Graph Theory GLM (GTG)
This Matlab toolbox calculates & runs a GLM on graph theory properties (i.e., invariants) derived from brain networks. The GLM accepts continuous & categorical between-participant predictors & categorical within-participant predictors. Significance is determined via non-parametric permutation tests. Both fully connected & thresholded networks can be tested. GTG uses the Brain Connectivity Toolbox to calculate many of the graph properties. Several normalizing options are available (to reduce dependence of higher order properties on lower order attributes like density, total strength, etc.), including division of weights by the mean/median/max.
The toolbox also provides a data processing path for resting state & task fMRI data, along with calculating connectivity matrices from fMRI timeseries. Options for partialing nuisance signals include (but are not limited to): local & total white matter signal (Jo et al., 2013), PCA of white matter/ventricular signal (Muschelli et al., 2014), Saad et al. (2013)’s GCOR, & Chen et al. (2012)’s GNI. In addition, Power et al. (2014)’s motion scrubbing method & Patel et al. (2014)’s WaveletDespike are available. Several options for calculating connectivity are available, including standard Pearson’s correlation, several types of robust correlation, rank-order correlation, two types of partial correlation, mutual information, along with several copulas.
See the NITRC page to download the last stable version of the toolbox: www.nitrc.org/projects/metalab_gtg/
***We are currently testing a highly updated version that includes a number of useful improvements (e.g., normalization of properties via randomized matrices). This version is not ready for wide release, but is available by contacting us (email@example.com).
Conference abstract on toolbox:
Spielberg JM. (2014). Graph theoretic general linear model (GTG): a MATLAB toolbox. Brain Connectivity, 4, A1-A158. doi:10.1089/brain.2014.1501.abstracts
Resting state pathway & graph theory analysis:
Spielberg JM, McGlinchey RE, Milberg WP, Salat DH. (2015). Brain network disturbance related to posttraumatic stress & traumatic brain injury in veterans. Biological Psychiatry, 78, 210-216. doi:10.1016/j.biopsych.2015.02.013
Block-design task pathway & graph theory analysis:
Spielberg JM, Miller GA, Heller W, Banich MT. (2015). Flexible brain network reconfiguration supporting inhibitory control. Proceedings of the National Academy of Sciences, 112, 10020-10025. doi:10.1073/pnas.1500048112
Node Induced Centrality (NIC)
This tool computes the contribution of each node to the centrality of a target node, thus identifying nodes that support the target node in effective network function.
Matyi MA, Cioaba S, Banich MT, & Spielberg JM. (2021). Identifying brain regions supporting amygdalar functionality: Application of a novel graph theory technique. NeuroImage, 244, 118614. doi:10.1016/j.neuroimage.2021.118614
Note for all scripts and toolboxes: All scripts are beta versions. There are no known bugs. This software comes with no warranty (even the implied warranty of merchantability or fitness for a particular purpose). Therefore, USE AT YOUR OWN RISK!!! Copyleft 2014-2016. Software can be modified and redistributed, but modified, redistributed versions must have the same rights.
Matyi MA, Spielberg JM. (2021). Differential spatial patterns of structural connectivity of amygdala nuclei with orbitofrontal cortex. Human Brain Mapping.
Atlases (3d MNI & FreeSurfer FSAverage):
HCP diffusion-based connectivity matrices used in the publications below:
Matyi MA, Cioaba S, Banich MT, Spielberg JM. (2021). Identifying brain regions supporting amygdalar functionality: A complex anatomical network perspective.
Matyi MA, Spielberg JM (under review). The structural brain network topology of episodic memory