Comparison of Sensor-Space and Source-Space ICAs in Reconstructing Resting State Networks from EEG Data

Abstract

Resting state networks (RSNs) has been extensively studied using functional magnetic resonance imaging (fMRI). Recently, novel methods to study RSNs using electroencephalography (EEG) have also been developed. In these studies, independent component analysis (ICA) is an important step in deriving RSNs and it can be applied to sensor space or source space data obtained through inverse source imaging techniques. Differences have been observed in results via both techniques even using same data. In the present study, we compared the sensor-space ICA and source-space ICA using simulated and real resting-state EEG data. The present results revealed that the source-space ICA has better performance on reconstructing both spatial and temporal features of RSNs.