DynamicBC: a MATLAB toolbox for dynamic brain connectome analysis - PubMed (original) (raw)
DynamicBC: a MATLAB toolbox for dynamic brain connectome analysis
Wei Liao et al. Brain Connect. 2014 Dec.
Abstract
The brain connectome collects the complex network architectures, looking at both static and dynamic functional connectivity. The former normally requires stationary signals and connections. However, the human brain activity and connections are most likely time dependent and dynamic, and related to ongoing rhythmic activity. We developed an open-source MATLAB toolbox DynamicBC with user-friendly graphical user interfaces, implementing both dynamic functional and effective connectivity for tracking brain dynamics from functional MRI. We provided two strategies for dynamic analysis: (1) the commonly utilized sliding-window analysis and (2) the flexible least squares based time-varying parameter regression strategy. The toolbox also implements multiple functional measures including seed-to-voxel analysis, region of interest (ROI)-to-ROI analysis, and voxel-to-voxel analysis. We describe the principles of the implemented algorithms, and then present representative results from simulations and empirical data applications. We believe that this toolbox will help neuroscientists and neurologists to easily map dynamic brain connectomics.
Keywords: brain connectome; dynamic; effective connectivity; functional connectivity; resting-state fMRI.
Figures
**FIG. 1.
The framework of the DynamicBC toolbox. A graphical user interface (GUI) can be started by calling the “DynamicBC'” function in the command window of the MATLAB. The following procedures include three parts: the selection of connectivity types (the functional connectivity [FC] and effective connectivity [EC]), selection of dynamic analysis strategies (the sliding-window and flexible least squares [FLS]), and selection of connectivity measures (the seed-to-voxel, region of interest [ROI]-to-ROI, and voxel-to-voxel analysis). The subsequent brain connectomes are then visualized. Color images available online at
**FIG. 2.
The simulated data and dynamic interactions. (A) Signals are generated according to equation (7). The blue, green, and red lines denote signals x1, x2, and y, respectively. They are normalized to a mean=0 and a variance=0.3. (B, C) The static FC (s-FC) and dynamic FC (d-FC) by sliding-window analysis is drawn as the peach-puff and green line, respectively. The 95% confidence intervals for β coefficient (red line) estimated by the Kalman filtering (KF) method is shown as the yellow-filled area, and the blue line indicated β estimated by the FLS method. Color images available online at
**FIG. 3.
The differential β amplitudes as estimated by the FLS method with different penalty weights μ, which increase by powers of ten: 0.01, 0.10, 1, 10, 100, 1000, and 10,000. The gray line indicates the ground truth β. Color images available online at
**FIG. 4.
Illustration of the seed-to-voxel-wise s-FC and d-FC. (A) Group-averaged s-FC map following the linear Pearson's correlation analysis using the full-length of the resting-state blood-oxygenation level-dependent (BOLD) functional magnetic resonance imaging (fMRI) signal with a seed placed in the posterior cingulate cortex (PCC; Montreal Neurological Institute [MNI] coordinates: −2, −48, 28, 6 mm radius sphere). (B) d-FC map following FLS analysis with the PCC seed of a representative healthy subject. Two target ROIs with a 6 mm radius sphere were placed in the medial prefrontal cortex (mPFC, MNI coordinates: 6, 51, 9), and right superior parietal lobule (SPL, MNI coordinates: 63, −33, 39). The d-FC (solid line) and s-FC (dashed line) time series of mPFC (red line) and SPL (blue line) varied in a time-dependent manner. Warm and cool colors indicate brain regions with positive and negative temporal correlations with the PCC seed, respectively. Color scales represent the group-averaged correlation coefficient value (Z) of the s-FC map and individual β amplitude values of the d-FC map, respectively. See Supplementary Movie S1; Supplementary Data are available online at
www.liebertpub.com/brain Color
images available online at
**FIG. 5.
Illustration for seed-to-voxel-wise static EC (s-EC) and dynamic EC (d-EC). (A) Group-averaged s-EC map following linear residual-based Granger causality analysis (GCA) using full length of resting-state BOLD fMRI signal with a seed placed in the PCC (MNI coordinates: −2, −48, 28, 6 mm radius sphere). (B) d-EC map following sliding-window analysis of linear residual-based GCA with the PCC seed of a representative healthy subject. One target ROI with 6 mm radius sphere placed in the mPFC (MNI coordinates: 3, 39, 18). The d-EC (solid line) and s-EC (dashed line) time series of mPFC from the in map (red line) and out map (blue line) vary across time. Warm and cool colors indicate brain regions with in influence (from whole brain to the seed) and out influence (from seed to whole brain), respectively. Color scales represent group-averaged GC value (F) of s-EC map and individual β amplitude values of d-EC map, respectively. See Supplementary Movies S2 and S3. Color images available online at
**FIG. 6.
Illustration for ROI-to-ROI wise s-FC and d-FC. (A) Group-averaged pairwise correlation matrix (upper panel) following linear Pearson correlation analysis using full length of resting-state BOLD fMRI signals from 43 ROIs. This correlation matrix was visualized by brain network (bottom panel). (B) Dynamic functional correlation matrix (upper panel) and brain network (bottom panel) following FLS analysis among pairwise ROIs of a representative healthy subject. The d-FC (solid line) and s-FC (dashed line) time series between the PCC and other three ROIs vary across time. Warm and cool colors in correlation matrix indicate positive and negative correlations, respectively. See Supplementary Movie S4. Color images available online at
**FIG. 7.
d-FC strength (d-FCS). We calculated the d-FCS using FLS analysis strategy of pairwise voxels of a representative patient with absence epilepsy. The yellow shadow indicates ictal period marked by simultaneously collected electroencephalography data. The d-FCS time series of the PCC (red line) and thalamus (THA; blue line) vary preictal, ictal, and postictal periods. See Supplementary Movie S5. Color images available online at
References
- Biswal B, Yetkin FZ, Haughton VM, Hyde JS. 1995. Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magn Reson Med 34:537–541 -PubMed
Publication types
MeSH terms
LinkOut - more resources
Full Text Sources
Other Literature Sources
Medical