Design network structure:

[0030] Input layer: receives three-dimensional medical imaging data;

[0031] Convolution layer: extract features of the input image through multiple convolution kernels;

[0032] Activation layer: Use nonlinear activation function to introduce nonlinear factors into the network;

[0033] Pooling layer: downsample the feature map;

[0034] Fully connected layer: converts the features extracted by the convolutional layer into the final output;

[0035] Output layer: a segmentation map of the same size as the input image, where each pixel represents the probability that the point belongs to the tumor area;

[0036] D4. Input the training data into the network, obtain the output results through calculations at each layer of the network, calculate the loss function value based on the output results and the true label, evaluate the prediction performance of the network, update the network weights based on the gradient information of the loss function to reduce the loss value, and repeat the process of forward propagation, loss calculation and back propagation until the network performance reaches the preset standard or the training rounds reach the upper limit.

[0037] Preferably, the calculation formula of the nonlinear activation function of the activation layer in step C3 is:

[0038] f(x)=max(0,x)

[0039] Among them, x is the input data, f(x) is the output result, when x is greater than or equal to 0, f(x) is equal to x, when x is less than 0, f(x) is equal to 0.

[0040] Preferably, the steps of identifying and segmenting the tumor area using the convolutional neural network algorithm model are:

[0041] E1. After the convolutional neural network training is completed, an independent test set is used to evaluate the convolutional neural network model to verify its accuracy and generalization ability in the tumor area segmentation task;

[0042] E2. Input the 3D medical image to be detected into the trained convolutional neural network model, and the model will automatically output the probability map of each pixel belonging to the tumor area;

[0043] E3, performing threshold processing and morphological operation post-processing steps on the probability map to obtain the final tumor region segmentation result;

[0044] E4. Superimpose the segmentation results onto the original three-dimensional medical image and display them to doctors for diagnosis and evaluation through a display interaction module.

[0045] Preferably, the convolutional neural network model needs to be optimized and improved by adjusting the network structure and introducing the attention mechanism according to specific task requirements and data characteristics, and training should be carried out in combination with data from multiple imaging modalities to improve the model's ability to capture tumor features and segmentation accuracy. The algorithm should be optimized through clinical examples to achieve real-time processing of tumor area segmentation and improve the clinical application value of the system.

[0046] Preferably, the image interaction module supports one-click generation of detailed reports containing three-dimensional medical images, annotation information, measurement data and diagnostic conclusions. Users can save or export the reports in multiple formats and share them with other doctors or patients via email or cloud storage.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] The present application realizes efficient 3D visualized medical image processing through the cooperation of data acquisition module, preprocessing module, 3D reconstruction module, intelligent analysis module and display interaction module, significantly improves the efficiency and accuracy of medical image processing, provides more intuitive and accurate information support for medical diagnosis, and helps doctors make more accurate treatment decisions. The data acquisition module supports the input of multiple medical image formats and can automatically identify and parse image data in different formats, greatly broadening the scope of application of the system and improving the flexibility and compatibility of data processing. The preprocessing module performs data denoising through median filtering, effectively reducing noise interference in the image, adopts contrast method to enhance the image, improves the clarity and contrast of the image, and uses feature point matching method for image registration, ensuring that image data from different sources or time series can be accurately aligned, which is conducive to subsequent 3D reconstruction. The 3D reconstruction module can generate accurate 3D medical images, providing doctors with the possibility to observe lesions from any angle, and helping to discover details that are difficult to detect with traditional 2D images. The intelligent analysis module uses a convolutional neural network algorithm to perform in-depth analysis of 3D medical images, realizing automatic identification and precise segmentation of tumor areas, greatly improving the accuracy and efficiency of diagnosis and reducing the workload of doctors. The display interaction module provides a user-friendly operation interface, supports multi-dimensional viewing, dynamic adjustment and interactive operation, allowing doctors to flexibly adjust the image display according to their needs and obtain more comprehensive and in-depth image information. This medical image processing software system not only improves the efficiency and accuracy of medical image processing, but also enhances the flexibility of the system and user experience, providing strong technical support for medical diagnosis and has significant beneficial effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is the framework diagram of this 3D visualization medical image processing software system. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] The present invention provides a technical solution: a highly efficient three-dimensional visualization medical image processing software system, the image processing software system comprising: a data acquisition module, a preprocessing module, a three-dimensional reconstruction module, an intelligent analysis module and a display interaction module;

[0052] The data acquisition module is used to collect and acquire medical imaging data;

[0053] The preprocessing module is used to preprocess medical images, including data denoising, image enhancement and image registration;

[0054] The 3D reconstruction module is used to perform 3D reconstruction on the pre-processed data to generate 3D medical images;

[0055] The intelligent analysis module is used to analyze the generated three-dimensional medical images through the convolutional neural network algorithm to automatically identify and segment the tumor area;

[0056] The display interaction module is used to provide a user-friendly operation interface, supporting multi-dimensional viewing, dynamic adjustment and interactive operation.

