EEG Quality: Origin and Reduction of the EEG Cardiac-Related Artefact

Abstract

With the advent of purpose-built “MR-compatible” EEG recording hardware, the simultaneous acquisition of EEG and fMRI has recently become more widespread (for reviews, see Herrmann and Debener 2007; Laufs et al. 2008). Nevertheless, the MRI scanner remains a hostile environment for EEG recordings, and ensuring good EEG signal quality can be a challenging task (e.g. Parkes et al. 2006). The level of EEG data quality that can be achieved from simultaneous recordings is a matter of ongoing investigation, but a common view is that a certain loss of quality is unavoidable and must be tolerated (Debener et al. 2007b). Nonetheless, combined EEG and fMRI data acquisitions in a single session are an attractive alternative to separate acquisitions in some circumstances, as discussed throughout this book (Babiloni et al. 2004; Debener et al. 2006; Horwitz and Poeppel 2002). The reduction of artefacts that contaminate the EEG signal as much as possible is necessary to make full use of the potential of EEG–fMRI. The MRI environment is known to introduce several different types of EEG artefact, among them the gradient artefact and the cardiac pulse-related (often referred to as ballistocardiogram or BCG) artefact. This chapter focuses on the pulse artefact: its origin, characteristics of it, and methods of reducing or eliminating it. After introducing the conceptual and statistical characteristics that define the pulse artefact, we discuss the mechanisms that give rise to the pulse artefact and present a two-factor pulse artefact model. We will review and compare different ways of removing it. We will focus on several recording and analysis details that have been barely acknowledged in the literature and can have a significant impact on the quality of the pulse artefact correction step, and thus on the final EEG data quality.