ICA+OPCA for artifact-robust classification of EEG data

EEG-based brain computer interface (BCI) provides a new communication channel between human brain and computer. An important task in an EEG-based BCI system is to analyze EEG patterns. EEG data is a multivariate time series, so hidden Markov model (HMM) might be a good choice for classification. However EEG is very noisy data and contains artifacts, thus the extraction of features that are robust to noise and artifacts is important. In this paper we present a method, which employ both independent component analysis (ICA) and oriented principal component analysis (OPCA) for artifact-robust feature extraction. The high performance of our method is confirmed by experimental study on classifying EEG into 4 categories, which consist of left/right/up/down movements during imagination.