Gradient-based tactile topology mapping model construction method

Abstract

The invention relates to a gradient-based tactile topology mapping model construction method. The method comprises the steps of obtaining functional imaging data and structural imaging data of magnetic resonance scanning of a testee; constructing a voxel-level functional connection matrix based on a voxel space in the functional imaging data; constructing a vertex-level structure connection matrix based on a cortex space in the structure imaging data; based on a gradient dimension reduction analysis processing method, matrix similarity calculation and dimension reduction analysis are carried out on the function connection matrix and the structure connection matrix, and a plurality of gradient components and variance interpretation quantities are obtained; and on the basis of the variance interpretation quantity, determining a functional main gradient and a second structural gradient, and constructing a tactile topology mapping model in combination with a tactile topology distribution rule. According to the method, the problem that the current tactile perception topological mapping is not accurate is solved, and more accurate mapping of tactile perception and brain activation is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of neural imaging data processing, and in particular to a method for constructing a gradient-based tactile topology mapping model. Background Art

[0002] In neuroscience, understanding how the brain processes and organizes tactile information is a long-standing and complex task. The somatosensory motor (SI) area, a core brain region responsible for tactile information processing, has long been a focus of neuroscience research, with its structural and functional distribution characteristics. Combining functional and structural studies of the brain provides a diverse perspective for in-depth exploration of the brain mechanisms underlying tactile perception. Advanced techniques such as functional magnetic resonance imaging (fMRI) have enabled a more comprehensive exploration of tactile-related brain regions and their dynamic processes. A key finding in tactile research is that the functional map of the SI area changes in response to adjustments in task requirements or instructions to participants. This finding suggests that the SI area not only plays a passive role in receiving tactile information but also possesses dynamic characteristics that are closely linked to behavioral tasks. Using anatomical criteria, researchers aim to assign functionally significant voxel clusters to specific cytoarchitectural regions, thereby delineating the distribution patterns of the functional topology of the tactile brain. This division is crucial for understanding the brain's processing of tactile information, but current topological maps of tactile perception remain imprecise.

[0003] Constructing a tactile topographic map model to accurately characterize tactile topological relationships in the human hand tactile cortex is challenging. High-resolution structural magnetic resonance imaging (MRI) has enabled researchers to precisely identify changes in myelination patterns within the cortical gray matter. These changes not only vary across regions but also define the anatomical boundaries of four subregions within the tactile brain area (BA1, BA2, BA3a, and BA3b). These four subregions each play distinct functional roles in tactile processing. Further investigation of these regions will advance our understanding of the neural mechanisms underlying tactile sensation. However, it is noteworthy that the correspondence between cytoarchitectonic boundaries and macroscopic anatomical features identified from structural MRI images exhibits high inter-subject variability. This variability suggests that, while common features can be identified probabilistically in group data analyses, anatomical features can differ significantly at the individual level. This high degree of individuality emphasizes the importance of individualized research and calls for caution in interpreting group data. Therefore, individualized analyses are necessary to construct a universal tactile topographic map model.

[0004] By studying functional and structural gradients, a tactile topological mapping model was constructed. This model aims to address the current inaccurate topological mapping of tactile perception and achieve a more precise mapping of tactile perception and brain activation. Functional gradients reveal the gradual changes in different functional areas during tactile processing, while structural gradients reveal the anatomical basis behind these functional changes. The combination of these two approaches can more comprehensively explain the tactile brain functional topology and reveal the processing pathways of tactile information in the brain. Summary of the Invention

[0005] The purpose of this invention is to provide a gradient-based tactile topology mapping model construction method to realize the construction and analysis of the brain tactile topology mapping model, aiming to solve the current problem of inaccurate tactile perception topology mapping and achieve a more accurate mapping of tactile perception and brain activation.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] A method for constructing a gradient-based tactile topology mapping model, comprising:

[0008] Acquire functional imaging data and structural imaging data from magnetic resonance imaging of subjects;

[0009] Based on the voxel space in functional imaging data, a voxel-level functional connectivity matrix is constructed;

[0010] Based on the cortical space in the structural imaging data, a vertex-level structural connectivity matrix is constructed;

[0011] Based on the gradient dimensionality reduction analysis processing method, matrix similarity calculation and dimensionality reduction analysis are performed on the functional connectivity matrix and the structural connectivity matrix to obtain several gradient components and variance explanations;

[0012] Based on the variance explanation, the functional main gradient and the structural secondary gradient are determined, and a tactile topological mapping model is constructed in combination with the tactile topological distribution law.

