Electroencephalogram (EEG) is a widely used signal for investigating the activities of brain. It is extensively being utilized for the diagnosis of different disorders of central nervous system such as Alzheimer�s, parkinson�s, seizures, epilepsy, and so forth. Ocular activity creates significant artifacts in electroencephalogram recordings. Analysis of the EEG and obtaining clinical information is being difficult because of these noise sources. This paper proposes wavelet-based denoising method with new statistical thresholding for single channel EEG signal. Two commonly used WT strategies, Discrete Wavelet Transform (DWT) and Stationary Wavelet Transform (SWT), were applied. Five WT basis functions, to be specific, daubechies, coiflet, symlet, bior and rbio were considered for artifact removal utilizing universal threshold (UT), statistical threshold (ST) and proposed thresholding (PT) techniques. This method is evaluated on EEG signals taken from polysomnographic records, eegmmidb database. The adequacy of the proposed threshold is quantitatively measured using parameters such as signal to noise ratio (SNR), Artifact rejection ratio (ARR) and contrasted with the existing thresholds. Consequence of this study demonstrated that DWT+PT+coif5 and SWT+PT+db5 combination removes the artifacts successfully from single channel EEG signals.

Username
Password