Neuro-fuzzy classification of brain computer interface data using phase based feature

Brain-Computer Interface is an interface technique between human and computer which can help severely motor-disabled persons to communicate and control their environment. In this study we have examined phase locking value as a possible feature to use in BCI systems based on the adaptive neuro-fuzzy inference system (ANFIS). To provide this feature, for classification of three motor imagery tasks, phase locking value was calculated for two pairs of electrodes, FCz-C3 and FCz-C4. The effect of different frequency bands was investigated as well. Result indicate that accuracy of classification between foot and hands movement imagery was more than the classification of right and left hand motor imagination. And broadband classification was more accurate than narrowband.