[0013] Optionally, obtaining functional imaging data and structural imaging data from a magnetic resonance imaging scan of a subject includes:

[0014] Perform magnetic resonance imaging on the subjects to obtain original functional imaging data and original structural imaging data;

[0015] The original functional imaging data and the original structural imaging data are preprocessed respectively to obtain the functional imaging data and the structural imaging data.

[0016] Optionally, preprocessing the raw functional imaging data includes:

[0017] Correcting the time deviation in the original functional imaging data caused by the difference in acquisition time of different slices during the scanning process to ensure the consistency of each voxel in the time series;

[0018] Aligning the images at each moment to a preset reference image, and compensating for image deviation caused by motion using a number of motion parameters, wherein the motion parameters include translation and rotation;

[0019] Register individual brain images to the standard brain template space to ensure the consistency of the spatial position of each individual brain image;

[0020] Smoothing and filtering the image.

[0021] Optionally, preprocessing the original structural imaging data includes:

[0022] Performing non-uniformity correction on the intensity of the original structural imaging data to correct the intensity non-uniformity phenomenon in the image caused by the non-uniform magnetic field;

[0023] The brain skull is stripped by identifying and separating the boundaries between brain tissue and non-brain tissue, removing the skull and other non-brain tissues to extract the brain tissue area;

[0024] Extracting the cerebral cortex from the brain tissue image after removing the skull;

[0025] The extracted cerebral cortex was segmented into white matter and gray matter regions.

[0026] Optionally, constructing a voxel-level functional connectivity matrix includes:

[0027] Based on the voxel space in functional imaging data, brain regions were divided and the somatosensory motor regions of the right brain were located;

[0028] Extracting voxel time series from the right somatosensory motor area, wherein each voxel time series includes every time point during the participant's task performance;

[0029] The extracted voxel time series were compared with the time series of 271,633 voxels in the whole brain to perform Pearson correlation analysis, and a voxel-level functional connectivity matrix was generated for each subject.

[0030] Optionally, constructing a vertex-level structural connectivity matrix includes:

[0031] Based on the cortical space in the structural imaging data, several cortical vertices in the right somatosensory motor region are identified, and morphological features of the cortical vertices in the right somatosensory motor region of each subject are extracted; wherein the morphological features include: gray matter volume, surface area, cortical thickness, Gaussian curvature, and sulcal depth;

[0032] Based on the morphological features, feature vectors are constructed, and the structural connectivity matrix is obtained through Pearson correlation analysis between the feature vectors.

[0033] Optionally, performing matrix similarity calculation and dimensionality reduction analysis on the functional connectivity matrix and the structural connectivity matrix includes:

[0034] For the functional connectivity matrix and the structural connectivity matrix, respectively calculating similarity matrices of the two connectivity matrices;

[0035] For the similarity matrix, gradient dimensionality reduction is performed using a preset expression to identify spatial axes representing changes in different connection patterns; wherein each of the spatial axes corresponds to a specific component and the spatial axes are regarded as gradients;

[0036] Based on the similarity matrix, calculate the covariance matrix;

[0037] Perform eigenvalue decomposition on the covariance matrix to obtain a set of eigenvalues and corresponding eigenvectors; where the eigenvalue represents the variance of each principal component, and the eigenvector represents the direction of the principal component;

[0038] The principal components are sorted according to the size of the eigenvalues, and the principal components with the first several eigenvalues are selected; wherein the variance explanation of each principal component is the ratio of the eigenvalue of the principal component to the sum of all eigenvalues.

[0039] Optionally, the preset expression is:

[0040] W α =d -1α Ad -1α

[0041]

[0042] Among them, A represents the similarity matrix, d represents the degree matrix of A, P α represents the p-dimensional feature vector, D α W α The degree matrix of , G represents the new low-dimensional representation obtained after dimensionality reduction, λ and g represent the eigenvalue and eigenvector respectively, t represents the time parameter of scaling, α represents the anisotropic diffusion parameter used by the diffusion operator, W α The normalized degree matrix is obtained by normalizing the cosine similarity matrix A by α, and m represents the target dimension after dimensionality reduction.

