Sparse spectrum enabled swarm decomposition for brain signal processing

Abstract

Brain signal analysis has a crucial role in the investigation of neuronal activity for diagnosis of neurological diseases and brain-computer interfaces (BCI) applications. The electroencephalogram (EEG) and magnetoencephalogram (MEG) are the most efficient brain signals that assist in the diagnosis of neurological diseases and also play an essential role in all neurosurgery related to the brain. Furthermore, it is a valuable tool for clinicians and researchers and has the potential to develop advanced BCIs and help in partial paralysis rehabilitation. The clinicians visually study these brain signal recordings to determine the manifestation of abnormalities in the brain. Hence adequate and accurate brain signal analysis of these signals is always a major challenge for researchers. The complex and nonstationary nature of EEG and MEG signals demands advanced signal analysis techniques for proper interpretation and extraction of brain information. In this perspective, adaptive nonstationary signal-processing algorithms like swarm decomposition (SWD) and iterative filtering provide useful signal representation for EEG and MEG, enabling easy interpretation and feature extraction.