EMD-BSS: A hybrid methodology combining Empirical Mode Decomposition and Blind Source Separation to eliminate the ocular artifacts from EEG recordings

In Electroencephalogram (EEG) research, physiological artifacts like muscle activity, heart rhythm, and eye movements or blinks continue to be a prominent issue. It is essential to deal with these artifacts without affecting the underlying neuronal information. Up until now, several studies have used a number of different signal-processing techniques in an attempt to achieve the optimal artifact rejection outcome. In this study, we propose a hybrid approach that combines Empirical Mode Decomposition (EMD) with five different Blind Source Separation (BSS) algorithms in an attempt to remove the ocular artifacts. We evaluate our method using four commonly used assessment features, namely the Spearman Correlation Coefficient (SCC), the Euclidean distance (ED), the Root Mean Square Error (RMSE), and the Signal-to-Artifact Ratio (SAR) of the clean reconstructed signal. The aim of this study is to generate an enhanced artifact removal methodology and to compare the performance of the BSS algorithms after combining them with the EMD method. The results demonstrate that the herein presented approach is more effective for ocular artifact rejection compared to solely applying a single Blind Source Separation (BSS) algorithm, and appoint the EMD-AMICA algorithm as the optimally performing technique in the context of the hybrid methodology (SCC = 0.95, RMSE = 9.51 ED = 736.7, and SAR = 1.92).