[0043] Optionally, determining the functional primary gradient and the structural secondary gradient based on the variance explanation, and constructing a tactile topology mapping model in combination with the tactile topology distribution law includes:

[0044] Draw the variance explanation curves of the first several gradient components;

[0045] Based on the variance explanation rate ranking of the first several components of the functional gradient in the variance explanation curve diagram, the first gradient component is selected as the main functional gradient;

[0046] According to the results of the variance explanation rate ranking of the first several components of the structural gradient, the first gradient component and the second gradient component are determined as the main structural gradient and the second structural gradient respectively;

[0047] Based on the functional main gradient and the structural second gradient, a tactile topology mapping model is constructed in combination with the tactile topology distribution law.

[0048] The beneficial effects of the present invention are:

[0049] The present invention first obtains functional imaging data and structural imaging data from a subject's magnetic resonance imaging scan. Next, based on the voxel space in the functional imaging data, a voxel-level functional connectivity matrix is constructed. Next, based on the cortical space in the structural imaging data, a vertex-level structural connectivity matrix is constructed. Using a gradient dimensionality reduction analysis method, matrix similarity calculation and dimensionality reduction analysis are performed on the functional and structural connectivity matrices to obtain several gradient components and explained variance. Finally, based on the explained variance, the primary functional gradient and secondary structural gradient are determined, and a tactile topological mapping model is constructed based on the tactile topological distribution patterns. Compared with existing technologies, the present invention can solve the problem of inaccurate tactile perception topological mapping and achieve a more accurate mapping of tactile perception and brain activation. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. 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 any creative work.

[0051] Figure 1 A schematic flow chart of a method for constructing a gradient-based tactile topology mapping model according to an embodiment of the present invention;

[0052] Figure 2 A schematic diagram of performing matrix similarity calculation and dimensionality reduction to obtain gradient principal components based on a constructed functional connectivity matrix and a constructed structural connectivity matrix according to an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of determining the functional primary gradient and the structural secondary gradient based on the variance explanation and constructing a tactile topology mapping model in combination with the tactile topology distribution law according to an embodiment of the present invention;

[0054] Figure 4Schematic diagram of the structure of a gradient-based tactile topology mapping model construction system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0056] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0057] like Figure 1 As shown, this embodiment proposes a gradient-based tactile topology mapping model construction method, including:

[0058] S1. Obtain functional imaging data and structural imaging data from magnetic resonance imaging of the subject;

[0059] Furthermore, in S1, functional imaging and structural image scans of the subject are obtained, and the original brain images are preprocessed using a high-precision image registration processing method.

[0060] The MRI scan includes task-state functional MRI data and high-resolution structural images of T1-weighted imaging, which are used for alignment with task-state functional MRI and MNI standard brain space to solve the problem of poor direct alignment caused by the low spatial resolution of functional imaging.

[0061] Specifically, in S1, this embodiment designs a set of image processing methods to achieve high-precision image registration, specifically including: a method for processing structural images: 1. Performing non-uniformity correction on the intensity of the structural image to correct the intensity non-uniformity caused by the non-uniform magnetic field in the image. This non-uniformity may cause uneven grayscale in the image, thereby affecting the accuracy of subsequent processing. 2. By identifying and separating the boundaries between brain tissue and non-brain tissue, the brain skull is stripped and the skull and other non-brain tissues are removed to accurately extract the brain tissue area. Skull stripping helps reduce the interference of noise and artifacts, ensuring that subsequent analysis focuses on the brain tissue area. 3. Extracting the cerebral cortex from the brain tissue image after removing the skull for more detailed structural analysis. 4. Further dividing the extracted cortical area into white matter and gray matter areas for more detailed analysis. Functional image processing methods: 1. Correct the time deviation caused by the difference in the acquisition time of different slices during the scanning process to ensure the consistency of each voxel in the time series; 2. Align the images at each moment to a reference image, and use a variety of motion parameters (translation and rotation) to compensate for the image deviation caused by motion to reduce the impact of motion artifacts. The head motion parameter is set to <3mm. 3. Align individual brain images to the standard brain template space to ensure the consistency of the spatial position of each individual brain image, which is convenient for comparison and analysis between groups; 4. Smooth the image to improve the signal-to-noise ratio, reduce noise, and enhance the spatial coherence of the signal in the image. 5. Remove low-frequency drift in the time series and high-frequency noise generated by the 7T nuclear magnetic equipment through filtering, retain the signal frequency band of interest, and exclude interference at the same time. The bandpass frequency range (0.01-0.1Hz).

