Imagined 3D Hand Movement Trajectory Decoding from Sensorimotor EEG Rhythms

Reconstruction of the three-dimensional (3D) trajectory of an imagined limb movement using electroencephalography (EEG) poses many challenges. However, if achieved, more advanced non-invasive brain-computer interfaces (BCIs) for the physically impaired could be realized. The most common motion trajectory prediction (MTP) BCI employs a time-series of band-pass filtered EEG potentials for reconstructing the 3D trajectory of limb movement using multiple linear regression (mLR). Most MTP BCI studies report the best accuracy using low delta (0.5-2Hz) band-pass filtered EEG potentials. In a recent study, we showed spatiotemporalpower distribution of theta (4-8Hz), mu (8-12Hz), and beta (12-28Hz) EEG frequency bands contain richer information associated with movement trajectory. This finding is in line withthe results in the extensive literature on traditional sensorimotor rhythm (SMR) based multiclass (MC) BCI studies, which report the best accuracy of limb movement classification using power values of mu and beta frequency bands. Here, we show the reconstruction of actual and imagined 3D limb movement trajectory with an MTP BCI using a time-series of bandpower values (BTS model). Furthermore, we show the proposed BTS model outperforms the standard potential time-series model (PTS model). The BTS model yielded best results in the mu and beta bands (R~0.5 for actual and R~0.2 for imagined movement reconstruction) and not in the low delta band, as previously reported for MTP studies using the PTS model. Our results show for the first time how mu and beta activity can be used for decoding imagined 3D hand movement from EEG.