White matter fiber bundle structure-function connection coupling method based on distance function

Abstract

The invention discloses a white matter fiber bundle structure-function connection coupling method based on a distance function. The method comprises the steps that brain imaging data are collected and preprocessed; extracting white matter fiber bundle structure attributes and function signals of the preprocessed brain imaging data; constructing a structure connection matrix and an individualized function connection matrix according to the white matter fiber bundle structure attributes and the function signals; measuring the connection weight similarity between corresponding columns of the structure connection matrix and the function connection matrix through a distance function, and carrying out white matter fiber bundle structure-function connection coupling according to the connection weight similarity; through the method, the relationship of the white matter fiber bundle structure and function of the brain can be quantified, and the method is suitable for brain structure and function analysis based on the white matter fiber bundle and research on brain development and mental diseases.

Description

Technical Field

[0001] The present application relates to the field of medical image processing, and in particular to a white matter fiber bundle structure-functional connection coupling method based on a distance function. Background Art

[0002] Currently, the research on brain structural networks and functional networks in the field of neuroscience is becoming increasingly in-depth. The traditional structural network is defined as the average fractional anisotropy value of the voxels passing through the connecting fibers between two regions based on the results of fiber bundle imaging. When there is at least one streamline fiber and both of its endpoints are located in two regions, the two regions are considered to be structurally connected. Traditional methods only focus on a single structural attribute. In fact, various structural attributes have different neural and genetic bases, and multiple structural attributes may provide further suggestions for the field of neuroscience.

[0003] At the same time, studies have shown that structural connectivity and functional connectivity are interrelated, which can provide comprehensive insights into subtle changes in brain activity. Traditional methods for quantifying structural-functional connectivity coupling use diffusion fiber tractography based on diffusion tensors reconstructed in white matter to measure structural features, while functional connectivity is estimated separately using functional magnetic resonance imaging in gray matter. Because diffusion fiber tractography-based structural connectivity and functional connectivity between gray matter regions are very different in nature, direct integration of these features provides limited insights into understanding the structural connectivity that underpins functional interactions. How to accurately quantify the relationship between white matter fiber tract structural connectivity and white matter functional connectivity remains a challenge.

[0004] In addition, the traditional quantitative structure and function relationship mainly uses the Pearson correlation method. However, this method can only measure the linear relationship between two variables and cannot adapt to the multidimensional space of structure and function. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a white matter fiber bundle structure-function connection coupling method based on distance function, which solves the problem that the traditional connection coupling method cannot intuitively measure the spatial distance between points and cannot adapt to the multi-dimensional space of structure and function.

[0006] In order to achieve the above-mentioned purpose of the invention, the technical solution adopted by the present invention is: a white matter fiber bundle structure-functional connection coupling method based on a distance function, comprising:

[0007] S1, collect brain imaging data and perform preprocessing;

[0008] S2, extracting the structural properties and functional signals of white matter fiber bundles from preprocessed brain imaging data;

[0009] S3. Construct a structural connectivity matrix and an individualized functional connectivity matrix based on the structural properties and functional signals of white matter fiber bundles;

[0010] S4, measuring the connection weight similarity between the corresponding columns of the structural connection matrix and the functional connection matrix by the distance function;

[0011] S5. Complete the white matter fiber bundle structure-functional connection coupling based on the connection weight similarity measurement value.

[0012] Further: In S1, the brain imaging data includes diffusion tensor imaging data and functional magnetic resonance imaging data.

[0013] Further: For diffusion tensor imaging data, the preprocessing steps include: denoising, B1 field correction, eddy current correction, DTI-T1 weighted alignment, model fitting, and inter-regional probabilistic fiber bundle tracking.

[0014] Further: For functional magnetic resonance imaging data, the preprocessing steps include: slice time correction, motion correction, normalization, delinear trending, temporal bandpass filtering, interference regression and smoothing.