[0062] S2. Construct a voxel-level functional connectivity matrix based on the voxel space in the functional imaging data;

[0063] Specifically, in S2, this embodiment divides the brain regions based on voxel space and locates the somatosensory motor brain region of the right brain. This region consists of 33 sub-regions, and its functional and structural network is constructed. The voxel time series of the somatosensory motor brain region of the right brain is extracted, totaling 2083 voxels. Each voxel time series contains every time point (836 time points in total) when the participant is performing the task, thereby reflecting the dynamic brain functional activity of the somatosensory motor brain region during the entire scanning period. By performing Pearson correlation analysis on the time series data of the extracted right brain somatosensory motor brain region and the time series of 271,633 voxels of the whole brain, a voxel-level correlation matrix (271,633×2083) is generated for each subject to characterize the functional connectivity of the somatosensory motor brain region with the whole brain.

[0064] S3, constructing a vertex-level structural connectivity matrix based on the cortical space in the structural imaging data;

[0065] Specifically, in S3, for vertex-level structural network construction, the 24,044 cortical vertices contained in the right somatosensory motor region were first identified. Five morphological features of the cortical vertices in each subject's right somatosensory motor region were extracted, including gray matter (GM) volume, surface area (SA), cortical thickness (CT), Gaussian curvature (GC, reflecting the geometric shape of the cortex), and cerebral sulcus (CS, reflecting the folding complexity of the cortex). Based on these morphological features, feature vectors were constructed, and Pearson correlation analysis was performed between the feature vectors to obtain a 24,044×24,044 structural connectivity matrix.

[0066] S4. Based on the gradient dimensionality reduction analysis method, matrix similarity calculation and dimensionality reduction analysis are performed on the functional connectivity matrix and the structural connectivity matrix to obtain several gradient components and variance explanations;

[0067] Specifically, in step S4, this embodiment designs a set of gradient dimensionality reduction analysis methods. The specific analysis process is as follows: Figure 2 As shown. For each individual, based on the constructed functional and structural connectivity matrices, the similarity matrices of the two connectivity matrices are calculated respectively. Among them, the calculation of the similarity matrix: taking the functional connectivity matrix (not a square matrix and asymmetric) as an example, the cosine similarity between each column in the connectivity matrix and each other column is calculated. The elements in the similarity matrix can represent the similarity or distance between two data, and identify the patterns and associations between the data. After the similarity calculation, the original connectivity matrix can be converted into a symmetrical square matrix, reducing the dimension of the data and reducing the complexity of the calculation.

[0068] Specifically, the functional connectivity matrix is converted into a square matrix with a size of 2083×2083, while the structural connectivity matrix is originally a square matrix, so its size remains unchanged. The similarity matrix is used to measure the similarity of the connection patterns between different regions, where the higher the similarity value, the more similar the connection patterns between regions. In order to characterize this spatial topological structure, gradient dimensionality reduction is finally used to identify a series of spatial axes representing the changes in different connection patterns. Each axis corresponds to a specific component, and these axes are collectively referred to as "gradients." The analysis of gradient dimensionality reduction is shown in formulas (1), (2), and (3).

[0069] W α =d -1α Ad -1α (1)

[0070]

[0071] The diffusion operator in formula (1) uses the anisotropic diffusion parameter α∈[0,1], A represents the similarity matrix, and W α It is obtained by normalizing the similarity matrix A through α, and d represents the degree matrix of A. In formula (2), P α Represents the p-dimensional feature vector, and it is necessary to map the high-dimensional connection matrix into a low-dimensional space to retain the important connectivity features in the connectivity pattern. α It's W α degree matrix. In formula (3), G represents the new low-dimensional representation obtained after applying dimensionality reduction, that is, the gradient; λ and g represent the eigenvalue and eigenvector, respectively; and t represents the time parameter of the scale. α determines the effect of the sampling point density on dimensionality reduction: when α = 0, the sampling point density has the greatest effect on dimensionality reduction; when α = 1, there is no effect. In this embodiment, α = 0.5 and t = 0 are selected to preserve the global relationship between data points in the embedding space.

