Gray matter microstructural imaging method based on high-resolution diffusion magnetic resonance imaging
By using the combination of tensor deterministic algorithm and the MRtrix/FSL platform in diffusion magnetic resonance imaging, the problem of fiber tracking is solved, accurate tracking of extremely fine fibers and high-resolution image analysis is achieved, and early diagnosis of diseases is improved ability.
Patent Information
- Application Number
- CN202411095438.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-12
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-08-12
AI Technical Summary
Diffusion magnetic resonance imaging is difficult to track or mistrack extremely small fiber bundles, especially at the ash-white matter junction and the anatomical location of the sulcus or civic, resulting in missed imaging evaluation in early or subclinical stages of the disease.
Using a method based on high-resolution diffusion magnetic resonance imaging, fiber trajectories are generated by formulating tensor deterministic algorithms, and pre-processing and visual analysis of gray matter microstructure images are combined with MRtrix and FSL platforms to remove noise and artifacts, and eddy current correction and Spline interpolation correction, tracking U-shaped fibers in the junction of gray matter-white matter cortex and tiny fibers in the gray matter cortex.
It realizes stable and reliable tracking of extremely fine fibers, improves the signal-to-noise ratio and spatial resolution of the image, ensures the accuracy and reliability of fiber tracking, and enhances the ability to diagnose early diseases.
Smart Images

Figure CN118628488B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of medical image processing, and in particular to an imaging method of gray matter microstructure images based on high-resolution diffusion magnetic resonance imaging. Background Art
[0002] Diffusion MRI is currently the only MRI method that can perform non-invasive, in vivo quantitative measurements of brain microstructures. However, diffusion MRI currently has the following drawbacks:
[0003] (1) Only thicker white matter fibers can be tracked and displayed. Due to the limitation of spatial resolution, the early DTI imaging technology was unable to track relatively small fiber bundles (especially the U-shaped fibers at the gray-white matter junction and the extremely fine fibers that play a conductive role at the neuronal connection in the gray matter cortex). This can easily lead to missed diagnosis of lesions during imaging evaluation in the early or subclinical stages of the disease.
[0004] (2) It is easy for white matter fibers to turn too far, leading to incorrect tracking of protein fibers. Due to the anatomical position of the sulci or cisterns, some nerve fibers may make large turns or sharp turns. Past post-processing techniques cannot adapt to such large turns, making it easy for incorrect tracking to occur. Summary of the invention
[0005] The technical problem to be solved by the present invention is that fibers cannot be tracked or are tracked incorrectly in diffusion magnetic resonance imaging. The present invention provides an imaging method for gray matter microstructure images based on high-resolution diffusion magnetic resonance imaging. A tensor deterministic algorithm is formulated to generate fiber trajectories, and a gray matter microstructure image area is generated based on the original image of the gray matter microstructure image and the fiber trajectory. The FSL platform is used to perform quantitative analysis of the gray matter microstructure image visualization, thereby achieving fiber tracking.
[0006] In order to achieve the above object, the present invention is implemented by the following technical solutions:
[0007] The imaging method of gray matter microstructure image based on high-resolution diffusion magnetic resonance imaging comprises the following steps:
[0008] Step 1): optimizing a diffusion magnetic resonance imaging sequence to generate a high-resolution diffusion magnetic resonance imaging image or picture of gray matter microstructure;
[0009] Step 2): Based on the MRtrix software platform and the FSL platform, the gray matter microstructure image noise and artifacts are removed, and the gray matter microstructure image is corrected by eddy current correction and spline interpolation correction;
[0010] Step 3): preprocessing the gray matter microstructure image to obtain an attribute image representing the gray matter microstructure;
[0011] Step 4): According to step 3), based on the MRtrix software platform, the tckgen tool is used to track and visualize the U-shaped fibers at the gray matter-white matter cortex junction of the gray matter microstructure and the tiny fibers in the gray matter cortex;
[0012] Step 5): According to step 3), the preprocessed gray matter microstructure image is converted into a brain surface grid using the AFNI software platform;
[0013] Step 6): Quantitative analysis of gray matter microstructure.
