Data-driven MEG analysis to extract fMRI resting-state networks

The electrophysiological basis of resting-state networks (RSN) is still under debate. In particular, no principled mechanism has been determined that is capable of explaining all RSN equally well. While magnetoencephalography (MEG) and electroencephalography are the methods of choice to determine the electrophysiological basis of RSN, no standard analysis pipeline of RSN yet exists. In this article, we compare the two main existing data-driven analysis strategies for extracting RSNs from MEG data and introduce a third approach. The first approach uses phase-amplitude coupling to determine the RSN. The second approach extracts RSN through an independent component analysis of the Hilbert envelope in different frequency bands, while the third new approach uses a singular value decomposition instead. To evaluate these approaches, we compare the MEG-RSN to the functional magnetic resonance imaging (fMRI)-RSN from the same subjects. Overall, it was possible to extract RSN with MEG using all three techniques, which matched the group-specific fMRI-RSN. Interestingly the new approach based on SVD yielded significantly higher correspondence to five out of seven fMRI-RSN than the two existing approaches. Importantly, with this approach, all networks-except for the visual network-had the highest correspondence to the fMRI networks within one frequency band. Thereby we provide further insights into the electrophysiological underpinnings of the fMRI-RSNs. This knowledge will be important for the analysis of the electrophysiological connectome.