LAUSR

Abstract In this paper, power suppression detection in mu band known as event-related desynchronization during a simultaneous EEG–fMRI recording is addressed. Simultaneous EEG-fMRI experiments provide great opportunities to recognize relation between the active areas in the brain, discovered by fMRI, and EEG events. However, decoding EEG is challenging as a result of artifacts which MRI scanner applies on EEG. Ballistocardiogram (BCG) is one of the main destructive artifacts which deteriorates EEG rhythms specially in mu band. The proposed method, is a supervised time-frequency decomposition algorithm which estimates underlying rhythms of EEG signals recorded in MRI scanner. In the proposed method, a multi layer decomposition approach using three criteria is developed to extract brain oscillations. This goal is achieved by eliminating the artifact related components at each decomposition layer as a result of removing useless extracted components in each layer. Performance of the proposed method is evaluated using synthetic and real EEG. The achieved results from synthetic data confirm the ability of the proposed method to extract mu rhythms when different levels of noise exist. In addition, proposed algorithm is applied on a set of real EEG acquired in a simultaneous EEG–fMRI recording. Estimated ERD confirms the superiority of the developed algorithm in terms of estimating changes of mu rhythms. Moreover, an EEG–fMRI integration is performed to explore the correlation between blood oxygenation level dependent (BOLD) in fMRI and ERD of mu rhythm in EEG.