[0072] The variance explanation is obtained by calculating the similarity matrix; the variance explanation is calculated to select the principal components for dimensionality reduction;

[0073] Obtaining the amount of variance explained includes:

[0074] 1. Calculate the covariance matrix: The covariance matrix reflects the linear relationship and interdependence between variables. Each element in the covariance matrix represents the covariance between two variables.

[0075] Calculation of covariance:

[0076] For the similarity matrix X, calculate the covariance matrix C;

[0077]

[0078] Where n is the number of columns or rows of the matrix (because it is a square matrix);

[0079] 2. Eigenvalue decomposition: Perform eigenvalue decomposition on the covariance matrix to obtain a set of eigenvalues and corresponding eigenvectors. The eigenvalue represents the variance of each principal component, and the eigenvector represents the direction of the principal component.

[0080] 3. Select principal components: Sort the principal components by eigenvalue and select the first few with the largest eigenvalues. These principal components can explain most of the variance in the data, thus achieving the purpose of dimensionality reduction. Here, the number of principal components we want to select determines the number of eigenvalues, which in turn constitute the variance explained.

[0081] The variance explained by each principal component is the ratio of the eigenvalue of the principal component to the sum of all eigenvalues:

[0082]

[0083] Among them, λ i is the eigenvalue of the i-th principal component, and p is the total number of features;

[0084] The variance explained by each gradient component reflects its importance in describing the functional and structural connectivity characteristics of somatosensory and motor brain regions. For example, a component with a higher variance explained indicates a greater contribution to the data variation and is therefore more representative in understanding and explaining the functional and structural changes in that brain region.

[0085] S5. Based on the variance explanation, determine the functional primary gradient and structural secondary gradient, and construct a tactile topological mapping model based on the tactile topological distribution law.

[0086] Specifically, in S5, this embodiment selects to map the first two components (called the first gradient and the second gradient) back to the anatomical space of the somatosensory-motor brain area according to the size of the variance explained, so as to facilitate the visualization of the spatial representation of these gradients. Figure 3 As shown in the figure, the variance explanation curves of the first 10 gradient components are plotted. The variance explanation quantifies the contribution of each component in explaining the total variation of functional and structural data. Gradient components with high variance explanation mean that they are more significant and representative in describing the functional and structural connection characteristics of the somatosensory motor brain areas. Figure 3 The first 10 components of the functional gradient were ranked by variance explained, and the first gradient component was selected as the principal gradient. This gradient component explained approximately 20% of the total connectivity variance. Furthermore, based on the variance explained by the first 10 components of the structural gradient, the first and second gradient components were identified as the principal and secondary structural gradients, respectively. The principal and secondary structural gradients explained approximately 76% and 8% of the total connectivity variance, respectively. The principal and secondary structural gradients were consistent with the tactile topology map, and were subsequently selected to construct the tactile topology map model.

[0087] In this embodiment, based on a gradient-based tactile topological mapping model construction method of the present invention, the functional and structural gradient characteristics of the tactile brain areas of 10 healthy young people were analyzed using autonomously collected 7T task-state functional and structural magnetic resonance data, and then a tactile topological mapping model was constructed.

[0088] like Figure 2 and Figure 3 As shown in Figure 2, the construction process of the tactile topology mapping model includes the following:

[0089] 1. During task-based functional data acquisition, gradient echo planar imaging (GEPI) was used with the following parameters: TR = 2000 ms, TE = 25 ms, FA = 67°, and FoV = 200 mm × 200 mm. The functional imaging baseline was parallel to the anterior commissure (AC) and posterior commissure (PC) lines (AC-PC). A multislice EPI sequence was used with a multiband acceleration factor of 3 and an isotropic resolution of 1.5 mm³, resulting in a total of 96 slices. Whole-brain high-resolution anatomical images were acquired using the MP2RAGE sequence with the following parameters: TR = 4300 ms, TE = 2.35 ms, FA = 4°, FoV = 230 mm × 210 mm, first inversion time (TI1) = 1000 ms, and isotropic voxel size = 0.64 mm.

