Development of a Brain-Computer Interface system based on sensorimotor rhythms

In this work we developed a Brain-Computer Interface system using brain signals from somatosensory cortex. As stimulus, arrows were shown to the right and left to five subjects, who perform imaginary hand movement tasks to the side indicated by the arrow. Subject-specific parameters were selected using DSLVQ (Distinctive Sensitive Learning Vector Quantization) and Lateralization Index (LI), in order to extract the most relevant features, resulting in a higher hit rate in the classification stage. Online classification was performed with Linear Discriminant Analysis resulting in hit rates between 81 and 92.1%.