ICA Gives Higher-Order Functional Connectivity of Brain

Previously, only pairwise temporal correlation and Principal Component Analysis have been exploited for functional connectivity of the brain, which were based on second-order statistics. In this work, we developed a new concept of functional connectivity in the higher-order statistical sense and used Independent Component Analysis (ICA) for detecting connected brain regions. The fMRI datasets from resting brain were processed with eigenimage analysis and ICA, respectively, which showed an accordance with previous observations. The patterns of functional connectivity from both techniques were also compared. The applicability of ICA was demonstrated by finding higher-order functional connectivity with more number of components and less noisy imperfection. Keywords-Functional connectivity, higher-order connectivity, eigenimage analysis, Independent Com- ponent Analysis