[0090] 2. High-precision image registration processing methods are used to correct intensity non-uniformity, perform skull stripping, cortex extraction, and gray-white matter segmentation on the original images. Functional imaging is time-corrected, motion-corrected, normalized, smoothed, and filtered, and then linearly registered to the high-resolution T1 structural image.

[0091] 3. Based on the voxel space, the brain regions were divided and the somatosensory motor brain region of the right brain was located. For the construction of the functional network of each individual in the task state, the voxel time series of the somatosensory motor brain region of the right brain was first extracted, totaling 2083 voxels. Each voxel time series contains every time point when the participant is performing the task (a total of 836 time points), thus reflecting the dynamic brain functional activity of the somatosensory motor brain region during the entire scanning period. By performing Pearson correlation analysis on the time series data of the extracted right brain somatosensory motor brain region and the time series of 271,633 voxels in the whole brain, a voxel-level correlation matrix (271,633×2083) was generated for each individual to characterize the functional connectivity between the somatosensory motor brain region and the whole brain.

[0092] 4. Based on cortical space, we determined that the right somatosensory motor region contains 24,044 cortical vertices. We extracted five morphological features of these cortical vertices for each individual, including gray matter (GM) volume, surface area (SA), cortical thickness (CT), Gaussian curvature (GC, reflecting cortical geometry), and cerebral sulcus (CS, reflecting cortical folding complexity), to construct a structural network. Based on these morphological features, we constructed feature vectors, and through correlation analysis between these feature vectors, we obtained a 24,044×24,044 structural connectivity matrix.

[0093] 5. For each individual, the similarity matrix between the functional and structural connectivity matrices was calculated based on the constructed functional and structural connectivity matrices. Specifically, the functional connectivity matrix was converted into a square matrix of size 2083×2083, while the structural connectivity matrix was originally a square matrix, so its size remained unchanged. A dimensionality reduction analysis was then performed to obtain 10 principal components of the gradient. The functional primary gradient and structural secondary gradient were then determined based on the variance explained. Finally, these two gradient components were used to construct a tactile topological mapping model, addressing the issue of inaccurate topological mapping of tactile perception before touch.

[0094] like Figure 4 As shown, this embodiment also provides a gradient-based tactile topology mapping model construction system. The computing system includes a magnetic resonance scanning and data acquisition module for scanning and reconstructing functional imaging and structural image data; a data processing module for data preprocessing and functional and structural gradient analysis and calculation, for data preprocessing, calculation and analysis of functional connectivity, structural connectivity, functional gradients, and functional and structural gradients; and a visualization module for visualizing functional and structural gradients on the brain and visualizing the tactile topology mapping model.

[0095] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A method for constructing a gradient-based tactile topology mapping model, characterized in that: include: Obtaining functional imaging data and structural imaging data from magnetic resonance imaging of subjects; Based on the voxel space in functional imaging data, a voxel-level functional connectivity matrix is constructed; Based on the cortical space in the structural imaging data, a vertex-level structural connectivity matrix is constructed; Based on the gradient dimensionality reduction analysis processing method, matrix similarity calculation and dimensionality reduction analysis are performed on the functional connectivity matrix and the structural connectivity matrix to obtain several gradient components and variance explanations; Based on the variance explanation, the functional main gradient and the structural secondary gradient are determined, and a tactile topological mapping model is constructed in combination with the tactile topological distribution law.

2. The method for constructing a gradient-based tactile topology mapping model according to claim 1, wherein: Acquiring functional imaging data and structural imaging data from magnetic resonance imaging of subjects includes: Perform magnetic resonance imaging on the subjects to obtain original functional imaging data and original structural imaging data; The original functional imaging data and the original structural imaging data are preprocessed respectively to obtain the functional imaging data and the structural imaging data.

3. The method for constructing a gradient-based tactile topology mapping model according to claim 2, wherein: Preprocessing the raw functional imaging data includes: Correcting the time deviation in the original functional imaging data caused by the difference in acquisition time of different slices during the scanning process to ensure the consistency of each voxel in the time series; Aligning the images at each moment to a preset reference image, and compensating for image deviation caused by motion using a number of motion parameters, wherein the motion parameters include translation and rotation; Register individual brain images to the standard brain template space to ensure the consistency of the spatial position of each individual brain image; Smoothing and filtering the image.

