Quantitative method and device for brain structure-function coupling based on brain network gradients

Abstract

The present invention discloses a quantitative method and device for brain structure-function coupling based on brain network gradients. The method comprises: using principal component analysis to reduce the dimensionality of a similarity matrix to obtain a structural gradient matrix, with each column representing a principal component, i.e., a structural connectivity gradient; using principal component analysis to reduce the dimensionality of a similarity matrix to obtain a functional gradient matrix, with each column representing a principal component, i.e., a functional connectivity gradient; using principal component analysis to reduce the dimensionality of the structural connectivity matrix and the functional connectivity matrix to obtain a structural gradient matrix and a functional gradient matrix; calculating the Pearson correlation coefficient between each row of the low-frequency structural gradient matrix and the low-frequency functional gradient matrix to obtain a low-frequency gradient coupling index; and calculating the Pearson correlation coefficient between each row of the high-frequency structural gradient matrix and the high-frequency functional gradient matrix to obtain a high-frequency gradient coupling index; and capturing subtle changes in brain structure or function using the low-frequency gradient coupling index and the high-frequency gradient coupling index. The device comprises a processor and a memory.

Description

Technical Field

[0001] The present invention relates to the field of medical image signal processing, and in particular to a quantitative method and device for brain structure-function coupling based on brain network gradients. Background Art

[0002] According to classical neuroanatomical theory, the spatial arrangement of brain regions is not random but rather the result of developmental mechanisms shaped by evolutionary selection. Recent neuroimaging and network neuroscience research has made significant progress in mapping spatial gradients in both human and non-human brains. A gradient is an axis of variation in cortical features that is spatially continuous. The similarity between two regions is strongly correlated with their relative position along the cortical surface, meaning that regions with similar characteristics tend to be located close together along the gradient. In recent years, new methods for studying spatial gradients in brain tissue have emerged. Several studies have applied dimensionality reduction algorithms to functional and structural connectivity data to derive principal components, or gradients, that describe smooth transitions between different cortical systems. These principal components have been applied to the study of various neurodegenerative diseases. For example, abnormal gradients have been found in conditions such as major depressive disorder, autism, schizophrenia, acute stroke, childhood bipolar disorder, Alzheimer's disease, and epilepsy. Therefore, gradients are considered biomarkers for a variety of neurodegenerative diseases.

[0003] Structure-function coupling has been a hot topic in recent research. Numerous studies have applied various approaches, including communication models, biophysical models, and statistical models, to this approach. It has been widely applied to the study of brain development, cognitive function, gene expression, and various diseases, including epilepsy, depression, schizophrenia, bipolar disorder, and traumatic brain injury. Compared to single structural or functional networks, structure-function coupling methods can capture more subtle changes in the brain and more comprehensively reveal the complex mechanisms of brain activity.

[0004] Existing structural-functional coupling indices do not use principal component analysis to reduce data dimensions, require large storage space and high computational cost, and contain redundancy and noise. Summary of the Invention

[0005] This invention provides a quantitative method and device for brain structure-function coupling based on brain network gradients. This invention combines structure-function coupling with structural and functional connectivity gradients to more comprehensively and sensitively reflect brain activity and capture subtle changes in brain structure and function. This method is expected to provide new research ideas for exploring the physiological mechanisms of healthy or disordered brain activity. Details are described below:

[0006] In the first aspect, a quantitative method for brain structure-function coupling based on brain network gradients, the method comprising:

[0007] Gradient echo sequence was used to acquire brain imaging data; axial single spin echo EPI sequence was used to acquire DTI data; and preprocessing operations were performed on DTI data and brain imaging data respectively.

[0008] The similarity matrix is ​​reduced in dimension using principal component analysis to obtain a structural gradient matrix, where each column is a principal component, i.e., the structural connection gradient; the similarity matrix is ​​reduced in dimension using principal component analysis to obtain a functional gradient matrix, where each column is a principal component, i.e., the functional connection gradient;

[0009] The structural connectivity matrix and the functional connectivity matrix are reduced in dimension using principal component analysis to obtain the structural gradient matrix and the functional gradient matrix;

[0010] The Pearson correlation coefficient between each row of the low-frequency structural gradient matrix and the low-frequency functional gradient matrix was calculated to obtain the low-frequency gradient coupling index. The Pearson correlation coefficient between each row of the high-frequency structural gradient matrix and the high-frequency functional gradient matrix was calculated to obtain the high-frequency gradient coupling index.

