An automatic segmentation method of medical images based on mixed reality technology

Through the automatic segmentation method based on mixed reality technology, including data preprocessing and spectral clustering algorithm, the problems of low efficiency and poor consistency of medical image segmentation are solved, and efficient and accurate medical image segmentation is achieved to meet the needs of modern medical diagnosis.

CN114972386BActive Publication Date: 2024-11-15SHENZHEN UNIV
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202210791510.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-05
Publication Date
2024-11-15
Estimated Expiration
2042-07-05

AI Technical Summary

Technical Problem

The existing medical imaging segmentation methods are inefficient and the segmentation results are inconsistent, making it difficult to meet the needs of modern medical diagnosis for efficient and accurate.

Method used

Using automatic segmentation methods based on mixed reality technology, including data preprocessing, image correction and registration, deterministic and probabilistic fiber tracking, spectral clustering algorithm and PCA algorithm, the image features are realized by building connection matrix and reconstruction mapping. Extract and segment, and finally perform results comparison verification.

Benefits of technology

The efficiency and consistency of image segmentation are improved, and good segmentation effect is obtained, and human organs and lesions can be detected and segmented more accurately.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114972386B_ABST
    Figure CN114972386B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of information processing technology, including a medical image automatic segmentation method based on mixed reality technology, firstly, data preprocessing is performed, specifically including image correction and image registration, and then image deterministic fiber tracking and probabilistic fiber tracking are performed to extract image structure and connection information; image segmentation is performed, and the connection between each voxel point in the region of interest and the whole brain is obtained based on image features; the result of extracting the connection value by using a matrix pair algorithm is downsampled to represent, and a connection matrix of the segmented brain region of interest at the individual spatial level is constructed, and correlation analysis is performed based on the connection information, and the result is used as a feature input of a clustering segmentation algorithm; the mapping is reconstructed by a PCA algorithm, and the distance between the original input x and the reconstructed vector is minimized by the L2 norm, and finally the mapping W is solved; the model is established and trained by an SC algorithm; and a feature image is input. The present invention has good consistency and good segmentation effect.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an automatic medical image segmentation method based on mixed reality technology. Background Art

[0002] Medical image processing includes the process of adjusting the brightness, contrast, edge enhancement, magnification, and editing of the acquired images to achieve the best quality. Medical images are images that reflect the internal structure of the human body and are one of the main bases for modern medical diagnosis. At present, medical image processing tasks are mainly concentrated in four aspects: image detection, image segmentation, image registration, and image fusion.

[0003] Medical image data is available, high-quality, large in volume, and has unified standards, making the application of artificial intelligence relatively mature. Image processing technology is used to analyze and process images to achieve position detection, segmentation extraction, 3D reconstruction, and 3D display of human organs, soft tissues, and lesions. It can also perform qualitative and even quantitative analysis on regions of interest (ROI), thereby greatly improving the efficiency, accuracy, and reliability of clinical diagnosis. It can also play an important auxiliary role in medical teaching, surgical planning, surgical simulation, and various medical research. Therefore, a medical image automatic segmentation method based on mixed reality technology is needed. Summary of the invention

[0004] The purpose of the present invention is to provide a medical image automatic segmentation method based on mixed reality technology;

[0005] The present invention is implemented as follows: the present invention provides a medical image automatic segmentation method based on mixed reality technology;

[0006] S1: First, data preprocessing is performed, including image correction and image registration, and then image deterministic fiber tracking and probabilistic fiber tracking are performed to extract image structure and connection information;

[0007] S2: Perform image segmentation, and obtain the connection between each voxel point in the region of interest and the whole brain based on the image features; use the matrix pair algorithm to downsample the result of the connection value extraction, construct the connection matrix of the segmented brain region of interest at the individual spatial level, and perform correlation analysis based on the connection information. The result is used as the feature input of the clustering segmentation algorithm;

[0008] S3: Reconstruct the mapping through the PCA algorithm, minimize the distance between the original input x and the reconstructed vector through the L2 norm, and finally solve the mapping W;

[0009] S4: Model building and training through SC algorithm;

[0010] S5: Input feature image and perform image segmentation;

[0011] S6: Compare and verify the results.

[0012] Further, in step S1, mutual mapping between images is realized based on transformation rules, as shown in formula (1);

[0013] Formula (1)

[0014] Where Ii and Ij represent the given images; F (.) represents the mapping function of the image; m represents the similarity measurement of the transformation; the goal of registration is to find the mapping transformation function based on the maximum m( ) through the optimization algorithm.

[0015] Further, in step S6, the result comparison and verification specifically includes aligning the image segmentation result to the standard space, calculating the maximum probability segmentation result of the image at the group level, estimating the segmentation mode consistency of the image segmentation result, and comparing and evaluating it with the result of the SCN method; as shown in formula (2);

[0016] Formula (2)

[0017] Among them, Q represents the number of subjects. The segmentation results of the individual diffuse space are converted to the segmentation results of the subject Mq in the standard space through reverse registration. i , i represents the i-th image partition.

[0018] Furthermore, when performing image correction, first, a reference image b0 without diffusion gradient pulse is extracted based on the magnetic resonance image, and then the diffusion signal is modeled based on the Gaussian process with the b0 image as the standard. The distortion and motion distortion signals of the image slices are adjusted and corrected through eddy current correction, and finally the diffusion tensor model of the voxel is calculated to obtain the eigenvalues ​​in different directions and the tensor estimation and diffusion property distribution of each voxel.

[0019] Furthermore, in the process of image standardization, the image normalization tool ANTs is specifically used, which can realize image registration under multimodal conditions based on mean square error, cross-correlation and mutual information.