[0014] Optionally, in step 1), a high-resolution DTI imaging device of the optimized magnetic resonance diffusion imaging sequence is used to perform brain scanning, with a scanning time in the range of 0-23 minutes to generate a high-resolution diffusion magnetic resonance imaging image of gray matter microstructure.
[0015] Furthermore, the optimized parameters of the optimized MRI diffusion imaging sequence in the high-resolution DTI imaging device are: MRI repetition time / MRI echo time is 23083ms / 66.7ms, flip angle is 0-90°, lens viewing angle is 0-230×230mm 2 The gray matter microstructure layer thickness is 0-0.9mm, the gray matter microstructure is not less than 150 layers, the image matrix is 120×120, and the voxel size is not higher than 0.9×0.9×0.9mm 3 , the magnetic resonance b value is 800s / mm 2 , 13 diffusion gradient encoding directions, a multi-band acceleration factor of 4, the number of excitations of 1, and a scan time of 20 minutes.
[0016] Optionally, in step 2), the code dwidenoise based on the MRtrix software platform and the FSL platform is used to remove image noise of gray matter microstructure and improve the signal-to-noise ratio of the image; the code mrdegibbs based on the MRtrix software platform and the FSL platform is used to remove artifacts in the image data; and the code eddy_correct based on the MRtrix software platform and the FSL platform is used to perform eddy current correction and spline interpolation correction on the image after noise removal.
[0017] Optionally, in step 3), the gray matter microstructure image is preprocessed using the following steps:
[0018] Step 31): Use mrgrid code to perform grid resampling on DWI data;
[0019] Step 32): extracting the first reference image without diffusion weighting of the resampled DWI data using fslroi code for subsequent analysis;
[0020] Step 33): using the bet tool of the FSL platform to extract brain tissue from the resampled b0 image;
[0021] Step 34): Use the dwi2tensor code to calculate the diffusion tensor of the resampled DWI data, and specify the file containing the b vector and bz value to ensure the correct use of gradient information, wherein only the voxels within the brain mask are calculated during the calculation of the diffusion tensor of the resampled DWI data;
[0022] Step 35): quantifying the image based on the diffusion tensor to calculate various diffusion indices to characterize the properties of the gray matter microstructure and generate a gray matter microstructure image;
[0023] Step 36): Use the flirt tool to align the gray matter microstructure image from the T1 structural image to the b0 image.
[0024] Furthermore, in step 31), the DWI data is grid-resampled and magnified by 2 times, and cubic interpolation is used.
[0025] Optionally, in step 4), a fiber trajectory is generated by formulating a tensor deterministic algorithm; wherein the tensor deterministic algorithm is a combination operation of the position of the seed area and the whole-brain tensor model, and a certain fiber is tracked by combining the position of the seed area and the whole-brain tensor model, thereby determining the fiber trajectory, and generating a fiber trajectory file with a suffix of .tck, and the fiber trajectory file is visualized by FSLview software; in the fiber tracking process, the fiber direction in each voxel on the fiber trajectory or fiber pathway is uniquely determined, and the fiber is tracked by the fiber direction in each voxel on the fiber trajectory.
[0026] Optionally, in step 5), a gray matter microstructure image region is generated by the original image and fiber trajectory of the gray matter microstructure image. The former (the original image of the gray matter microstructure image) is based on the original image of the gray matter microstructure image, and the smoothwm surface model and the pial surface model of the standard left / right hemisphere are respectively converted according to the transformation matrix file to generate a normalized file. The smoothwm surface model data is converted into voxel data based on the 3dsurf2vol code of the AFNI tool, and the left / right hemisphere information is merged using the 3dcalc model to generate a rim mask, thereby performing a brain surface grid conversion process on the gray matter microstructure image, marking the brain pia mater surface and gray matter / white matter surface and defining the area between the two surfaces as a filling area, establishing a gray matter-cerebrospinal fluid boundary and a gray matter-white matter boundary, and determining the gray matter microstructure image region;
[0027] The latter (fiber trajectory) uses a tensor deterministic algorithm to determine the seed point position and the whole-brain tensor model, then traces the determined fibers and the trajectory of each voxel in the determined fiber pathway. Finally, the fibers pass through the trajectory of each voxel to determine the gray matter microstructure image area that should be analyzed.