[0015] Further: S2 includes:

[0016] S21. Divide the whole brain into N brain regions;

[0017] S22, extracting the anisotropy fraction, axial diffusivity, radial diffusivity, mean diffusivity and local diffusion uniformity of the preprocessed diffusion tensor imaging data as white matter fiber bundle structural properties;

[0018] S23. Extract the mean blood oxygen level dependence signal time series of each brain region from the preprocessed functional magnetic resonance imaging data as the functional signal.

[0019] Further: In S3, the structural connection matrix construction method includes:

[0020] S301, using the whole brain white matter fiber bundles as network nodes and the structural attributes of the white matter fiber bundles as structural feature vectors;

[0021] S303, constructing a structural connectivity matrix through the Spearman correlation of the structural feature vectors between any two brain regions.

[0022] Further: In S3, the method for constructing an individualized functional connectivity matrix includes:

[0023] S311. Using the whole brain white matter fiber bundles as network nodes, calculate the Pearson correlation coefficient of the paired averages of the mean blood oxygen level dependence signal time series of N brain regions in the whole brain;

[0024] S312. Construct an individual-level functional network based on the Pearson correlation coefficient of the paired average values ​​of the time series of the mean blood oxygen level dependence signals of N brain regions in the whole brain.

[0025] Further: In S4, the distance functions include Euclidean distance, standardized Euclidean distance, Manhattan distance, Minkowski distance, Chebyshev distance and Mahalanobis distance.

[0026] The beneficial effects of the present invention are:

[0027] 1. The structural similarity of white matter fiber bundles is evaluated based on five structural attributes: fractional anisotropy, axial diffusivity, radial diffusivity, average diffusivity, and local diffusion uniformity. This combines multiple attribute information to provide more comprehensive physiological significance.

[0028] 2. Provides evidence of the interaction between structure and function at the white matter level: The present invention is based on the division of fiber bundles in the white matter of the brain, and simultaneously quantifies the relationship between structure and function as nodes of the structural and functional networks, and can explore the complex interaction between structure and function at the white matter level;

[0029] 3. The current method of quantifying structure-function connection coupling has been improved, measuring six distance functions including Euclidean distance, standardized Euclidean distance, Manhattan distance, Minkowski distance, Chebyshev distance and Mahalanobis distance, and weighted averaging to quantify the relationship between structure and function, which is more prepared to reflect the association between the two modal signals from diffusion tensor imaging and functional magnetic resonance imaging. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Flowchart of the white matter fiber bundle structure-functional connectivity coupling method based on distance function. DETAILED DESCRIPTION

[0033] The specific implementation modes of the present invention are described below to facilitate those skilled in the art to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0034] like Figure 1 As shown, in one embodiment of the present invention, a white matter fiber bundle structure-functional connection coupling method based on a distance function is provided, comprising:

[0035] S1, collect brain imaging data and perform preprocessing;

[0036] Among them, brain imaging data include diffusion tensor imaging data and functional magnetic resonance imaging data;

[0037] S2, extracting the structural properties and functional signals of white matter fiber bundles from preprocessed brain imaging data;

[0038] S3. Construct a structural connectivity matrix and an individualized functional connectivity matrix based on the structural properties and functional signals of white matter fiber bundles;

[0039] S4, measuring the connection weight similarity between the corresponding columns of the structural connection matrix and the functional connection matrix by the distance function;

[0040] S5. Complete the white matter fiber bundle structure-functional connection coupling based on the connection weight similarity measurement value.

[0041] As shown in Figure 2, for diffusion tensor imaging data, the preprocessing steps include: denoising, B1 field correction, eddy current correction, DTI-T1 weighted alignment, model fitting, and inter-regional probabilistic fiber bundle tracking; for functional magnetic resonance imaging data, the preprocessing steps include: slice time correction, motion correction, normalization, delinear trending, temporal bandpass filtering, interference regression, and smoothing.

[0042] Specifically, S2 includes:

[0043] S21. Divide the whole brain into N brain regions;

[0044] S22, extracting the anisotropy fraction, axial diffusivity, radial diffusivity, mean diffusivity and local diffusion uniformity of the preprocessed diffusion tensor imaging data as white matter fiber bundle structural properties;

[0045] S23. Extract the mean blood oxygen level dependence signal time series of each brain region from the preprocessed functional magnetic resonance imaging data as the functional signal.

