Real Time Eye Blink Artifacts Removal in Electroencephalogram Using Savitzky-Golay Referenced Adaptive Filtering

Abstract

Eye blink artifacts cause a major problem to electroencephalograph (EEG) signals since they introduce serious distortion to the signals. The implementation of algorithm to eliminate those artifacts automatically and at real time becomes highly important when considering the recent focus on portable EEG applications. In this paper, we propose a real-time eye blinks artifact removal using adaptive filtering without an additional reference electrode. The proposed method utilizes a Savitzky-Golay (SG) filter to estimate a blink component from the noisy EEG signals, and then is used as a reference in adaptive noise cancellation system. We demonstrate the reliability of the proposed method by using both simulated and real EEG datasets. The estimated blink artifacts by the proposed method are compared to the original blink signals measured by electrooculograph (EOG). The performance of the filters in terms of removing the blink components from the noisy signals are also compared between the proposed SG referenced and a conventional EOG referenced adaptive filtering. The results show that the proposed method is able to estimate the blink artifacts with high correlation to the original blink signals and remove the artifact successfully.