[0057] Furthermore, the data acquisition module supports the input of multiple medical imaging formats, including but not limited to CT, MRI, and PET, and can automatically identify and parse imaging data in different formats. Medical imaging data is often stored in multiple formats due to its different imaging principles and application scenarios. The data acquisition module is compatible with imaging files from various manufacturers and different devices, greatly enhancing the compatibility and flexibility of the system.

[0058] Furthermore, the preprocessing module performs preprocessing in the following manner:

[0059] A1. Use the median filter method to denoise the traversal volume data image. The median filter method needs to determine the filter window first. Common window shapes are square and rectangular. The size is selected according to the density and distribution of the noise, usually an odd number, so that the middle value can be directly taken. Then move the filter window to a certain position in the image, read the grayscale values ​​of all pixels in the window, and sort the read pixel grayscale values ​​in ascending order. According to the sorting result of the pixel grayscale values ​​in the window, take the middle value as the new grayscale value of the center pixel of the window, move the filter window row by row and column by column on the image according to the preset step size, and repeat the above steps until all pixels in the image are processed. After the above steps, each pixel is assigned a new grayscale value, thereby generating a denoised image;

[0060] A2. Use contrast method to enhance the traversal volume data image. Contrast enhancement methods include histogram equalization, histogram regulation, linear contrast stretching and nonlinear contrast enhancement. After preliminary enhancement, the image needs to be further processed, including smoothing and quality assessment. It should be noted that when performing contrast enhancement, attention should be paid to parameter settings to avoid excessive enhancement leading to image distortion or loss of key information. During the processing, it is necessary to combine multiple pre-processing and post-processing methods to optimize algorithms and parameters to obtain the best processing effect.

[0061] A3. Use feature point matching method to perform image registration processing on the traversal volume data image. Feature points are usually corner points, edge points or high curvature points in the image. For each extracted feature point, a corresponding feature descriptor is generated. The feature descriptor is a vector that describes the local image features around the feature point, including but not limited to gradient direction and texture information.

[0062] Furthermore, the specific steps of generating a 3D medical image by the 3D reconstruction module are as follows:

[0063] B1, the preprocessing module transmits the processed data to the 3D reconstruction module;

[0064] B2. Traverse each cube in the volume data, set the density and intensity of the tumor as thresholds, determine the isosurface in each cube, determine the geometric shape of the isosurface inside the cube through a lookup table according to the state of the cube vertices, use a triangulation algorithm to divide the isosurface into triangular facets, connect the triangular facets in adjacent cubes, and form a continuous three-dimensional surface model;

[0065] B3. Use a smoothing algorithm to remove jagged edges and noise from the generated 3D surface model, and repair topological errors in the model, including holes and cracks. By mapping the texture information in the original image to the 3D model, the realism of the model is increased, and the 3D model is post-processed and optimized.

[0066] B4. The optimized three-dimensional surface model is rendered and visualized using graphic rendering technology to generate a three-dimensional medical image that can be observed and diagnosed by doctors.

[0067] Furthermore, the detailed steps of the triangulation algorithm in step B2 include:

[0068] C1. Construct a super triangle. To ensure that all the concentrated points are inside the triangle, the vertices of the triangle can be selected on the boundary of the point set or far enough away from the point set.

[0069] C2. Take out points one by one from the original point set, find the triangle containing the new point, delete the common edges of the affected triangle, connect the new point with all the vertices of the affected triangle, and form a new triangle;

[0070] C3. Check whether the circumscribed circles of adjacent triangles contain other points and swap the diagonals.

[0071] C4. In the final segmentation result, delete the edges and triangles related to the super triangle.

[0072] Furthermore, the specific implementation method of the intelligent analysis module generating the convolutional neural network algorithm model is as follows:

[0073] D1, passing the 3D medical image generated by the 3D reconstruction module to the display interaction module;

[0074] D2. Professional medical imaging experts annotate 3D medical images to clarify the location and boundaries of the tumor area, provide label data for the training of convolutional neural networks, and increase the diversity of training data through rotation, scaling, and translation operations to improve the generalization ability of convolutional neural networks;

[0075] D3. Design network structure:

[0076] Input layer: receives 3D medical imaging data, which usually needs to be adjusted to a fixed size that the network can handle;

[0077] Convolution layer: extracts features from the input image through multiple convolution kernels, ensuring that each convolution kernel can capture specific features in the image. As the depth of the network increases, the extracted features are gradually increased from low-level to high-level.

