Feature extraction method for EEG during motor imagery of limbs

抄録

Recently, many researchers have been studying Brain Computer Interface (BCI) by which we can operate objects with analyzing Electroencephalogram (EEG) instead of hands. One of them, by using event related synchronization from EEG during motor imagery of each hand, discriminated right or left hand's movement [1]. However, these methods enable us to discriminate only a few classes. Furthermore, they need a long-term training for improving their discrimination accuracy. In this paper, we construct a system for discriminating motor imaged limbs using Movement-Related Cortical Potentials (MRCPs), such as BereitschaftsPotential (BP), Negative Slope (NS) and Contingent Negative Variation (CNV), in which it does not need a long-term training.