[0028] Optionally, in step 6), the data of anisotropy fraction, mean diffusivity, radial diffusivity and axial diffusivity are extracted through the selected gray matter microstructure image area according to the customized brain template, and quantitative analysis is performed in SPSS software, and visualized using MRICron software or FSLview software; the results of fiber tracking are quantitatively analyzed in the FSL platform and visualized using FSLview software.
[0029] Beneficial effects of the present invention:
[0030] The present invention generates fiber trajectories by formulating a tensor deterministic algorithm, and generates a gray matter microstructure image area based on the original image of the gray matter microstructure image and the fiber trajectory to ensure the stability and reliability of fiber tracking; the FSL platform is used to perform quantitative analysis of gray matter microstructure image visualization to achieve fiber tracking. BRIEF DESCRIPTION OF THE DRAWINGS
[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0032] Figure 1 An image of gray matter microstructure obtained by high-resolution diffusion magnetic resonance imaging of the present invention;
[0033] Figure 2 The image of the gray matter microstructure image with noise and artifacts removed and eddy current correction and spline interpolation correction performed by the present invention;
[0034] Figure 3 Preprocessing the gray matter microstructure image of the present invention;
[0035] Figure 4 It is the attribute image characterizing the gray matter microstructure of the present invention;
[0036] Figure 5 The tracking image of the U-shaped fibers in the gray matter-white matter cortex junction area and the micro-fibers in the gray matter cortex of the gray matter microstructure of the present invention;
[0037] Figure 6Establishing gray matter-cerebrospinal fluid boundary and gray matter-white matter boundary images for the present invention;
[0038] Figure 7 A quantitative analysis image of the gray matter microstructure of the present invention;
[0039] Figure 8 It is the work flow chart of the present invention. DETAILED DESCRIPTION
[0040] The embodiments of the present application are described in detail below with reference to the accompanying drawings. Example
[0041] Reference Figure 1-Figure 8 This embodiment provides an imaging method for gray matter microstructure imaging based on high-resolution diffusion magnetic resonance imaging, comprising the following steps (eg Figure 8 shown):
[0042] Step 1): Figure 1 As shown, a magnetic resonance diffusion imaging sequence (i.e., a magnetic resonance diffusion imaging Multiplexed Sensitivity-Encoding (MUSE) imaging sequence) is optimized to generate a high-resolution diffusion magnetic resonance imaging image or picture of gray matter microstructure;
[0043] Step 2): Figure 2 As shown, the gray matter microstructure image noise and artifacts are removed based on the MRtrix software platform and the FSL platform, and the gray matter microstructure image is corrected by eddy current correction and spline interpolation correction;
[0044] Step 3): Figure 3 As shown in , the gray matter microstructure image is preprocessed (grid resampling, brain tissue extraction), as shown in Figure 4 As shown, the attribute image representing the gray matter microstructure is obtained ( Figure 4 Fractional anisotropy (FA);
[0045] Step 4): According to step 3), Figure 5 As shown, based on the MRtrix software platform, the tckgen tool was used to track and visualize the U-shaped fibers at the gray matter-white matter cortex junction of the gray matter microstructure and the tiny fibers in the gray matter cortex;
[0046] Step 5): According to step 3), Figure 6 As shown, the pre-processed gray matter microstructure image is transformed into a brain surface grid using the AFNI software platform;
[0047] Step 6): Figure 7 As shown, quantitative analysis of gray matter microstructure was performed.
