Classification and interactive segmentation of EEG synchrony patterns

This paper presents a novel methodology for the exploratory analysis of power and synchronization patterns in EEG data from psychophysiological experiments. The methodology is based on the segmentation of the time–frequency plane in regions with relatively homogeneous synchronization patterns, which is performed by means of a seeded region-growing algorithm, and a Bayesian regularization procedure. We have implemented these methods in an interactive application for the study of cognitive experiments, although some of the techniques discussed in this work can also be applied to other multidimensional data sets. To demonstrate our methodology, results corresponding to a figure and word categorization EEG experiment are presented.