[0011] The low-frequency gradient coupling index and high-frequency gradient coupling index are used to capture subtle changes in brain structure or function.

[0012] Wherein, the low-frequency gradient coupling index is:

[0013]

[0014] Among them, lr x is the Pearson correlation coefficient of each row of the low-frequency structural gradient and low-frequency functional gradient matrices, x = 1, 2, 3, ...., N.

[0015] Wherein, the high frequency gradient coupling index is:

[0016]

[0017] Among them, hr x is the Pearson correlation coefficient of each row of the high-frequency structural gradient and high-frequency functional gradient matrices, x = 1, 2, 3, ...., N.

[0018] The method further includes calculating the ratio of the high-frequency gradient coupling index to the low-frequency gradient coupling index, which is called the gradient coupling fraction:

[0019]

[0020] Among them, coindex ratio is the gradient coupling score matrix, hr x / lr x is the ratio of the high-frequency gradient coupling index to the low-frequency gradient coupling index, x=1, 2, 3, ..., N.

[0021] A device for quantifying brain structure-function coupling based on brain network gradients, the device comprising: a processor and a memory, wherein program instructions are stored in the memory, and the processor calls the program instructions stored in the memory to enable the device to execute any method described in the first aspect of the claim.

[0022] A computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is caused to execute any one of the methods described in the first aspect.

[0023] The beneficial effects of the technical solution provided by the present invention are:

[0024] 1. The present invention obtains the structural connectivity matrix and functional connectivity matrix of the brain by processing diffusion tensor imaging (DTI) data and functional magnetic resonance imaging (fMRI) data respectively. The principal component analysis algorithm is used to reduce the dimensionality of the two matrices and extract all principal components and eigenvectors. Each column is a principal component, i.e., a "gradient." By performing Pearson correlation analysis on the structural connectivity gradient, functional connectivity gradient matrix, low-frequency gradient, and high-frequency gradient, an indicator reflecting characteristic information of brain structure-function coupling can be obtained.

[0025] 2. This paper proposes a new method for calculating the structure-function gradient coupling index. By calculating the correlation between the structural connectivity gradient and the functional connectivity gradient at the brain region level, multiple structure-function gradient coupling indices are obtained, thereby achieving the goal of more comprehensively and sensitively capturing subtle changes in the brain, providing a new biomarker for studying the mechanism of brain activity.

[0026] 3. The present invention provides a multimodal fusion-based gradient-based structure-function coupling index. After acquiring and processing fMRI and DTI signals, the gradient coupling index of the present invention is used to calculate and obtain a series of gradient coupling indices. This can capture the changes in the structure and function of the brain during development and evolution, providing new methods and ideas for understanding the physiological mechanisms of brain development.

[0027] 4. Applying the gradient coupling index proposed in this invention to the study of abnormalities in brain tissue structure and function in neurodegenerative diseases may significantly improve the sensitivity of identifying differences between healthy subjects and diseased individuals, and can more sensitively and comprehensively detect pathological changes of the disease, more accurately locate lesions, and capture more subtle connections between structural abnormalities and functional abnormalities. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 Flowchart for calculation of the structure-function gradient coupling index.

[0029] Figure 2 is the calculation result diagram of the gradient coupling index;

[0030] Figure 3 This is the calculation result of the low-frequency gradient coupling index;

[0031] Figure 4 This is the calculation result of high-frequency gradient coupling index;

[0032] Figure 5 This is the calculation result of the gradient coupling fraction. DETAILED DESCRIPTION

[0033] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention are described in further detail below.

[0034] 1. Data Collection Plan

[0035] Brain imaging data were acquired on a 3T Siemens MRI scanner with an 8-channel head coil using a gradient echo (EPI) sequence. Imaging parameters were as follows: for DTI data, an axial single-shot spin echo EPI sequence was used to acquire diffusion data (b = 0 s / mm 2 , 1000s / mm 2 , 2000s / mm 2 , 30 gradient directions), TR=10.5s, TE=103ms, FOV=239×240mm 2 , slice thickness = 1.8 mm, number of slices = 38, matrix size = 128 × 128, acquisition time = 3 min 35 s, voxel size = 1.8 × 1.8 × 1.8 mm 3 For fMRI data, repetition time (TR) = 3s, echo time (TE) = 30ms, field of view (FOV) = 175×238mm 2 , flip angle (FA) = 90°, slice thickness = 3 mm, number of slices = 38, matrix size = 50 × 64, acquisition time = 11 min 30 s, voxel size = 3.4 × 3.4 × 3.4 mm 3 .

