Ocular Artifact Removal from EEG Using ANFIS

Electroencephalogram (EEG) signals are often contaminated with various artifacts, especially electrooculogram (EOG) or ocular artifacts that cannot be avoided consciously and largely degrade the clinical interpretation of the signals. This paper presents a study on adaptive noise cancellation (ANC) based on adaptive neuro-fuzzy inference system (ANFIS) for EOG artifacts removal, especially when time delay is significant and on real contaminated EEG signal. The performance is first evaluated using simulated EEG and EOG signals, further investigation on the effect of time delay and tests on real data are also performed. The results illustrate that ANFIS provides a promising approach to ocular artifact removal with the best performance in comparison with ANC using adaptive filtering and ADALINE.