Eeg Artifact Removal and Noise Suppression Using Hybrid Glct -Ica Technique

Abstract

Electroencephalogram signals are often contaminated by non-cerebral sources like muscle artifacts, eye movement and instrumentation noise due to which cerebral information loss occurs and interpretation of signals become challenging. This paper presents a novel noise suppression and artifact removal technique for Electroencephalogram signal records. The proposed hybrid technique is based on joint usage of Fast-Power ICA and General Linear Chirplet Transform. In present work, to separate blind sources of contaminated Electroencephalogram activity Fast-Power ICA technique is employed. Further, Artifactual Independent Components are identified and corrected by GLCT transformation technique. The efficacy of proposed work is estimated on simulated Electroencephalogram signals by qualitative evaluation. The results demonstrate that proposed artifact and noise suppression technique is capable of identifying non-cerebral sources of artifact present in Electroencephalogram activity. Also, it effectively removes such sources from recorded Electroencephalogram activity and makes signals contamination free.