[0078] Activation layer: Use nonlinear activation function to introduce nonlinear factors into the network and enhance the network's expressive ability;

[0079] Pooling layer: downsamples the feature map to reduce the spatial dimension of the data, reduce the amount of calculation, and improve the robustness of the features;

[0080] Fully connected layer: converts the features extracted by the convolutional layer into the final output;

[0081] Output layer: a segmentation map of the same size as the input image, where each pixel represents the probability that the point belongs to the tumor area;

[0082] D4. Input the training data into the network, obtain the output results through calculations at each layer of the network, calculate the loss function value based on the output results and the true label, evaluate the prediction performance of the network, update the network weights based on the gradient information of the loss function to reduce the loss value, and repeat the process of forward propagation, loss calculation and back propagation until the network performance reaches the preset standard or the training rounds reach the upper limit.

[0083] Furthermore, the calculation formula of the nonlinear activation function of the activation layer in step C3 is:

[0084] f(x)=max(0,x)

[0085] Among them, x is the input data, f(x) is the output result, when x is greater than or equal to 0, f(x) is equal to x, when x is less than 0, f(x) is equal to 0.

[0086] Furthermore, the steps of the convolutional neural network algorithm model to identify and segment the tumor area are as follows:

[0087] E1. After the convolutional neural network training is completed, an independent test set is used to evaluate the convolutional neural network model to verify its accuracy and generalization ability in the tumor area segmentation task;

[0088] E2. Input the 3D medical image to be detected into the trained convolutional neural network model, and the model will automatically output the probability map of each pixel belonging to the tumor area;

[0089] E3, performing threshold processing and morphological operation post-processing steps on the probability map to obtain the final tumor region segmentation result;

[0090] E4. Superimpose the segmentation results onto the original three-dimensional medical image and display them to doctors for diagnosis and evaluation through a display interaction module.

[0091] Furthermore, it will be necessary to optimize and improve the convolutional neural network model by adjusting the network structure and introducing the attention mechanism according to specific task requirements and data characteristics, and to combine data from multiple imaging modalities for training to improve the model's ability to capture tumor features and segmentation accuracy. By optimizing the algorithm through clinical examples, real-time processing of tumor area segmentation can be achieved, thereby improving the clinical application value of the system.

[0092] Furthermore, the image interaction module supports one-click generation of detailed reports containing 3D medical images, annotation information, measurement data and diagnostic conclusions. Users can save or export the reports in multiple formats and share them with other doctors or patients via email or cloud storage. The image interaction module allows users to freely switch and view different perspectives of 3D medical images. It also integrates a series of interactive annotation and measurement tools that enable doctors to directly annotate 3D medical images and obtain accurate measurement data. At the same time, it supports manual outlining or correction of tumor areas to improve the accuracy of segmentation results.

[0093] 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.

[0094] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An efficient 3D visualization medical image processing software system, characterized in that: The image processing software system includes: data acquisition module, preprocessing module, 3D reconstruction module, intelligent analysis module and display interaction module; The data acquisition module is used to collect and acquire medical imaging data; The preprocessing module is used to preprocess medical images, including data denoising, image enhancement and image registration; The 3D reconstruction module is used to perform 3D reconstruction on the pre-processed data to generate 3D medical images; The intelligent analysis module is used to analyze the generated three-dimensional medical images through the convolutional neural network algorithm to automatically identify and segment the tumor area; The display interaction module is used to provide a user-friendly operation interface, supporting multi-dimensional viewing, dynamic adjustment and interactive operation.

2. The highly efficient 3D visualization medical image processing software system according to claim 1, characterized in that: The data acquisition module supports the input of multiple medical image formats, including but not limited to CT, MRI, and PET, and can automatically identify and parse image data in different formats.

3. The efficient three-dimensional visualization medical image processing software system according to claim 1, characterized in that: The preprocessing module performs preprocessing as follows: A1. Use median filtering method to denoise the traversal volume data image; A2, using contrast method to perform image enhancement processing on the traversal volume data image; A3. Use feature point matching method to perform image registration processing on the traversal volume data image.

