Reduction of ocular artifacts from sleep EEG signals using the independent component analysis method

We present a statistical method using Independent Component Analysis (ICA) that is able to isolate the ocular artifacts from the electroencephalographic (EEG) data, separating EEG into their independent components. Some techniques, rejecting large portions of the recordings, can often cause loss of information and become impractical for many applications. After ICA is introduced it is shown how this technique can be used to remove ocular artifacts in EEG recording during sleep. Experiments on real EEG data from a sleep database show that ICA can effectively reduce ocular artifacts from even strongly contaminated EEG recordings. The method seems to be appropriate for other artifactual signals.