A Polarity Alignment Method for Group-Averaging of Event-Related Neural Signals at Source Level in MEG Beamforming Applications.

Brain waveforms reconstructed at source level, like in beamforming, suffer polarity indeterminacy, which precludes direct group averaging of associated waveforms. We describe a polarity alignment method as an alternative of averaging rectified (i.e. absolute value) waveforms. Using MEG from an auditory localisation task, we compare the ability of the two approaches to enable signal detection in the primary auditory cortex over increasing sample size. The two methods are comparable in signal detection sensitivity, but the alignment method preserves group-average polarity alternation.