fMRI based Granger causality as a measure of effective connectivity in macaque visual cortex
Bernstein Center for Computational Neuroscience, Humboldt-Universität zu Berlin, Philippstraße 13, 10115 Berlin, Germany, Germany
Institut für Theoretische Physik, Technische Universität Berlin, Germany
Institute of Neuroscience, Newcastle university, United Kingdom
Granger causality (GC) analysis of fMRI blood oxygen level dependent (BOLD) time series has been proposed as a statistical tool to analyse directional influences between remote brain regions. Ongoing BOLD activity is treated as a set of stochastic time series (STS) generated by stochastic processes whose properties can be captured by autoregressive models. GC quantifies the improvement in predicting one brain region STS by inclusion of that from another region. Here, we show how GC can be tested using the ensemble of single-voxel time series in a region of interest (ROI). Granger causality was tested on F statistics derived from linear regression models using BOLD time series. Ongoing BOLD activity was extracted from four ROIs: three in V1 and one in MT/V5 area of the macaque visual cortex. fMRI scans were made under three stimulus conditions, whereby (1) a single (‘centre’) grating was presented, (2) the single grating was flanked by 2 additional gratings, and (3) only the flanking gratings were shown. ROIs were not spatially homogenous in their influence on each other; we found a variable number of voxel pairs with significant F statistics. GC analysis showed that the V1 ROI which represented the ‘centre’ grating was influenced by the MT/V5 ROI when centre-flanker stimuli were shown, but not by the V1 ROIs representing the flanking gratings. It thus appears that centre-flanker interactions (at distances >3 deg of visual angle) arise mainly through feedback, not through lateral intra-areal connections.
BC11 : Computational Neuroscience & Neurotechnology Bernstein Conference & Neurex Annual Meeting 2011, Freiburg, Germany, 4 Oct - 6 Oct, 2011.
neurons, networks and dynamical systems (please use "neurons, networks and dynamical systems" as keywords)
(2011). fMRI based Granger causality as a measure of effective connectivity in macaque visual cortex.
Front. Comput. Neurosci.
BC11 : Computational Neuroscience & Neurotechnology Bernstein Conference & Neurex Annual Meeting 2011.
19 Aug 2011;
04 Oct 2011.
Dr. Vesna Vuksanovic, Bernstein Center for Computational Neuroscience, Humboldt-Universität zu Berlin, Philippstraße 13, 10115 Berlin, Germany, Berlin, Germany, 10115, Germany, firstname.lastname@example.org