Disclosing brain functional connectivity from electrophysiological signals with phase slope based metrics

The characterization of the coupling direction between brain regions is fundamental for disclosing brain functioning. To this end, several computational methods have been developed that exploit either the temporal or the spectral characteristics of electrophysiological signals measured by e.g. EEG and MEG. Among these methods, the Phase Slope Index (PSI) estimates the directionality of frequency-specific neural interactions by relying on the sine of the phase slopes of the complex coherencies between time series, which is just an approximation for small angles of the actual phase slopes. The purpose of our study is to: 1) build a directionality estimator, namely Psi(id), which directly takes into account the non-approximated phase slopes; 2) assess the performance in estimating the coupling direction of PSI and Psi(id) in exhaustive simulations. Our findings show that while Psi(id) obtains better performance than PSI for the no noise case, a Signal-to-Noise Ratio equal or lower than one completely reverses the results in favour of PSI.