Task-irrelevant alpha component analysis in motor imagery based brain computer interface

In motor imagery based BCI, the alpha rhythm shares the same frequency band with sensorimotor rhythm (SMR), and does not correlate with mental task, which contaminates the SMR recording. Independent component analysis (ICA) was applied to decompose original EEG signal into source components, and a comprehensive method was proposed to discriminate those source components by combining temporal, frequency, spatial, and class label information. Task-irrelevant alpha components were sorted out and their projections were reduced by proper bipolar electrode placement for improving the BCI performance.