Rapid detection of voluntary movements in a self-paced brain-computer interface via compressive sensing

This study employs compressive sensing (CS) for accelerating the overall data processing in a brain-computer interface (BCI). CS is a joint signal acquisition and compression scheme. By projecting EEG data streams onto a lower dimensional basis that is incoherent with their inherent structure, CS compresses the data and preserves their salient information. This paper presents a novel self-paced BCI (SBCI) that extracts compact and relevant features from EEG using CS. To detect an intentional control (IC) state from user´s EEG, the signal structure of a specific voluntary movement that is common in all trials is identified. This compressed subset of basis functions is constructed during the BCIs training mode and is then used to acquire features during its testing mode. Experimental results show that our proposed method can efficiently detect voluntary movements for a SBCI, with potential speed increases up to 12 times from the state-of-art prototype.