Elimination of ECG Artifacts from a Single-Channel EEG Using Sparse Derivative Method

The electroencephalogram (EEG) signal has been widely studied in a brain computer interface (BCI) or a polysomnography (PSG) study. However, the EEG can be easily contaminated by electrocardiogram (ECG) artifacts due to high electrical energy of ECG. These artifacts obstruct the analysis of EEG signal. This paper proposes sparse derivative method for eliminating ECG artifacts from a single-channel EEG. In order to evaluate the performance of our proposed method, we compared the proposed method with the combined ensemble empirical mode decomposition and independent component analysis (EEMD+ICA) method by simulation. In addition, we applied the sparse derivative method to real ECG contaminated EEG datasets. Our proposed method could not only estimate and remove ECG artifacts, but also preserve spectral contents of the original EEG. The results of simulation and real data indicate that our proposed method performs better than the EEMD+ICA method.