[0048] Specifically, in step 1), a high-resolution DTI imaging device with an optimized magnetic resonance diffusion imaging sequence is used to perform brain scanning (i.e., a magnetic resonance diffusion imaging sequence is set or optimized in the high-resolution DTI imaging device, and the optimized magnetic resonance diffusion imaging sequence is then scanned by the high-resolution DTI imaging device), and the scanning time is within the range of 0-23 minutes (on the premise of a complete scan of the brain, the scanning time can be limited to about 23 minutes to ensure that most subjects can complete the scan) to generate a high-resolution diffusion magnetic resonance imaging image of the gray matter microstructure.
[0049] Furthermore, the optimized parameters of the optimized MRI diffusion imaging sequence in the high-resolution DTI imaging device are: MRI repetition time / MRI echo time is 23083ms / 66.7ms, flip angle is 0-90°, lens viewing angle is 0-230×230mm 2 The gray matter microstructure layer thickness is 0-0.9mm, the gray matter microstructure is not less than 150 layers, the image matrix is 120×120, and the voxel size is not higher than 0.9×0.9×0.9mm 3 (The voxel size is 0.875×0.875×0.875mm 3 Sub-millimeter ultra-high resolution DTI scan), magnetic resonance b value is 800s / mm 2 , 13 diffusion gradient encoding directions, a multi-band acceleration factor of 4, the number of excitations of 1, and a scan time of 20 minutes.
[0050] In step 2), the code dwidenoise based on the MRtrix software platform and the FSL platform is used to remove the image noise of the gray matter microstructure and improve the signal-to-noise ratio of the image; the code mrdegibbs based on the MRtrix software platform and the FSL platform is used to remove artifacts in the image data; and the code eddy_correct based on the MRtrix software platform and the FSL platform is used to perform eddy current correction and spline interpolation correction on the image after noise removal.
[0051] In step 3), the gray matter microstructure image is preprocessed using the following steps:
[0052] Step 31): Use mrgrid code to perform grid resampling on DWI data;
[0053] Step 32): extracting the first reference image without diffusion weighting of the resampled DWI data using fslroi code for subsequent analysis;
[0054] Step 33): Use the bet tool of the FSL platform to extract brain tissue from the resampled b0 image;
[0055] Step 34): Use the dwi2tensor code to calculate the diffusion tensor of the resampled DWI data, and specify the file containing the b vector and bz value to ensure the correct use of gradient information, wherein only the voxels within the brain mask are calculated during the calculation of the diffusion tensor of the resampled DWI data;
[0056] Step 35): quantifying the image based on the diffusion tensor to calculate various diffusion indices to characterize the properties of the gray matter microstructure and generate a gray matter microstructure image;
[0057] Step 36): Use the flirt tool to align the gray matter microstructure image from the T1 structural image to the b0 image.
[0058] In step 31), the DWI data is grid-resampled and magnified by 2 times (scaling factor is 2), and cubic interpolation is used.
[0059] In step 4), a fiber trajectory is generated by formulating a tensor deterministic algorithm; wherein the tensor deterministic algorithm is a combined operation of the position of the seed region and the whole-brain tensor model, and by combining the position of the seed region and the whole-brain tensor model, a certain fiber is tracked, thereby determining the fiber trajectory, and generating a fiber trajectory file with the suffix .tck, and the fiber trajectory file is visualized by FSLview software; in the fiber tracking process, the fiber orientation in each voxel on the fiber trajectory or fiber path is uniquely determined, and the fiber is tracked by the fiber orientation in each voxel on the fiber trajectory to ensure the stability and reliability of fiber tracking. In addition, the opposite technology to fiber tracking is called fiber probabilistic tracking, that is, the fiber orientation is determined by calculating the probability of fiber connection between two voxels, and it is considered that the fiber orientation in each voxel on the fiber path is probabilistically distributed.
