Motor Tasks Classification Using Phase Locking Value in a BCI Based EEG Paradigm

Brain-computer interface (BCI) is developing very quickly with applications extending to medical and non-medical fields. Electroencephalography (EEG) is used in BCI to detect and analyze brain signals. An approach based on phase synchronization was tested on two datasets (one with EEG signals recorded from 15 healthy subjects and one with EEG signals recorded from 9 subjects with disabilities). Phase locking value was tested as feature extraction method from EEG signals. k-nearest neighbor (KNN) and support vector machine (SVM) classifiers were applied for discrimination between tasks (right hand motor imagery, left hand motor imagery and feet motor imagery). Classification rates above 81% obtained with kNN and 92% achieved with SVM indicate that phase synchronization based method can be exploited in developing BCI systems for controlling and assisting people with upper and lower limb disabilities.