[0046] The structural connection matrix construction method includes:

[0047] S301, using the whole brain white matter fiber bundles as network nodes and the structural attributes of the white matter fiber bundles as structural feature vectors;

[0048] S303, constructing a structural connectivity matrix through the Spearman correlation of the structural feature vectors between any two brain regions.

[0049] The method of constructing an individualized functional connectivity matrix includes:

[0050] S311. Using the whole brain white matter fiber bundles as network nodes, calculate the Pearson correlation coefficient of the paired averages of the mean blood oxygen level dependence signal time series of N brain regions in the whole brain;

[0051] S312. Construct an individual-level functional network based on the Pearson correlation coefficient of the paired average values ​​of the time series of the mean blood oxygen level dependence signals of N brain regions in the whole brain.

[0052] In S4, the distance functions include Euclidean distance, standardized Euclidean distance, Manhattan distance, Minkowski distance, Chebyshev distance and Mahalanobis distance.

[0053] As shown in FIG3 , in one embodiment of the present invention, the distance function-based white matter fiber bundle structure-functional connectivity coupling method described in the present application is applied to depression and healthy controls;

[0054] Diffusion tensor imaging (DTI) and functional magnetic resonance imaging (fMRI) data of depression and healthy controls were obtained using magnetic resonance imaging equipment such as GE 750W.

[0055] During the entire scanning process, participants were instructed to keep their eyes closed, stay awake, and remain as still as possible. To rule out disease, routine axial brain T2-weighted imaging was performed; brain structure and function data were further preprocessed separately.

[0056] DTI images were registered to correct for motion artifacts and distortion. Fiber paths were reconstructed using a deterministic fiber tracking algorithm. The fiber tracking process started from the deep white matter (WM) region and was terminated if the crossing angle exceeded 35° in two consecutive translation directions or the fractional anisotropy (FA) was out of range (0.1-1).

[0057] Resting-state fMRI data preprocessing included discarding the first 10 time points to eliminate magnetization nonequilibrium effects and to allow subjects to acclimate to the scanning environment. The remaining data were corrected for time point and head motion artifacts. Participants with mean inter-frame displacement (FD) values ​​greater than 2 mm and deviation angles greater than 2° were excluded from the study. Functional images were normalized to the Montreal Neurological Institute (MNI) and resampled to 3 × 3 × 3 mm 3 The voxel size was 2000 × 1000 pixels, linear trend was removed and temporal bandpass filtering (0.01–0.08) was performed to remove high-frequency noise. Cerebrospinal fluid and head motion noise signals were regressed out, and white matter and global brain signals were not removed to avoid eliminating important neural signals. Finally, a 6 mm full-width smoothing kernel was applied to spatially smooth the images to reduce inter-individual variability.

[0058] Extraction of structural properties and functional signals at the white matter level: 48 white matter regions of interest were defined based on the JHU ICBM-DTI-81 white matter atlas. Fractional anisotropy, axial diffusivity, radial diffusivity, mean diffusivity, and local diffusion uniformity were extracted from the preprocessed DTI data, and we obtained a 48×5 feature vector representing the structural feature vector of each subject; the extraction of functional signals was to extract the mean blood oxygen level dependent signal (BOLD) time series of the brain region from the preprocessed fMRI data.

[0059] Using 48 fiber bundles of whole-brain white matter as network nodes, we calculated the Spearman correlation of the structural feature vectors between any two brain regions and constructed an individual-level structural similarity network. We obtained a symmetrical 48×48 connection matrix, representing the brain structural network of each subject. Subsequently, the present invention uses sparsity (threshold) to measure the significance of the correlation between any two brain regions. If the correlation between two brain regions exceeds a specific threshold (the threshold is in [-1,1] and can also be set according to statistical results), then the two regions are considered to be structurally connected. On the contrary, if the correlation between two brain regions is lower than the threshold, it indicates that they are not connected, and then a weighted symmetrical 48×48 structural connection matrix of each subject is obtained.