[0060] In step 5), the gray matter microstructure image region is generated by the original image and fiber trajectory of the gray matter microstructure image. In the former, based on the original image of the gray matter microstructure image, the smoothwm surface model and the pial surface model of the standard left / right hemisphere are converted according to the transformation matrix file to generate a normalized file, and the smoothwm surface model data is converted into voxel data based on the 3dsurf2vol code of the AFNI tool. The 3dcalc model is used to merge the left / right hemisphere information and generate a rim mask, thereby converting the gray matter microstructure image into a brain surface grid, marking the brain pia mater surface and gray matter / white matter surface and defining the area between the two surfaces as a filling area, establishing the gray matter-cerebrospinal fluid boundary and the gray matter-white matter boundary, and determining the gray matter microstructure image region;
[0061] The latter, through the tensor deterministic algorithm, determines the seed point position and the whole brain tensor model, and then traces the determined fibers and the trajectory of each voxel in the determined fiber pathway. Finally, the fiber passes through the trajectory of each voxel to determine the gray matter microstructure image area that should be analyzed.
[0062] In step 5), the gray matter microstructure image region is generated by the fiber trajectory mainly through the tensor deterministic algorithm, and the fiber tracking is performed after the trajectory of each voxel in the fiber trajectory is determined. The gray matter microstructure image region is determined by the original image of the gray matter microstructure image and the fiber trajectory.
[0063] Optionally, in step 6), a visual quantitative analysis of gray matter microstructure is performed. The data of anisotropy fraction (FA), mean diffusivity (MD), radial diffusivity (RD) and axial diffusivity (AD) are extracted through the selected gray matter microstructure image area according to the custom brain template, and quantitative analysis is performed in SPSS software, and visualized using MRICron software or FSLview software; the results of fiber tracking are quantitatively analyzed in the FSL platform and visualized using FSLview software. Example
[0064] Based on Example 1, the present invention improves the diffusion magnetic resonance imaging with a spatial resolution of 2 mm in routine clinical work to diffusion magnetic resonance with a spatial resolution of 0.9 mm.
[0065] With respect to the decrease in signal-to-noise ratio after the spatial resolution is improved, the present invention can improve the signal-to-noise ratio by removing image noise.
[0066] In addition, the present invention has also developed a complete post-processing quantitative analysis and visualization technology chain, which can perform quantitative analysis on gray matter microstructure, fiber tracing or tracking of U-shaped fibers at the gray-white matter junction and microfibers in the gray matter cortex, ultimately achieving the goal of improving clinical work efficiency.
[0067] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope of the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. A method for imaging gray matter microstructure based on high-resolution diffusion magnetic resonance imaging, characterized in that: The steps include: Step 1): Optimizing a diffusion-weighted magnetic resonance imaging sequence to generate a high-resolution diffusion-weighted magnetic resonance imaging image or picture of gray matter microstructure; Step 2): Remove noise and artifacts from gray matter microstructure images based on the MRtrix software platform and the FSL platform, and perform eddy current correction and spline interpolation correction on the gray matter microstructure images; Step 3): preprocessing the gray matter microstructure image to obtain an attribute image representing the gray matter microstructure; In step 3), the gray matter microstructure image is preprocessed using the following steps: Step 31): Use mrgrid code to perform grid resampling on the DWI data; Step 32): using the fslroi code to extract the first non-diffusion-weighted reference image of the resampled DWI data for subsequent analysis; Step 33): using the bet tool of the FSL platform to extract brain tissue from the resampled b0 image; Step 34): Calculate the diffusion tensor of the resampled DWI data using the dwi2tensor code and specify the file containing the b vector and bz value to ensure the correct use of gradient information, wherein only the voxels within the brain mask are calculated during the diffusion tensor calculation of the resampled DWI data; Step 35): Calculating various diffusion indices based on image quantification of the diffusion tensor to characterize the properties of gray matter microstructure and generate a gray matter microstructure image; Step 36): Use the flirt tool to register the gray matter microstructure image from the T1 structural image to the b0 image; Step 4): Based on step 3), the tckgen tool is used on the MRtrix software platform to track and visualize the U-shaped fibers at the gray