Detection of movement intentions through a single channel of electroencephalography

Abstract

In this study, the aim was to estimate the performance of a braincomputer interface (BCI) system by detecting movement intentions using only a single monopolar channel of electroencephalography (EEG). Seven healthy subjects performed four types of cued palmar grasps with two levels of force and speed. The movement intentions were detected using a technique where a template of the initial negative phase of the movement-related cortical potential (movement intention) was matched with the continuous EEG. On average 78 % of the movements were detected (true positive rate) ~150 milliseconds before the task onset. The number of false positive detections was 1.5 per minute. The estimated system performance, using only a single monopolar channel, indicates that the proposed setup can be used for neuromodulation paradigms in BCI where the movement intention is paired with somatosensory feedback from e.g. functional electrical stimulation or robot-assisted movements.