Quality advancement of EEG by wavelet denoising for biomedical analysis

Interests on the human body have never decreased and research on it has never stopped since hundreds years ago. A study of EEG for analysis of composition of the brain and cognitive processes for biomedical applications is ongoing topic for research. For proper diagnosis of many neurological diseases such as epilepsy, tumors, problems associated with trauma accurate analysis of EEG signals is essential. In addition, to enhance the efficacy of Brain Computer Interface (BCI) systems it is required to determine methods of increasing the signal-to-noise ratio (SNR) of the observed EEG signals. EEG measured by placing electrodes on scalp usually has very small amplitude in microvolts, so the analysis of EEG data and the extraction of information from this data is a difficult problem. This problem become more complicated by the introduction of artifacts such as line noise from the power grid, eye blinks, eye movements, heartbeat, breathing, and other muscle activity. Discrete wavelet transform offers an effective solution for denoising nonstationary EEG signals. In this paper, wavelet denoising is applied to EEG acquired during performing different mental tasks. First decomposition of the EEG signal from database using five different types of wavelets viz. Haar, Daubechies, Symlet, Coiflet,Dmey is carried out. In denoising process, the thresholding method used for removing noise from contaminated EEG. Our objective to find best suitable wavelet type to particular task which gave better performance measure such as larger signal-to-Noise Ratio (SNR). The EEG database from the Colorado state university is used for experimentation.