Ica Based Artifacts Removal from Eeg Signal Using High Resolution Double Density Wavelet Structure

Abstract

The physiological signal as Electroencephalography (EEG) can be contaminated by artifacts during acquisition process which will obstruct the features and quality of interest in the signal. The health-care diagnostic procedures need strong and viable biomedical signals, hence elimination of artifacts from EEG is important. In this report, an improved ensemble approach is proposed for single channel EEG signal motion artifacts removal. The single channel EEG signal is preferred in order to reduce the system complexity. newlineThe EEG signal is processed with Ensemble Empirical Mode Decomposition (EEMD) to convert signal channel signal into the multi-channel signals. These multi-channel signals are provided to BSS-CCA (Blind Source Separation-Canonical Correlation Analysis) approach for separation of signal and artifact sources. The motion artifacts randomness are still present in the EEG signal even after this two-stage filtering approach (EEMD-CCA). Therefore, Stationary Wavelet Transform (SWT) is applied to the EEMD-CCA processed EEG signal. This filtering technique is tested against currently available artifact removal techniques. The obtained results indicate that the proposed technique is robust enough to efficiently process the EEG signal. Moreover, this technique is capable enough to remove the motion artifacts effectively without diminishing the EEG neural information. The performance of the proposed artifact removal technique is evaluated both qualitatively and quantitatively. newlineThe quantitative parametric analysis is done by using efficiency matrices such as Del Signal to Noise Ratio (DSNR), Lambda (and#955;), Spectral Distortion (Pdis) and Root Mean Square Error (RMSE). Comparison of the proposed method with state of the art artifact removal methods indicate that the proposed algorithm is suitable for use as a supplement to algorithms currently in use. The result offers significant improvements in parameters like DSNR by 91%, lambda by 27.87%, PSD by 88.33% and reduction in RMSE parameter by 31.96% which demonstrates th