Cortico-Muscular Coupling Allows to Discriminate Different Types of Hand Movements

Cortico-muscular coupling (CMC) could be used as potential input of a novel hybrid Brain-Computer Interface (hBCI) for motor re-learning after stroke. Here, we aim of addressing the design of a hBCI able to classify different movement tasks taking into account the interplay between the cerebral and residual or recovered muscular activity involved in a given movement. Hence, we compared the performances of four classification methods based on CMC features to evaluate their ability in discriminating finger extension from grasping movements executed by 17 healthy subjects. We also explored how the variation in the dimensionality of the feature domain would influence the different classifier performances. Results showed that, regardless of the model, few CMC features (up to 10) allow for a successful classification of two different movements type. Moreover, support vector machine classifier with linear kernel showed the best trade-off between performances and system usability (few electrodes). Thus, these results suggest that a hBCI based on brain-muscular interplay holds the potential to enable more informed neural plasticity and functional motor recovery after stroke. Furthermore, this CMC-based BCI could also allow for a more "natural control" (l.e., that resembling physiological control) of prosthetic devices.