matter-white matter cortical junction and the microfibers within the gray matter cortex; Step 5): According to step 3), the pre-processed gray matter microstructure image is converted into a brain surface grid using the AFNI software platform; In the step 5), a gray matter microstructure image region is generated by using the original image and fiber trajectory of the gray matter microstructure image. In the former, based on the original image of the gray matter microstructure image, the smoothwm surface model and the pial surface model of the standard left / right hemisphere are converted according to the transformation matrix file to generate a normalized file, the smoothwm surface model data is converted into voxel data based on the 3dsurf2vol code of the AFNI tool, the left / right hemisphere information is merged using the 3dcalc model, and a rim mask is generated, thereby performing a brain surface mesh conversion process on the gray matter microstructure image, marking the brain pia mater surface and gray matter / white matter surface and defining the area between the two surfaces as a filling area, establishing the gray matter-cerebrospinal fluid boundary and the gray matter-white matter boundary, and determining the gray matter microstructure image region; The latter, through a tensor deterministic algorithm, determines the seed point position and the whole-brain tensor model, and then traces the determined fibers and the trajectory of each voxel in the determined fiber pathway. Finally, the fiber passes through the trajectory of each voxel to determine the gray matter microstructure image area that should be analyzed; Step 6): Quantitative analysis of gray matter microstructure; In step 6), data on fractional anisotropy, mean diffusivity, radial diffusivity, and axial diffusivity are extracted from the selected gray matter microstructure image area according to a custom brain template, quantitatively analyzed in SPSS software, and visualized using MRICron software or FSLview software. The fiber tracking results are quantitatively analyzed in the FSL platform and visualized using FSLview software.
2. The imaging method of gray matter microstructure based on high-resolution diffusion magnetic resonance imaging according to claim 1, characterized in that: In step 1), a brain scan is performed using a high-resolution DTI imaging device of the optimized diffusion magnetic resonance imaging sequence, with a scanning time ranging from 0 to 23 minutes, to generate an image of gray matter microstructure of the high-resolution diffusion magnetic resonance imaging.
3. The imaging method of gray matter microstructure based on high-resolution diffusion magnetic resonance imaging according to claim 2, characterized in that: The optimized parameters of the optimized magnetic resonance diffusion imaging sequence in the high-resolution DTI imaging device are: magnetic resonance repetition time / magnetic resonance echo time is 23083ms / 66.7ms, flip angle is 0-90°, and lens viewing angle is 0-230×230mm. 2 The gray matter microstructure layer thickness is 0-0.9mm, the gray matter microstructure is not less than 150 layers, the image matrix is 120×120, and the voxel size is not higher than 0.9×0.9×0.9mm 3 , the magnetic resonance b value is 800s / mm 2 , 13 diffusion gradient encoding directions, a multi-band acceleration factor of 4, the number of excitations was 1, and the scanning time was 20 minutes.
4. The imaging method of gray matter microstructure based on high-resolution diffusion magnetic resonance imaging according to claim 1, characterized in that: In step 2), the image noise of gray matter microstructure is removed based on the code dwidenoise of the MRtrix software platform and the FSL platform to improve the signal-to-noise ratio of the image; the artifacts in the image data are removed based on the code mrdegibbs of the MRtrix software platform and the FSL platform; and the eddy_correct code based on the MRtrix software platform and the FSL platform performs eddy current correction and spline interpolation correction on the image after noise removal.
5. The imaging method of gray matter microstructure based on high-resolution diffusion magnetic resonance imaging according to claim 1, characterized in that: In step 31), the DWI data is grid-resampled and magnified by a factor of 2, and cubic interpolation is used.
6. The imaging method of gray matter microstructure based on high-resolution diffusion magnetic resonance imaging according to claim 1, characterized in that: In step 4), a fiber trajectory is generated by formulating a tensor deterministic algorithm; wherein, the tensor deterministic algorithm is a combined operation of the position of the seed area and the whole-brain tensor model. By combining the position of the seed area and the whole-brain tensor model, a certain fiber is tracked, thereby determining the fiber trajectory, and generating a fiber trajectory file with the suffix .tck. The fiber trajectory file is visualized by FSLview software; during the fiber tracking process, the fiber direction in each voxel on the fiber trajectory or fiber pathway is uniquely determined, and the fiber is tracked by the fiber direction in each voxel on the fiber trajectory.
