Decoding Speed and Direction of Imagined Hand Movement from EEG-BCI

Motor Imagery-based Brain Computer Interface (BCI) system, that decodes imagined movements from non invasive electroencephalogram (EEG) are of significant importance, as it can enhance neurorehabilitation and human- computer interaction. Decoding the kinematic parameters of imagined movement is essential to realize BCI systems with higher degrees of freedom of movement and precise control over external effectors. In this work, we propose an efficient algorithm to decode imagined bi-directional movements of hand at two different speeds, slow and fast, from EEG-based BCI. EEG is recorded from fourteen healthy subjects, while they imagined slow and fast movement of their right hand towards the right or left direction. Wavelet-Common Spatial Pattern (WCSP) features and Movement Related Cortical Potential (MRCP) features are extracted from the EEG and a subset of these features are further identified based on the subject-specificity of discriminative subbands and channels. Selected WCSP and MRCP features are then concatenated and used to decode the imagined slow and fast bi-directional movements. Binary classification of the speed-direction pairs resulted in an average classification accuracy of 68.1% across fourteen subjects. To our knowledge, this is the first work addressing decoding of imagined speed and direction of hand movements using EEG-BCI.