Metrological performance of a single-channel brain-computer interface based on motor imagery

In this paper, the accuracy in classifying Motor Imagery (MI) tasks for a Brain-Computer Interface (BCI) is analyzed. Electroencephalographic (EEG) signals were taken into account, notably by employing one channel per time. Four classes were to distinguish, i.e. imagining the movement of left hand, right hand, feet, or tongue. The dataset '2a' of BCI Competition IV (2008) was considered. Brain signals were processed by applying a short-time Fourier transform, a common spatial pattern filter for feature extraction, and a support vector machine for classification. With this work, the aim is to give a contribution to the development of wearable MI-based BCIs by relying on single channel EEG.