Advanced EEG Signal Processing

Abstract

The study investigates the performance of some EEG signal processing methods in detecting the signal variations within the Event-Related Potential (ERP) and in extracting the EEG effective connectivity, and the obtained results are discussed. The advantage of applying the Independent Component Analysis (ICA) in EEG analysis is also considered. The EEG data are recorded in the framework of BCI 2005 competition, during a motor imager task, and includes segments of Event-Related (De)synchronization, revealed be the proposed signal processing methods: Event-Related Spectral Perturbation, , the Inter-Trial Phase Coherence, the Inter-Trial Linear Coherence and the Event Related Cross- Coherence. The effective connectivity is analyzed in time and frequency domain, by applying the Granger Causality Index (GCI) and the Partial-Directed Coherence (PDC) respectively, as time-variant or time-invariant methods.