Automatic Electroencephalogram Channel Selection of Independent Component Analysis Based Motor Imagery Brain-computer Interface

Independent component analysis (ICA) is widely used to separate movement-related independent components (MRICs) from multi-channel motor imagery electroencephalogram (EEG) signals. The distribution of channels is a crucial question in the design of brain-computer interface using ICA algorithm. The traditional methods on channel selection mainly use identical channels which close to the cortex area for all subjects. However, the optimal channels are subject-dependent. It is crucial to determine which channel combination should be used to reach optimal performance. In this study, based on our previous research related to the ICA-based motor imagery BCI (MIBCI) algorithm frame, we developed a new strategy to select optimal channel combination. The experiment was based on 24 runs three class motor imagery EEG datasets, in addition, the results of common spatial pattern (CSP) were compared with our channel optimization algorithm too. Experimental results revealed that higher stability and performance can be obtained after optimizing channel distribution by employed our novel channel selection strategy.