Inferring task-related networks using independent component analysis in magnetoencephalography

► We outline two methods for detecting functional networks in task positive MEG data. ► First is an analysis to find the optimum time-frequency window for FC detection. ► The second combines independent component analysis with the general linear model. ► Methods are applied to resting state and working memory task data. ► ICA/GLM reduces multiple comparisons and localises hippocampi.