[0060] The 48 fiber bundles in the whole brain white matter were used as network nodes. The Pearson correlation coefficient of the pairwise average of the BOLD time series of the 48 white matter regions of the brain was calculated to construct the individual-level functional network, which was converted into z scores using Fisher's r to z transformation to obtain a standardized 48×48 functional connectivity matrix for each subject.

[0061] The similarity of the structural and functional network connection weights of each brain region was calculated based on six distance functions: Euclidean distance, standardized Euclidean distance, Manhattan distance, Minkowski distance, Chebyshev distance and Mahalanobis distance.

[0062] For each distance, the distance between each column in the structural connectivity matrix and the corresponding column in the functional connectivity matrix represents the regional structural connectivity-functional connectivity coupling for each individual. Then, each individual obtained a vector of length 48, in which each element represents the structural connectivity-functional connectivity coupling value between the region and all other regions in the whole brain region of interest. Finally, based on the target, a weight is assigned to each distance metric, and the weighted average is used to obtain the final structural connectivity-functional connectivity coupling.

[0063] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A distance function-based white matter fiber bundle structure-functional connectivity coupling method, characterized in that: include: S1, collect brain imaging data and perform preprocessing; S2, extracting the structural properties and functional signals of white matter fiber bundles from preprocessed brain imaging data; S3. Construct a structural connectivity matrix and an individualized functional connectivity matrix based on the structural properties and functional signals of white matter fiber bundles; S4, measuring the connection weight similarity between the corresponding columns of the structural connection matrix and the functional connection matrix by the distance function; S5. Complete the white matter fiber bundle structure-functional connection coupling based on the connection weight similarity measurement value.

2. The method for coupling white matter fiber bundle structure-function connectivity based on distance function according to claim 1, characterized in that: In S1, brain imaging data include diffusion tensor imaging data and functional magnetic resonance imaging data.

3. The method for coupling white matter fiber bundle structure-function connectivity based on distance function according to claim 2, characterized in that: For diffusion tensor imaging data, preprocessing steps included: denoising, B1 field correction, eddy current correction, DTI-T1 weighted alignment, model fitting, and inter-regional probabilistic fiber tract tracking.

4. The method for coupling white matter fiber bundle structure-function connectivity based on distance function according to claim 2, characterized in that: For fMRI data, preprocessing steps included slice time correction, motion correction, normalization, delinear trending, temporal bandpass filtering, noise regression, and smoothing.

5. The method for coupling white matter fiber bundle structure-function connectivity based on distance function according to claim 1, characterized in that S2 include: S21. Divide the whole brain into N brain regions; S22, extracting the anisotropy fraction, axial diffusivity, radial diffusivity, mean diffusivity and local diffusion uniformity of the preprocessed diffusion tensor imaging data as white matter fiber bundle structural properties; S23. Extract the mean blood oxygen level dependence signal time series of each brain region from the preprocessed functional magnetic resonance imaging data as the functional signal.

6. The method for coupling white matter fiber bundle structure-function connectivity based on distance function according to claim 5, characterized in that: In S3, the structural connectivity matrix construction method includes: S301, using the whole brain white matter fiber bundles as network nodes and the structural attributes of the white matter fiber bundles as structural feature vectors; S303, constructing a structural connectivity matrix through the Spearman correlation of the structural feature vectors between any two brain regions.

7. The method for coupling white matter fiber bundle structure-function connectivity based on distance function according to claim 5, characterized in that: In S3, the method for constructing an individualized functional connectivity matrix includes: S311. Using the whole brain white matter fiber bundles as network nodes, calculate the Pearson correlation coefficient of the paired averages of the mean blood oxygen level dependence signal time series of N brain regions in the whole brain; S312. Construct an individual-level functional network based on the Pearson correlation coefficient of the paired average values ​​of the time series of the mean blood oxygen level dependence signals of N brain regions in the whole brain.

8. The method for coupling white matter fiber bundle structure-function connectivity based on distance function according to claim 1, characterized in that: In S4, the distance functions include Euclidean distance, standardized Euclidean distance, Manhattan distance, Minkowski distance, Chebyshev distance and Mahalanobis distance.

Images