A K-means multivariate approach for clustering independent components from magnetoencephalographic data

► We define an automatic strategy to capture spatio-temporal dynamics of brain activity. ► We propose a new set of fingerprints to summarize global properties of MEG-ICs. ► IC-fingerprints are clustered across sessions and subjects. ► The improved approach automatically computes parameter weights and cluster number. ► Consistency of clusters spatial topography and ERD-gram is tested on two protocols.