Localizing and estimating causal relations of interacting brain rhythms

Estimating brain connectivity and especially causality between different brain regions from EEG or MEG is limited by the fact that the data are a largely unknown superposition of the actual brain activities. Any method, which is not robust to mixing artifacts, is prone to yield false positive results. We here review a number of methods that allow for addressing this problem. They are all based on the insight that the imaginary part of the cross-spectra cannot be explained as a mixing artifact. First, a joined decomposition of these imaginary parts into pairwise activities separates subsystems containing different rhythmic activities. Second, assuming that the respective source estimates are least overlapping, yields a separation of the rhythmic interacting subsystem into the source topographies themselves. Finally, a causal relation between these sources can be estimated using the newly proposed measure Phase Slope Index (PSI). This work, for the first time, presents the above methods in combination; all illustrated using a single, simulated data set.