[0036] 2. Data Preprocessing Solution

[0037] The preprocessing steps of DTI data include: removing head motion artifacts, eddy current correction, creating brain tissue masks, diffusion tensor reconstruction, diffusion tensor fitting using the least squares method, fiber tract tracking, image registration to standard space, and smoothing with a Gaussian kernel.

[0038] The preprocessing steps of fMRI data include: removing head motion artifacts, time series alignment, spatial filtering and smoothing, removing noise interference, correcting brain image distortion, temporal filtering, normalizing the image to the MNI standard space, removing linear and nonlinear trends in the data, and smoothing with a Gaussian kernel.

[0039] 3. Construction scheme of structural connection gradient

[0040] The steps for constructing the structural connectivity gradient are as follows:

[0041] 1) Divide the cerebral cortex into N brain regions; extract the white matter fiber bundles connecting every two brain regions to construct a structural connectivity matrix, obtaining an N×N structural connectivity matrix;

[0042] 2) Calculate the similarity matrix of the structural connection matrix; scale the similarity matrix into a normalized angle matrix;

[0043] 3) Use the principal component analysis algorithm to reduce the dimension of the similarity matrix and obtain the structural gradient matrix, where each column is a principal component, i.e., the structural connection gradient.

[0044] 4. Construction of functional connectivity gradient

[0045] The steps for constructing the functional connectivity gradient are as follows:

[0046] 1) Divide the cerebral cortex into N brain regions; extract the mean time series signal of all voxels in each brain region;

[0047] 2) Calculate the Pearson correlation coefficient between the time series signals of each two brain regions to obtain an N×N functional connectivity matrix; set a 90% threshold for each row of the functional connectivity matrix, that is, retain the top 10% of connections in each region to eliminate false links and weak connections;

[0048] 3) Use cosine similarity to construct a similarity matrix; scale the similarity matrix to a normalized angle matrix to avoid negative values;

[0049] 4) Use the principal component analysis algorithm to reduce the dimension of the similarity matrix to obtain the functional gradient matrix, where each column is a principal component, i.e., the functional connectivity gradient.

[0050] 5. Calculation scheme of the structural-functional gradient coupling index

[0051] First, the structural connectivity matrix and the functional connectivity matrix are reduced in dimension using principal component analysis to obtain the structural gradient matrix and the functional gradient matrix.

[0052]

[0053] Among them, the structure gradient matrix is ​​represented by SCG, and the elements in the structure gradient matrix are represented by s ij The functional gradient matrix is ​​represented by FCG, and the elements in the functional gradient matrix are represented by f ij Indicates. scg i represents the i-th row of the structure gradient matrix, fcg i Represents the i-th row of the functional gradient matrix, i=1, 2, 3, ...., N, j=1, 2, 3, ...., N-1.

[0054] Among them, the ath row of the structural gradient matrix and the ath row of the functional gradient matrix are selected, respectively using scg a and fcg a express:

[0055] scg a =[s1 … s m ], fcg a =[f1 … f m ]

[0056] Among them, scg a The elements in s k Indicates that fcg a The elements in f k Indicates that k = 1, 2, 3,…, N-1.

[0057] Calculate the matrix scg separately a and fcg a The mean of:

[0058]

[0059] Calculate the Pearson correlation coefficient:

[0060]

[0061] Among them, r a represents the Pearson correlation coefficient between the a-th row of the structural gradient matrix and the a-th row of the functional gradient matrix, 1≤a≤N.

[0062] Similarly, the Pearson correlation coefficient is calculated for each other row according to the above process to obtain the gradient coupling index, such as Figure 2 shown.

[0063]

[0064] Among them, coindex is the gradient coupling index, r i represents the Pearson correlation coefficient between the i-th row of the structural gradient matrix and the i-th row of the functional gradient matrix, i = 1, 2, 3, ..., N.

[0065] Next, take U structural gradients and functional gradients with larger eigenvalues, which are called low-frequency gradients.

[0066]

[0067] Among them, U = 30% × N, the low-frequency structural gradient matrix is ​​represented by SCG_L, the elements in the low-frequency structural gradient matrix are represented by Sxy, the low-frequency functional gradient matrix is ​​represented by FCG_L, and the elements in the low-frequency functional gradient matrix are represented by f xy Indicates that x=1, 2, 3,…, N, y=1, 2, 3,…, U.