4. The efficient three-dimensional visualization medical image processing software system according to claim 1, characterized in that: The specific steps of the 3D reconstruction module to generate 3D medical images are as follows: B1, the preprocessing module transmits the processed data to the 3D reconstruction module; B2. Traverse each cube in the volume data, set the density and intensity of the tumor as thresholds, determine the isosurface in each cube, determine the geometric shape of the isosurface inside the cube through a lookup table according to the state of the cube vertices, use a triangulation algorithm to divide the isosurface into triangular facets, connect the triangular facets in adjacent cubes, and form a continuous three-dimensional surface model; B3. Use a smoothing algorithm to remove jagged edges and noise from the generated 3D surface model, and repair topological errors in the model, including holes and cracks. By mapping the texture information in the original image to the 3D model, the realism of the model is increased, and the 3D model is post-processed and optimized. B4. The optimized three-dimensional surface model is rendered and visualized using graphic rendering technology to generate a three-dimensional medical image that can be observed and diagnosed by doctors.

5. The efficient three-dimensional visualization medical image processing software system according to claim 4 is characterized in that: The detailed steps of the triangulation algorithm in step B2 include: C1. Construct a super triangle; C2, take out points one by one from the original point set, find the triangle containing the new point, delete the common edges of the affected triangle, connect the new point with all the vertices of the affected triangle, and form a new triangle; C3. Check whether the circumscribed circles of adjacent triangles contain other points and swap the diagonals. C4. In the final segmentation result, delete the edges and triangles related to the super triangle.

6. The efficient three-dimensional visualization medical image processing software system according to claim 1, characterized in that: The specific implementation method of the intelligent analysis module to generate a convolutional neural network algorithm model is as follows: D1, passing the 3D medical image generated by the 3D reconstruction module to the display interaction module; D2. Professional medical imaging experts annotate 3D medical images to clarify the location and boundaries of the tumor area, provide label data for the training of convolutional neural networks, and increase the diversity of training data through rotation, scaling, and translation operations to improve the generalization ability of convolutional neural networks; D3. Design network structure: Input layer: receives three-dimensional medical imaging data; Convolution layer: extract features of the input image through multiple convolution kernels; Activation layer: Use nonlinear activation function to introduce nonlinear factors into the network; Pooling layer: downsample the feature map; Fully connected layer: converts the features extracted by the convolutional layer into the final output; Output layer: a segmentation map of the same size as the input image, where each pixel represents the probability that the point belongs to the tumor area; D4. Input the training data into the network, obtain the output results through calculations at each layer of the network, calculate the loss function value based on the output results and the true label, evaluate the prediction performance of the network, update the network weights based on the gradient information of the loss function to reduce the loss value, and repeat the process of forward propagation, loss calculation and back propagation until the network performance reaches the preset standard or the training rounds reach the upper limit.

7. The efficient three-dimensional visualization medical image processing software system according to claim 6, characterized in that: The calculation formula of the nonlinear activation function of the activation layer in step C3 is: f(x)=max(0,x) Among them, x is the input data, f(x) is the output result, when x is greater than or equal to 0, f(x) is equal to x, when x is less than 0, f(x) is equal to 0.

8. The efficient three-dimensional visualization medical image processing software system according to claim 6, characterized in that: The steps of the convolutional neural network algorithm model to identify and segment the tumor area are: E1. After the convolutional neural network training is completed, an independent test set is used to evaluate the convolutional neural network model to verify its accuracy and generalization ability in the tumor area segmentation task; E2. Input the 3D medical image to be detected into the trained convolutional neural network model, and the model will automatically output the probability map of each pixel belonging to the tumor area; E3, performing threshold processing and morphological operation post-processing steps on the probability map to obtain the final tumor region segmentation result; E4. Superimpose the segmentation results onto the original three-dimensional medical image and display them to doctors for diagnosis and evaluation through the display interaction module.

9. The highly efficient three-dimensional visualization medical image processing software system according to claim 8, characterized in that: In the future, it will be necessary to optimize and improve the convolutional neural network model by adjusting the network structure and introducing the attention mechanism according to specific task requirements and data characteristics, and combine data from multiple imaging modalities for training to improve the model's ability to capture tumor features and segmentation accuracy. Through clinical example optimization algorithms, real-time processing of tumor area segmentation can be achieved to improve the clinical application value of the system.

10. The highly efficient 3D visualization medical image processing software system according to claim 1, characterized in that: The image interaction module supports one-click generation of detailed reports containing 3D medical images, annotation information, measurement data and diagnostic conclusions. Users can save or export reports in multiple formats and share them with other doctors or patients via email or cloud storage.

Citation Information

Cited By

  • Simulated medical image visualization system and method based on MR technology

    CN119229705A

  • Simplified magnetic resonance breast intelligent recognition system based on deep learning

    CN119314655A

  • Automatic sketching system for nasopharyngeal carcinoma tumor

    CN119763782A

  • Three-dimensional medical image processing model construction method and device based on AI algorithm

    CN120279194A

  • Medical knowledge visual propagation method based on 3D medical image and AI interaction

    CN120279213A