Eye-Blink Artifact Removal from EEG Signal using Machine Learning and De-noising Techniques

EEG (Electroencephalogram) is the most extensively used non-invasive system to record brain signal for brain computer interface but it is vulnerable to different type of artifacts. In which, eye-blink artifact can be more prominent in some cases, which needs to be removed for better performance of BCI system. At first, contaminated EEG signal is identified using SVM classifier and k-means clustering. Then, A collection of de-noising techniques consisting Independent Component Analysis (ICA), Discrete Wavelet Transform (DWT) and Empirical Mode Decomposition (EMD) have been applied to remove eye-blink artifact from EEG signal. Different statistical measure which includes Root Mean Square Error (RMSE), Signal to Artifact Ratio (SAR) and Correlation coefficient (CC) are used to compare performance of each de-noising technique.