Multivariate Granger causality analysis of fMRI data

Abstract

This article describes the combination of multivariate Ganger causality analysis, temporal down-sampling of fMRI time series, and graph theoretic concepts for investigating causal brain networks and their dynamics. As a demonstration, this approach was applied to analyze epoch-to-epoch changes in a hand-gripping, muscle fatigue experiment. Causal influences between the activated regions were analyzed by applying the directed transfer function (DTF) analysis of multivariate Granger causality with the integrated epoch response as the input, allowing us to account for the effects of several relevant regions simultaneously. Integrated responses were used in lieu of originally sampled time points to remove the effect of the spatially varying hemodynamic response as a confounding factor; using integrated responses did not affect our ability to capture its slowly varying affects of fatigue. We separately modeled the early, middle, and late periods in the fatigue. We adopted graph theoretic concepts of clustering and eccentricity to facilitate the interpretation of the resultant complex networks. Our results reveal the temporal evolution of the network and demonstrate that motor fatigue leads to a disconnection in the related neural network.

Whispering - The hidden side of auditory communication

“…All data showed covariance stationarity as indicated by the KPSS test, which is a test with the null hypothesis of no unit root. Finally, a model order of two was chosen based on previous studies with approximately half the sampling rate using a model order of one (Deshpande et al, 2009;Sato et al, 2010), and model validity was confirmed by the Durbin-Watson test. Significant directed connections based on F statistics between brain regions were determined at a threshold of p = 0.0001 (FDR corrected).…”

Investigations of Brain Network Alterations in Epilepsy Using Functional Magnetic Resonance Imaging

“…In several instances fMRI signal change in the DMN started prior to the event onset on EEG. Another study (Szaflarski, et al 2010) found parietal (but not necessarily DMN) activation occurring prior to thalamic activation and also detected similar causal links using Granger causality measures of fMRI data (Deshpande, et al 2009, Goebel, et al 2003. Another causal methodology, dynamic causal modeling , can estimate the influence of one system on another.…”

The impact of hemodynamic variability and signal mixing on the identifiability of effective connectivity structures in BOLD fMRI

“…After voxel‐wise application of GC, the result can be averaged over all voxel‐wise GC scores between the two regions. This voxel‐wise application of GC can be performed in multiple ways: the GC score can be averaged with LASSO regularization (Tang, Bressler, Sylvester, Shulman, & Corbetta, 2012), through a multivoxel pattern‐based causality mapping as proposed by Kim, Kim, Ahmad, & Park (2013), or through hierarchical clustering as proposed by Deshpande, LaConte, James, Peltier, & Hu (2009). This strategic twist is already being used in some studies (Katwal, Gore, Gatenby, & Rogers, 2013; Zhao et al., 2016), however, it is not the most common approach in the field yet [voxel‐wise modeling is, however, increasingly popular for finding activation patterns in cognition from fMRI data (Huth, de Heer, Griffiths, Theunissen, & Gallant, 2016)].…”