[0020] Furthermore, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a main controller, the method described in any one of the above is implemented.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] The present invention extracts the biometric information of the image and uses the spectral clustering algorithm to achieve image region segmentation. Experiments show that the algorithm not only makes image segmentation more efficient, but also has better consistency in segmentation results compared with other methods, and has a good segmentation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0024] Figure 1 It is a schematic diagram of the method process of the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the invention claimed for protection, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0026] See also Figure 1 ,An automatic medical image segmentation method based on mixed reality technology,S1: First, data preprocessing is performed, specifically including image correction and image registration, and then image deterministic fiber tracking and probabilistic fiber tracking are performed to extract image structure and connection information;

[0027] S2: Perform image segmentation, and obtain the connection between each voxel point in the region of interest and the whole brain based on the image features; use the matrix pair algorithm to downsample the result of the connection value extraction, construct the connection matrix of the segmented brain region of interest at the individual spatial level, and perform correlation analysis based on the connection information. The result is used as the feature input of the clustering segmentation algorithm;

[0028] S3: Reconstruct the mapping through the PCA algorithm, minimize the distance between the original input x and the reconstructed vector through the L2 norm, and finally solve the mapping W;

[0029] S4: Model building and training through SC algorithm;

[0030] S5: Input feature image and perform image segmentation;

[0031] S6: Compare and verify the results.

[0032] In this embodiment, in step S1, mutual mapping between images is achieved based on transformation rules, such as formula (1);

[0033] Formula (1)

[0034] Where Ii and Ij represent the given images; F (.) represents the mapping function of the image; m represents the similarity measurement of the transformation; the goal of registration is to find the mapping transformation function based on the maximum m( ) through the optimization algorithm.

[0035] In this embodiment, in step S6, the result comparison and verification specifically includes registering the image segmentation result to the standard space, calculating the maximum probability segmentation result of the image at the group level, estimating the segmentation mode consistency of the image segmentation result, and comparing and evaluating it with the result of the SCN method; as shown in formula (2);

[0036] Formula (2)

[0037] Among them, Q represents the number of subjects. The segmentation results of the individual diffuse space are converted to the segmentation results of the subject Mq in the standard space through reverse registration. i , i represents the i-th image partition.

[0038] In this embodiment, when performing image correction, first, a reference image b0 image without a diffusion gradient pulse is extracted based on the magnetic resonance image, and then the b0 image is used as a standard to model the diffusion signal based on the Gaussian process, and the distortion and motion distortion signals of the image slices are adjusted and corrected through eddy current correction. Finally, the diffusion tensor model of the voxel is calculated to obtain the eigenvalues ​​in different directions and the tensor estimation and diffusion property distribution of each voxel.

[0039] In this embodiment, during the image standardization process, an image standardization tool ANTs is specifically used, and image registration under multimodal conditions can be achieved based on mean square error, cross-correlation and mutual information.

[0040] In this embodiment, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the program is executed by the main controller to implement any of the methods described above.

[0041] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A medical image automatic segmentation method based on mixed reality technology, characterized in that: Please follow the steps below: S1: First, data preprocessing is performed, including image correction and image registration, and then image deterministic fiber tracking and probabilistic fiber tracking are performed to extract image structure and connection information; S2: Perform image segmentation, and obtain the connection between each voxel point in the region of interest and the whole brain based on the image features; use the matrix pair algorithm to downsample the result of the connection value extraction, construct the connection matrix of the segmented brain region of interest at the individual spatial level, and perform correlation analysis based on the connection information. The result is used as the feature input of the clustering segmentation algorithm; S3: Reconstruct the mapping through the PCA algorithm, minimize the distance between the original input x and the reconstructed vector through the L2 norm, and finally solve the mapping W; S4: Model building and training through SC algorithm; S5: Input feature image and perform image segmentation; S6: Compare and verify the results; In step S1, mutual mapping between images is achieved based on transformation rules, as shown in formula (1); Formula (1) Where Ii and Ij represent the given images; F (.) represents the mapping function of the image; m represents the similarity measurement of the transformation; the goal of registration is to find the mapping transformation function based on the maximum m () through the optimization algorithm; In step S6, the result comparison and verification specifically includes registering the image segmentation result to the standard space, calculating the maximum probability segmentation result of the image at the group level, estimating the segmentation mode consistency of the image segmentation result, and comparing and evaluating it with the result of the SCN method; as shown in formula (2); Formula (2) Among them, Q represents the number of subjects. The segmentation results of the individual diffuse space are converted to the segmentation results of the subject Mq in the standard space through reverse registration. i , i represents the i-th image partition.

2. The method for automatic segmentation of medical images based on mixed reality technology according to claim 1, characterized in that: When performing image correction, first, the reference image b0 image without diffusion gradient pulse is extracted based on the magnetic resonance image. Then, the b0 image is used as the standard, the diffusion signal is modeled based on the Gaussian process, and the distortion and motion distortion signals of the image slices are adjusted and corrected through eddy current correction. Finally, the diffusion tensor model of the voxel is calculated to obtain the eigenvalues ​​in different directions and the tensor estimation and diffusion property distribution of each voxel.

3. The method for automatic segmentation of medical images based on mixed reality technology according to claim 1, characterized in that: In the process of image standardization, the image normalization tool ANTs is specifically used, which can realize image registration under multimodal conditions based on mean square error, cross-correlation and mutual information.

4. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the main controller, the method according to any one of claims 1 to 2 is implemented.

Citation Information

Patent Citations

  • FMRI and DTI fusion-based vault white matter segmentation method

    CN106204562A

  • Brain limbic system map construction method

    CN108898135A

  • Thalamus function partitioning method based on subspace feature learning

    CN110599461A

  • Neural fiber extraction method, system and device for region of interest and storage medium

    CN111091561A

  • Method and processor for analyzing radioactivity concentration level of whole cerebral cortex

    CN114022463A