Online detection of horizontal hand movements from low frequency EEG components

Recent studies show that there is a correlation between electroencephalographic (EEG) signals and hand-reaching kinematic parameters when linear regression decoding models are applied to low frequency EEG components. However, the decoding performance is far from being sufficient to obtain an accurate control. In this paper, we propose the use of this methodology to detect horizontal hand movements. To that end, subjects are asked to reach two targets on a screen by moving the computer mouse. First, the decoding models are computed from fast training runs where both mouse trajectories and EEG signals are recorded. Then, these models are applied to classify the movement direction in real time and compare it to the actual performed trajectories. The results show that success rates are significantly above chance levels and suggest that this could be a suitable technique to infer hand movement intention with a very short training.