Application of independent component analysis (ICA) in multichannel spectral estimation of hippocampal EEG

Presents the result of employing independent component analysis (ICA) in multichannel spectral estimation of hippocampal EEG from subfields at CA1 and the dentate gyrus during REM sleep. A fast ICA algorithm was employed to remove possible cross-talk in hippocampal EEG recording from CA1 and the dentate gyrus. Multichannel spectral analysis was then performed to examine the effects of ICA.. It was observed in power spectra that significant energy was removed from all frequency bands due to ICA. However, it was also noted that energy components in the θ frequency band are highly coherent between CA1 and the dentate gyrus with or without ICA processing. These findings inspire immediate questions regarding the actual mechanisms that lead to significant coherence in θ band between CA1 and the dentate gyrus. The preliminary result shows that ICA can be utilized to assist the underlying research to better address the θ correlation between these two hippocampal formations.