Abstract
In this study, the right and left index fingers of the subjects were constantly stimulated by a mechanical vibration device to examine the possibility of an asynchronous brain machine interface (BMI) based on steady-state somatosensory evoked potentials (SSSEPs). In the experiments, the subjects were instructed to attend on the finger indicated by the visual que. EEGs were recorded from 61 electrodes on the entire head. First, SSSEPs in the range of 10 to 40 were measured and evaluation vectors were obtained. Secondly, narrow band pass filter, principal component analysis (PCA) and in dependent component analysis (ICA) were applied to obtain independent components (ICs). Lastly, the IC reflecting SSSEP was automatically selected based on the evaluation vector. The reconstructed time courses of the IC tended to be increased when subjects attended on the indicated finger. These results demonstrate that the presented method was useful for detecting the SSSEPs and an asynchronous BMI based on SSSEP may be promising.
