White-matter functional organization
Functional MRI signals from the brain's white-matter have been previously described but were largely neglected in research. We have recently demonstrated that white-matter signals, obtained during the resting state, can be clustered to uncover patterns of intrinsic connectivity within the white-matter. We describe the identification of 12 functional networks within the white-matter and their characterization.
Further details can be found in our paper "Evidence for functional networks within the human white matter", by Michael Peer, Mor Nitzan, Atira Bick, Netta Levin and Shahar Arzy (Journal of Neuroscience 2017).
Our full analysis scripts as well as the resulting functional networks and their 3D visualization results are described below.
for questions please contact michael.peer (at) mail.huji.ac.il .
Analysis software
Our full analysis codes, which can be used to uncover white-matter networks from resting-state data, can be downloaded here: white-matter_codes.zip
These scripts require Matlab and SPM to be installed. To use them, unzip the file contents to a new directory and add this directory and its subfolders to the Matlab path.
The zip file contains the following scripts:
- smooth_WM_GM_separately.m - pre-processing tool to apply smoothing separately to the white-matter and grey-matter, prior to the analysis.
- create_average_WM_and_GM_masks.m - script to obtain group averaged white-matter and grey-matter masks from SPM segmentation results.
- clustering_WM_networks.m - the main analysis script: uses the functional data from many subjects, computes correlations between white-matter voxels' signals, and performs K-means clustering on the results to identify functional networks within the white-matter.
- clustering_GM_networks.m - similar to the previous script but on grey-matter signals, to identify known cortical resting-state networks.
- check_clustering_for_different_numbers_of_subjects.m - measures clustering stability for groups of subjects of different sizes.
- compare_GM_and_WM_restingstate_networks.m - compares the signals in white-matter networks to that of grey-matter networks.
- compare_DTI_fibers_to_WM_restingstate_networks.m - calculates the spatial overlap of white-matter resting-state networks to anatomical DTI fibers.
- measure_components_symmetry.m - measures the symmetry of the functional components (left-right hemispheres correspondence).
- FFT_WM_GM_networks_timecourses.m - Computes Fourier transform for average signals from each white- and grey-matter network.
- compute_seed_connectivity.m - Performs a seed-based analysis in the white-matter and compares to clustering results.
- clusters_borders_correlation_analysis.m - Checks whether borders of clusters are characterized by abrupt changes in correlation between voxels.
White-matter functional networks data
The results of application of clustering on resting-state signals of 176 subjects from the NKI-Rockland database can be downloaded here: white-matter_data.zip
The zip file contains the following nifti files (results of our analyses, normalized to MNI space, as described in the paper):
- WM_clustering_K12.nii - results of the clustering of the white-matter into 12 functional networks
- WM_clustering_K12.nii - results of the clustering of the grey-matter into 9 functional networks
- WMmask_group_segmentation06_above08subjsnotnan.nii,
GMmask_group_segmentation02_noWMmask_above08subjsnotnan.nii - group average white-matter and grey-matter masks used in the study to identify voxels for clustering
- Directories All_clustering_results_white_matter, All_clustering_results_grey_matter - contain the results of clustering to different numbers of networks, from 2 to 22
- Directory Average_DTI_fibers_nifti - group average white-matter anatomical fibers from 11 subjects, identified using the AFQ algorithm (Yeatman et al. 2012)
3D visualizations of clustering results
The 3D visualizations can be downloaded here: white-matter_visualization.zip
The results are saved in VTK format. To visualize, create the directory c:\white-matter\ and unzip the file contents there. Download the ParaView software (www.paraview.org/download/), and use "file->load state" to load any of the following PVSM files:
- WM_networks_K2/5/8/12.pvsm - the results of clustering to different numbers of clusters
- GM_networks_K9.pvsm - the results of grey-matter clustering to 9 networks
- DTI_fibers.pvsm - the group averaged DTI fibers
