Towards Classifying Motor Imagery Using a Consumer-Grade Brain-Computer Interface

Abstract

This research attempts to classify electroencephalogram (EEG) signals of motor imagery of left and right hand movement with a consumer-grade brain-computer interface device, which consists of four channels. For this purpose, we designed an interface to collect a total of approximately 600 samples for left and right hand motor imagery from two subjects. Hilbert-Huang Transform was used for feature extraction, and we applied support-vector machine (SVM) and k-nearest neighbors (k-NN) algorithms for learning the features and classification. Results show that these methods have some ability to classify left and right hand motor imagery EEG signals. This paper outlines the used methodology which could be a reference for future studies of the same nature.