Movement imagery classification on the basis of single-trial EEGs

Abstract

In order to develop a new brain-computer interface (BCI) using EEGs, the independent component analysis (ICA), multiple equivalent current dipole source localization (ECDL) method and then Hayashi’s quantification method II were applied to single-trial EEGs during movement imagery. Up to the 20th independent component, each of them contained one or two positive (or negative) peaks in terms of 32-channel EEGs reconstructed on the electrode surface by the deflation procedure, and these peaks were analyzed by 1- or 2-dipole source localization method. Moreover, the quantification method based on these ECDL results yielded a good discrimination between right- and left-hand movement imageries, even if regardless of the time course of the localization results, instead of only for one male subject.