Optimal electroencephalogram signals denoising using hybrid β-hill climbing algorithm and wavelet transform

In this paper, hybridization between β-hill climbing algorithm and wavelet transform (WT) are proposed for Electro Encephalo Gram (EEG) signal denoising problem. EEG is a graphical measurement for the brain electrical activity which is recording from the scalp. It represents the voltage fluctuations resulting from ionic current flows within the neurons of the brain. During recording time, there are several artifacts noises can corrupt the original EEG signals such as eye blink, eye movements, muscles activity, and interference of power line. Therefore, the EEG signals should be processed to remove these noises obtaining the efficient EEG features. Several techniques have been proposed for EEG noises reduction in which one of these techniques is an EEG signal denoising using wavelet transforms (WT). Selecting wavelet parameters is a challenging task that is usually performed based on empirical evidence or experience. Therefore, β-hill climbing is proposed to find optimal wavelet parameters for EEG signal denoising that can obtain the minimum mean square error (MSE) between the original and denoised EEG signals. The proposed method was tested using a standard EEG dataset which is established by Kiern and Aunon. The proposed hybrid method was also evaluated using five criteria which are: Signal-to-Noise- Ration (SNR), SNR improvement, Mean Square Error (MSE), Root Mean Square Error (RMSE), and percentage root mean square difference (PRD). Finally, βHCWT compares with WT without βHC to show the effect of β-hill climbing on WT performance. The proposed method reveals an outstanding noise removal performance for non-stationary signals.