[0068] The Pearson correlation coefficient between each row of the low-frequency structural gradient matrix and the low-frequency functional gradient matrix was calculated to obtain the low-frequency gradient coupling index, as Figure 3 shown.

[0069]

[0070] Among them, lr x is the Pearson correlation coefficient of each row of the low-frequency structural gradient and low-frequency functional gradient matrices, x = 1, 2, 3, …, N.

[0071] The U structural gradients and functional gradients with smaller eigenvalues ​​are called high-frequency gradients.

[0072]

[0073] Among them, U = 30% × N, the high-frequency structure gradient matrix is ​​represented by SCG_H, and the elements in the high-frequency structure gradient matrix are represented by s zw The high frequency functional gradient matrix is ​​represented by FCG_H, and the elements in the high frequency functional gradient matrix are represented by f zw Indicates that z = 1, 2, 3,…, N, w = 1, 2, 3,…, U.

[0074] The Pearson correlation coefficient between each row of the high-frequency structural gradient matrix and the high-frequency functional gradient matrix was calculated to obtain the high-frequency gradient coupling index, such as Figure 4 shown.

[0075]

[0076] Among them, hr x is the Pearson correlation coefficient of each row of the high-frequency structural gradient and high-frequency functional gradient matrix, x = 1, 2, 3, ..., N.

[0077] Finally, the ratio of the high-frequency gradient coupling index to the low-frequency gradient coupling index is calculated, which is called the gradient coupling score, as Figure 5 shown.

[0078]

[0079] Among them, coindex ratio is the gradient coupling score matrix, hr x / lr x is the ratio of the high-frequency gradient coupling index to the low-frequency gradient coupling index, x=1, 2, 3,…, N.

[0080] By employing the gradient coupling index proposed in this embodiment of the present invention, subtle changes in brain structure or function can be captured more comprehensively and sensitively. In clinical and scientific research, the gradient coupling index proposed in this invention can be used to identify developmental changes in the brains of healthy and differently ill children or adolescents, assisting in the evaluation of clinical interventions and treatment outcomes.

[0081] Furthermore, the present invention can be used to classify various brain diseases into subgroups and conduct statistical analysis to explore variations in brain structure and functional characteristics across disease subgroups, assisting in the development of personalized clinical treatment plans. The methods proposed in this invention can reveal the communication mechanisms between different brain functional states and different structures in healthy and diseased individuals of different age groups.

[0082] A device for quantifying brain structure-function coupling based on brain network gradients, the device comprising: a processor and a memory, wherein the memory stores program instructions, and the processor calls the program instructions stored in the memory to cause the device to perform the following method steps in Example 1:

[0083] Gradient echo sequence was used to acquire brain imaging data; axial single spin echo EPI sequence was used to acquire DTI data; and preprocessing operations were performed on DTI data and brain imaging data respectively.

[0084] The similarity matrix is ​​reduced in dimension using principal component analysis to obtain a structural gradient matrix, where each column is a principal component, i.e., the structural connection gradient; the similarity matrix is ​​reduced in dimension using principal component analysis to obtain a functional gradient matrix, where each column is a principal component, i.e., the functional connection gradient;

[0085] The structural connectivity matrix and the functional connectivity matrix are reduced in dimension using principal component analysis to obtain the structural gradient matrix and the functional gradient matrix;

[0086] The Pearson correlation coefficient between each row of the low-frequency structural gradient matrix and the low-frequency functional gradient matrix was calculated to obtain the low-frequency gradient coupling index. The Pearson correlation coefficient between each row of the high-frequency structural gradient matrix and the high-frequency functional gradient matrix was calculated to obtain the high-frequency gradient coupling index.

[0087] The low-frequency gradient coupling index and high-frequency gradient coupling index are used to capture subtle changes in brain structure or function.

[0088] Among them, the low-frequency gradient coupling index is:

[0089]

[0090] Among them, lr x is the Pearson correlation coefficient of each row of the low-frequency structural gradient and low-frequency functional gradient matrices, x = 1, 2, 3, …, N.

[0091] Among them, the high-frequency gradient coupling index is:

[0092]

[0093] Among them, hr x is the Pearson correlation coefficient of each row of the high-frequency structural gradient and high-frequency functional gradient matrix, x = 1, 2, 3, ..., N.

