Envelope detection based on online ICA algorithm and its application to motor imagery classification

Independent Component Analysis (ICA) is a promising tool for brain-computer interface (BCI). But most of ICA-based BCI researches only used batch ICA algorithm as the offline preprocessing step for EEG artifact removal and pattern enhancement. This paper explored a new approach of applying online ICA based on sliding window Infomax algorithm for BCI implementation. In addition to having good performance of blind source separation as traditional ICA, the proposed method has the characteristics that can synchronously realize the online envelope detection of multi-channel signals, which is that other methods do not have, such as Hilbert transform. Then the online ICA is applied to the envelope detection of mu rhythm evoked by motor imagery and good classification results of imagining left and right hand movement are achieved on real-life data.