Wavelet Thresholding Using GWO Optimization for EEG Denoising

Electroencephalography (EEG) is one of the most frequently utilized methods in clinical neurology and neuro-engineering applications. However, the contaminated noises in EEG, such as artifacts caused by eye movements, muscular activities, and other factors like different technological sources that interfere with the signal at a specific scalp electrode position, may diminish the accuracy of this approach. As a result, EEG denoising is an inevitable step in EEG signal processing applications. This paper focuses on electrooculogram (EOG) signal cancellation from EEG signals. EOG signals are low-frequency signals that occur in the frequency range of 0-10Hz. Wavelet Thresholding is one of the efficient denoising methods for signals like EEG. Here, Wavelet Thresholding is used to eliminate EOG signals from the artifacted EEG signals. Generally, the threshold value for wavelet denoising is determined arbitrarily and the same for every level of decomposition. Therefore, most of the time, the elimination of artifacts is not accomplished effectively. In this methodology, the threshold value is calculated separately for each level of decomposition. The threshold value is determined separately and optimized using the Grey Wolf Optimization algorithm.