Application of phase clustering index in brain-computer interface feature extraction

Brain-computer interface(BCI) can be used to establish direct communication channel between brain and outside for movement handicapped patients.Extraction of identification features in the classification of BCI is important to improving BCI performance.Phase clustering index(PCI),as a BCI feature based on the steady state visual evoked potentials(SSVEP),emphasizes the phase consistency in different stimulating cycles,rightly reflecting the physiological mechanism of the SSVEP.The PCI classification accuracy and the commonly used power spectrum density(PSD) features were compared.The results show that,in a 6-target BCI system,the classification accuracy of PCI is significantly(p0.05) better than PSD.Therefore,using the PCI is more effective than using the PSD in SSVEP-based BCI classification.