A novel wavelet based technique for detection and de-noising of ocular artifact in normal and epileptic electroencephalogram

The electroencephalogram (EEG) is a biological signal that represents the electrical activity of the brain. A commonly encountered problem in clinical practice during EEG recording is ´blanking´ of the EEG signal due to blinking of the user´s eyes. Eye-blinks and movements of the eyeballs produce electrical signals that are collectively known as ocular artifacts and these are 10-100 times stronger than the EEG signal which is being recorded. The effective filtering of these ocular artifacts is extremely difficult because their frequency spread (1-50 Hz) overlaps with that of the EEG. Another major drawback of existing frequency based de-noising techniques is that they require continuous recording of electrooculargram (EOG) signals as well. Recently, a stationary wavelet transform (SWT) of the corrupted EEG signal has been used to de-noise it. We present a novel and simple technique for the detection, and subsequent de-noising, of these ocular artifacts using Haar wavelets of high orders. A comprehensive error analysis has been carried out, in both the time domain based artifact detection and the frequency domain based SWT de-noising of EEG. This procedure also has the advantage of being highly artifact selective, and so we have applied it to detect and de-noise epileptic EEG signals.