4. The method for constructing a gradient-based tactile topology mapping model according to claim 2, wherein: Preprocessing the original structural imaging data includes: Performing non-uniformity correction on the intensity of the original structural imaging data to correct the intensity non-uniformity phenomenon in the image caused by the non-uniform magnetic field; The brain skull is stripped by identifying and separating the boundaries between brain tissue and non-brain tissue, removing the skull and other non-brain tissues to extract the brain tissue area; Extracting the cerebral cortex from the brain tissue image after removing the skull; The extracted cerebral cortex was segmented into white matter and gray matter regions.

5. The method for constructing a gradient-based tactile topology mapping model according to claim 1, wherein: Constructing a voxel-level functional connectivity matrix includes: Based on the voxel space in functional imaging data, brain regions were divided and the somatosensory motor regions of the right brain were located; Extracting voxel time series from the right somatosensory motor area, wherein each voxel time series includes every time point during the participant's task performance; The extracted voxel time series were compared with the voxel time series of the whole brain to perform Pearson correlation analysis to generate a voxel-level functional connectivity matrix for each subject.

6. The method for constructing a gradient-based tactile topology mapping model according to claim 1, wherein: Constructing the vertex-level structural connectivity matrix includes: Based on the cortical space in the structural imaging data, several cortical vertices in the right somatosensory motor region are identified, and morphological features of the cortical vertices in the right somatosensory motor region of each subject are extracted; wherein the morphological features include: gray matter volume, surface area, cortical thickness, Gaussian curvature, and sulcal depth; Based on the morphological features, feature vectors are constructed, and the structural connectivity matrix is obtained through Pearson correlation analysis between the feature vectors.

7. The method for constructing a gradient-based tactile topology mapping model according to claim 1, wherein: Performing matrix similarity calculation and dimensionality reduction analysis on the functional connectivity matrix and the structural connectivity matrix includes: For the functional connectivity matrix and the structural connectivity matrix, respectively calculating similarity matrices of the two connectivity matrices; For the similarity matrix, gradient dimensionality reduction is performed using a preset expression to identify spatial axes representing changes in different connection patterns; wherein each of the spatial axes corresponds to a specific component and the spatial axes are regarded as gradients; Based on the similarity matrix, calculate the covariance matrix; Perform eigenvalue decomposition on the covariance matrix to obtain a set of eigenvalues and corresponding eigenvectors; where the eigenvalue represents the variance of each principal component, and the eigenvector represents the direction of the principal component; The principal components are sorted according to the size of the eigenvalues, and the principal components with the first several eigenvalues are selected; wherein the variance explanation of each principal component is the ratio of the eigenvalue of the principal component to the sum of all eigenvalues.

8. The method for constructing a gradient-based tactile topology mapping model according to claim 7, wherein: The preset expression is: W α =d -1 / α Ad -1 / α Among them, A represents the similarity matrix, d represents the degree matrix of A, P α represents the p-dimensional feature vector, D α W α The degree matrix of , G represents the new low-dimensional representation obtained after dimensionality reduction, λ and g represent the eigenvalue and eigenvector respectively, t represents the time parameter of scaling, α represents the anisotropic diffusion parameter used by the diffusion operator, W α The normalized degree matrix is obtained by normalizing the cosine similarity matrix A by α, and m represents the target dimension after dimensionality reduction.

9. The method for constructing a gradient-based tactile topology mapping model according to claim 1, wherein: Based on the variance explanation, determining the functional primary gradient and the structural secondary gradient, and constructing a tactile topology mapping model in combination with the tactile topology distribution law includes: Draw the variance explanation curves of the first several gradient components; Based on the variance explanation rate ranking of the first several components of the functional gradient in the variance explanation curve diagram, the first gradient component is selected as the main functional gradient; According to the results of the variance explanation rate ranking of the first several components of the structural gradient, the first gradient component and the second gradient component are determined as the main structural gradient and the second structural gradient respectively; Based on the functional main gradient and the structural second gradient, a tactile topology mapping model is constructed in combination with the tactile topology distribution law.

Images