Topographic mapping and automatic classification of electroencephalographic signals

Abstract

An automatic classification system of electroencephalographic data, BIAC (brain imaging and automatic classification), is presented. Special emphasis is placed on exploring topographic imaging for classification purposes and performing correlation analysis between channels in order to find a pattern of normality. The methods were applied to visual-evoked potentials, and classification was made in terms of the normality of the signals from two populations: controls and patients with hepatic cirrhosis. It was shown that symmetry features, being simple measures on topographic maps for particular time instants, are able to discriminate between populations. Correlation features evidenced different patterns for the two populations under study. The selection of four of these features proved useful in discriminating between the populations.<>