[0094] The method further includes calculating a ratio of a high-frequency gradient coupling index to a low-frequency gradient coupling index, referred to as a gradient coupling fraction:

[0095]

[0096] Among them, coindex ratio is the gradient coupling score matrix, hr x / lr x is the ratio of the high-frequency gradient coupling index to the low-frequency gradient coupling index, x=1, 2, 3,…, N.

[0097] It should be noted here that the device description in the above embodiment corresponds to the method description in the embodiment, and the embodiment of the present invention will not be described in detail here.

[0098] The execution subjects of the above-mentioned processor and memory can be computers, single-chip microcomputers, microcontrollers and other devices with computing functions. In specific implementation, the embodiment of the present invention does not limit the execution subject and it can be selected according to the needs of actual application.

[0099] Data signals are transmitted between the memory and the processor via a bus, which will not be described in detail in the embodiment of the present invention.

[0100] Based on the same inventive concept, an embodiment of the present invention further provides a computer-readable storage medium, which includes a stored program, and when the program is running, controls the device where the storage medium is located to execute the method steps in the above embodiment.

[0101] The computer-readable storage medium includes but is not limited to a flash memory, a hard disk, a solid-state drive, and the like.

[0102] It should be noted here that the description of the readable storage medium in the above embodiment corresponds to the description of the method in the embodiment, and the embodiment of the present invention will not be described in detail here.

[0103] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions according to the embodiments of the present invention are generated in whole or in part.

[0104] The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in or transmitted via computer-readable storage media. Computer-readable storage media can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that integrates one or more available media. Available media can include magnetic media or semiconductor media, etc.

[0105] Unless otherwise specified, the embodiments of the present invention do not limit the models of the components. Any component that can perform the above functions may be used.

[0106] Those skilled in the art will understand that the accompanying drawings are only a schematic diagram of a preferred embodiment, and the serial numbers of the embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A quantitative method for brain structure-function coupling based on brain network gradients, characterized by: The method comprises: Gradient echo sequence was used to acquire brain imaging data; axial single spin echo EPI sequence was used to acquire DTI data; and preprocessing operations were performed on DTI data and brain imaging data respectively. The similarity matrix is ​​reduced in dimension using principal component analysis to obtain a structural gradient matrix, where each column is a principal component, i.e., the structural connection gradient; the similarity matrix is ​​reduced in dimension using principal component analysis to obtain a functional gradient matrix, where each column is a principal component, i.e., the functional connection gradient; The structural connectivity matrix and the functional connectivity matrix are reduced in dimension using principal component analysis to obtain the structural gradient matrix and the functional gradient matrix; The Pearson correlation coefficient between each row of the low-frequency structural gradient matrix and the low-frequency functional gradient matrix was calculated to obtain the low-frequency gradient coupling index. The Pearson correlation coefficient between each row of the high-frequency structural gradient matrix and the high-frequency functional gradient matrix was calculated to obtain the high-frequency gradient coupling index. The low-frequency gradient coupling index and high-frequency gradient coupling index are used to capture subtle changes in brain structure or function.

2. A quantitative method for brain structure-function coupling based on brain network gradient according to claim 1 The method is characterized in that The low-frequency gradient coupling index is: ； in, is the Pearson correlation coefficient of each row of the low-frequency structural gradient matrix and the low-frequency functional gradient matrix, .

3. The quantitative method for brain structure-function coupling based on brain network gradient according to claim 1, characterized in that: The high-frequency gradient coupling index is: ； in, is the Pearson correlation coefficient of each row of the high-frequency structural gradient matrix and the high-frequency functional gradient matrix, .

4. The quantitative method for brain structure-function coupling based on brain network gradient according to claim 1, characterized in that: The method further includes calculating a ratio of a high-frequency gradient coupling index to a low-frequency gradient coupling index, referred to as a gradient coupling fraction: ； in, is the gradient coupling score matrix, is the ratio of the high-frequency gradient coupling index to the low-frequency gradient coupling index, .

5. A device for quantifying brain structure-function coupling based on brain network gradients, characterized in that: The device includes: a processor and a memory, wherein program instructions are stored in the memory, and the processor calls the program instructions stored in the memory to enable the device to execute the method according to any one of claims 1 to 4.

6. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions. When the program instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 1 to 4.

Images