Many studies have shown over the past two decades that
people can communicate by using the scalp recorded
electroencephalogram (EEG) activity with no or very little
voluntary movement. This is commonly called the EEG
based brain-computer interface (BCI) systems. The BCI
systems measure certain features of EEG activity, and use
them to generate the control signal. Some systems use
potentials evoked by stereotyped stimuli [1]. Other systems
use EEG features in the frequency domain without specific
sensory events [2], [3]. This motor imagery based BCI uses
the sensorimotor rhythms (SMRs), including Mu (8-14Hz)
and/or Beta (15-30Hz) rhythms, recorded from the scalp
over the sensorimotor cortex. When subjects imagine the
left or right hand movement, a distinct feature, such as the
amplitude attenuation of Mu rhythm and an event related
desynchronisation (ERD) in the EEG signaling activity,
appears over the contralateral hand area at the sensorimotor
cortex [4].