Removal of ocular artifacts in EEG signals using adapted wavelet and adaptive filtering

Abstract

EEG is the recording of electrical activity along the scalp. The raw EEG signals generally contain a large amount of unwanted artifacts. The movement of eyeballs and eye blinks introduce artifacts, also known as electrooculogram (EOG) signal. In this study, the EEG signals were recorded from 24 subjects while they were performing a particular task in a noise free environment. For the removal of these ocular artifacts (OAs), pattern matching using a continuous adapted wavelet followed by adaptive filtering technique has been proposed to obtain a refined EEG signal. This new approach is able to identify multiple eye blinks at different instants within the specified interval of time. An existing method has also been applied to all the recorded signals. The proposed method is compared with this existing approach. The performance evaluation of these methods is done using Mean Square Error (MSE) and Signal-to-Noise ratio (SNR). This showed that the average SNR of the proposed method is 48.43% higher than the existing method.