Prediction of missing EEG channel waveform using LSTM

The postsynaptic membrane potential is captured on the surface of the brain's scalp with the help of EEG machine. Various data points over the scalp region generate different EEG channel signal profiles. A huge amount of research studies have already been invested into analysing those EEG channel information and are trying to churn out hidden discrepancies between normal and diseased subjects' EEG waveforms. But till now little has been accomplished to reconstruct any lost EEG channel information from the remaining ones. This research work concentrates on the prediction of an EEG channel from any other EEG channel using deep network - LSTM on a publicly available BNCI EEG database. The waveforms for various missing EEG recordings are predicted with high accuracy with a simple lightweight technology.