Recognition of motor imagery EEG patterns based on common feature analysis

Motor imagery (MI) is particularly attractive in brain-computer interface (BCI) in the sense that it does not need any external stimuli. However, the overall performance is often severely affected by subject's mental states. In this study, a method based on common feature analysis (CFA) was proposed for MI electroencephalogram (EEG) patterns recognition, which can not only improve the recognition accuracy but also help to find reliable and interpretable features associated with specific MI patterns. Evaluation using several open competition datasets justifies that the common features could more accurately identify MI characteristics and hence substantially benefit MI EEG patterns recognition.