Adaptive non-linear principal component analysis of a saccade related EEG component

Abstract

Saccades are rapid changes in the orientation of the eyes for realigning the visual axes on objects of interest. Dysfunction in this system may affect various visual functions such as depth perception and reading. A signal source separation method called nonlinear principal component analysis (NLPCA) was used to analyse saccade related EEG waveforms recorded from seven normal subjects. A number of components with the main features of the lambda wave were extracted from the parieto-occipital area of the visual cortex. The peak of activities of these components corresponded to discrete locations and showed a symmetry over the left and right hemisphere. In a previous study the application of another signal source separation method called independent component analysis (ICA) to the same data had indicated that the peak of activity of the lambda wave spread across the whole parieto-occipital area. The findings of this study indicated that ICA considered the whole parieto-occipital area to be a single source for the lambda-wave whereas NLPCA identified a number of symmetrical independent sources for the lambda wave in that region. (6 pages)