Multivariate synchronization index for frequency recognition of SSVEP-based brain-computer interface

Multichannel frequency recognition methods are prevalent in SSVEP-BCI systems. These methods increase the convenience of the BCI system for users and require no calibration data. A novel multi-variate synchronization index (MSI) for frequency recognition was proposed in this paper. This measure characterized the synchronization between multichannel EEGs and the reference signals, the latter of which were defined according to the stimulus frequency. For the simulation and real data, the proposed method showed better performance than the widely used canonical correlation analysis (CCA) and minimum energy combination (MEC), especially for short data length and a small number of channels. The MSI was also implemented successfully in an online SSVEP-based BCI system, thus further confirming its feasibility for application systems. Because fast and accurate recognition is crucial for practical systems, we recommend MSI as a potential method for frequency recognition in future SSVEP-BCI. (C) 2013 Elsevier B.V. All rights reserved.