k-NN and LDA based Motor Imagery EEG Classification using Phase Features

A Brain-Computer Interface (BCI) based on motor imagery (MI) can be developed to translate the motor intent of a person for controlling some device. For this purpose, changes occurring in alpha and beta bands are exploited. In this paper we utilize measurement of coherence and phase synchrony (Phase features) to inspect the activities of sensory motor cortex during imagination of hand and feet movement. An attempt is made to establish relationship between changes occurring in the motor cortex with respect to phase. After bandpass filtering and spatial filtering for noise removal analysis of phase characteristics of these signals is carried out. These features are classified with the help of Linear Discriminant Analysis (LDA) and k-Nearest Neighbor (k-NN) classifiers. The classification performance suggests that phase and coherence features prove to be an additional set of features for differentiating between various motor imagery signals.