Predicting Intention and Direction of Wrist Movement from EEG

Brain-computer interfaces (BCI) offer potential for individuals with a variety of movement and sensory disabilities to control their environment, communicate and control mobility aids. However, the key to BCI usability rests in being able to extract relevant time varying signals that can be classified into usable commands. In this study we report on the results of experiments investigating the ability to classify scalp EEG signals on the basis of a users intention to move (and imaging to move) their wrist in different directions. EEG activity recorded from the scalp overlying the sensorimotor cortex was examined in the frequency domain to identify pre-movement patterns of synchronisation and desynchronization. Based on this, a further classification of the EEG epochs was performed based on Principal Component Analysis for feature extraction and Euclidean distance for intention classification. Classification success rates between 70-90% have been obtained using this relatively simple method suggesting that classification of pre-movement potentials can realistically be achieved in real time.