A hybrid method incorporating new-grouped SSA with joint ICA and unsupervised clustering for removing multiple artifacts from single-channel EEG

Electroencephalogram (EEG) signals collected through ambulatory systems are frequently marred by a medley of disturbances, including electrooculogram (EOG), Motion Artifacts (MA), Electrical Shift and Linear Trend (ESLT), and Electromyography (EMG) artifacts. These artifacts considerably impede the precision of subsequent EEG analysis in practical applications. To date, various approaches have been devised, integrating decomposition methods and Blind Source Separation techniques, to address single or multiple artifacts. However, only a limited number of techniques have been developed for the simultaneous removal of low and high-frequency multiple artifacts from single-channel EEG recordings. It is worth noting that improperly denoised EEG signals can lead to misdiagnosis. In this work, we introduce a novel approach that leverages a new grouped Singular Spectrum Analysis (SSA) technique along with unsupervised k-means clustering controlled Blind Source Separation (BSS) to tackle the simultaneous removal of diverse artifacts from single-channel EEG data. Notably, our method operates without relying on statistical thresholds, thereby enhancing automation in the artifact removal process. The effectiveness of the proposed algorithm is validated using both synthesized and real-world EEG databases, and its performance is evaluated based on metrics such as <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><mml:mrow><mml:mo>Δ</mml:mo></mml:mrow></mml:math> SNR, <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><mml:mrow><mml:mi>η</mml:mi></mml:mrow><mml:mtext>,</mml:mtext></mml:math> and RRMSE.