INSOMNIA EEG SIGNAL PREPROCESSING USING ICA ALGORITHMS

Abstract

Polysomnography (PSG) is a technique involved on the sleep disorders diagnostic. The signals acquired in a PSG study contain at least the electroencephalogram, the electrocardiogram, the electromiogram,the electrooculogram. Component Independent Analysis is a blindsource separation technique that has been shown to be very effec-tive in removing noise and artifacts that contaminate EEG signals.Inthis article, we will discuss the different ICA algorithms and thenapply them to denoising the EEG signal. This lead to well making decision regarding to this kind of disorder. These algorithms will beapplied for the denoising of the EEG signal containing insomniadisorders. The database used is the “CAP Sleep database” which isa collection of 108 polysomnographic recordings recorded in theCenter of Sleep Disorders at Ospedale Maggiore in Parma, Italy.Finally, theoretical and simulation results are presented to comparethe differents ICA algorithms applied to Insomnia EEG signals