Wavelet-ICA based Denoising of Electroencephalogram Signal

Electroencephalogram (EEG) signals are having very small amplitudes and because of that they can be easily contaminated by different Artifacts. The presence of artifacts makes the analysis of EEG difficult for clinical evaluation. The major types of artifacts that affect the EEG are Power Line noise, eye movements, Electromyogram (EMG), and Electrocardiogram (ECG). Out of these artifacts Power Line noise is most prominent. To deal with these artifacts, there are numerous methods and techniques have been evolved by different researchers. In this paper, a new Independent Component Analysis (ICA) and Wavelet analysis based technique is presented based on joint use. This Wavelet-ICA combination is targeted to Single Channel EEG Signal. In the Algorithm, signal is decomposed into spectrally nonoverlapping components using wavelet decomposition. The ICA algorithm is then applied to derive the independent components. The wavelet-ICA components associated with artifactual event is selected and cancelled out. The artifact free wavelet components are reconstructed to form artifact free EEG. ICA is a multichannel technique. So it cannot be applied directly to Single channel EEG signal. Thus it needs a technique which can represent the single channel signal into virtual multichannel Signal. Stationary Wavelet Transform (SWT) is used to decompose the signal. The SWT decomposes single channel EEG signal into components based upon different frequency levels. The performance analysis of the algorithm is done using Signal to