Eye Movement Artefact Suppression Using Volterra Filter for Electroencephalography Signals
A. M. Torres,
E. M. Sanchez-Morla,
J. L. Santos
Ocular artefacts are the most important form of interference in electroencephalography (EEG) signals. These eye movements have high amplitude and a frequency band that overlaps that of physiological or brain signals. It is important to remove these ocular artefacts before analyzing EEG signals to obtain accurate information about brain activity and avoid mistakes in its interpretation. This study presents an ocular artefact removal method based on a Volterra filter (VF) with a multichannel structure. Based on applications with various real and synthetic signals, the accuracy of the VF is compared with those of existing methods based on principal component analysis, support vector machines, and independent component analysis. The VF method has lower values of normalized mean square error, correlation coefficient, and spectral content change, and higher eye reduction, providing a good trade-off between removing artefacts and preserving inherent brain activities. The VF method can thus serve as an effective framework for substantially removing eye interference in EEG recordings.