Coherences among all possible electrodes and all frequencies produce thousands of variables. Principal components analysis (PCA) allows objective reduction of coherence data dimensionality to a much smaller number of statistically independent coherence factors, typically no more than 40, with minimal loss of information content [32-36]. Furthermore, PCA reduction of coherence data sets obviates the need to reduce data on the basis of a priori specified brain connectivity selections, and thus avoids the potential of investigator bias.
