Efficient de-noising technique for electroencephalogram signal processing

Abstract

An electroencephalogram (EEG) is a recording of various frequencies of electrical activity in the brain. EEG signal is very useful for diagnosis of various brain related diseases at early stage to prevent severe issues which may lead to loss of life. The raw EEG signal captured through the leads contain different type of noises which is not susceptible for diagnosis. In this paper, an efficient algorithm is proposed to process the raw EEG signal to combat the noise. To obtain noiseless EEG data, the likelihood test ratio is applied to interference computation block. The likelihood ratio test converts EEG data signal into segmented data with nearly constant noise characteristics. This will aid in detecting the noise present in a tiny segment which ensures proper signal denoising. The processed signal is compared with the database of noiseless EEG of the same person using principal component analysis (PCA) classifier. The proposed algorithm is 99.01% efficient to identify and combat noise in the EEG signal.