Ballistocardiogram artifact reduction in the simultaneous acquisition of auditory ERPS and fMRI

• Published in: NeuroImage (Orlando, Fla.) Fla.), 2004-08, Vol.22 (4), p.1534-1542
• Main Authors: Ellingson, M.L, Liebenthal, E, Spanaki, M.V, Prieto, T.E, Binder, J.R, Ropella, K.M
• Format: Article
• Language: English
• Subjects: Algorithms; Ballistocardiogram; Ballistocardiography; Brain; Electrocardiography; Electrodes; Electroencephalography; Event-related potential; Evoked Potentials, Auditory - physiology; Humans; Image Processing, Computer-Assisted; Magnetic Resonance Imaging; Scanners; Scholarships & fellowships; Signal Processing, Computer-Assisted
• ISSN: 1053-8119; 1095-9572
• DOI: 10.1016/j.neuroimage.2004.03.033

Electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) are now being combined to analyze brain function. Confounding the EEG signal acquired in the MR environment is a ballistocardiogram artifact (BA), which is predominantly caused by cardiac-related body movement. The objective of this study was to develop and evaluate a method for reducing these MR-induced artifacts to retrieve small auditory event-related potentials (ERPs) from EEG recorded during fMRI. An algorithm for BA reduction was developed that relies on timing information obtained from simultaneous electrocardiogram (ECG) recordings and subsequent creation of an adaptive BA template. The BA template is formed by median-filtering 10 consecutive BA events in the EEG signal. The continuously updated template is then subtracted from each BA in the EEG. The auditory ERPs are obtained through signal averaging of the remaining EEG signal. Experimental and simulated ERP data were estimated to assess effectiveness of the BA reduction. Simulation showed that the algorithm reduced BA without significantly altering the morphology of a signal periodically inserted in the EEG. Auditory ERP data, obtained in a 1.5-T scanner during a passive auditory oddball paradigm and processed with the BA reduction algorithm, were comparable to data recorded in a mock scanner outside the magnetic field with the same experimental paradigm. It is concluded that through adequate reduction of the BA, relatively small auditory ERPs can be acquired in the MR environment.