Effective connectivity analysis of fMRI time-series based on Granger causality and complex network

Abstract

This paper develops a method to explore effective connectivity for time-series by using Granger causality and complex network. The Granger causality of multivariable time-series are analyzed based on VAR model, by which the weighed causality graph is built up to reveal a variety of causal relationship among components of time-series. Then the directed and weighted connectivity in Granger causality graph is described with complex network measures, and the statistical properties of multivariable time-series are characterized according to network topological parameters. Simulation and experiment analysis demonstrate that the proposed method is effective in testing the causality of fMRI time-series.