Quality of EEG in simultaneous EEG-fMRI for epilepsy Christian-G. Be ´nar, Yahya Aghakhani, Yunhua Wang, Aaron Izenberg, Abdullah Al-Asmi, Franc ¸ois Dubeau, Jean Gotman * Montreal Neurological Institute and Department of Neurology and Neurosurgery, McGill University, Montre ´al, Que ´bec, Canada Accepted 12 November 2002 Abstract It is now possible to record the EEG continuously during fMRI studies. This is a very promising methodology that combines knowledge about neuronal activity and its metabolic response. The EEG recorded inside the fMRI scanner is, however, heavily contaminated by artifacts caused by the high intensity magnetic field and rapidly changing field gradients. Methods have been reported in the literature to reduce or eliminate these artifacts, in particular the ballistocardiogram and the artifact caused by currents induced by rapidly changing magnetic gradients. Nevertheless, recording the EEG simultaneously with fMRI remains an extremely delicate operation. In addition the use of artifact removal methods has only been reported by the laboratories in which they were developed. We report here the practical procedures we developed to reduce artifacts in a series of 10 epileptic patients, in the context of the visualization of epileptic spikes. We illustrate the effectiveness of methods designed to remove the scanning artifact and present new methods for removing the ballistocardiographic artifact. We present and evaluate techniques to obtain an EEG of good quality when performing simultaneous EEG and fMRI studies. q 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: fMRI; EEG; Simultaneous recording; Epilepsy; Techniques 1. Introduction The possibility of recording the EEG inside an MR scanner was first reported by Ives et al. (1993), and later by Huang-Hellinger et al. (1995) and Lemieux et al. (1997). These reports established the safety of recording a good quality EEG inside the scanner, as well as the possibility to obtain high quality MR images despite the presence of EEG electrodes and equipment. These findings opened the way to functional MR imaging studies of spontaneous EEG events such as alpha activity or epileptic spikes. The MR scanner is a very difficult environment in which to record EEG. Any movement of the electrode wires inside the large static magnetic field or any variation of the field around the wires will induce currents that manifest as an artifact in the EEG. A common movement artifact is the ballistocardiogram, caused by the slight motion of the head that occurs after each heart beat (Ives et al., 1993; Allen et al., 1998). An example of a field variation artifact is the very large signal caused by gradient switching during fMRI image acquisition, which renders the EEG apparently non- interpretable. Subtraction techniques have been presented to eliminate the ballistocardiogram or reduce its effect on EEG quality (Allen et al., 1998; Goldman et al., 2000). Another approach in removing this artifact involves the application of a spatial filter optimizing the output signal to noise ratio (Bonmassar et al., 1999). The first method used to circumvent the gradient artifact relied on reading the undistorted EEG between image acquisitions. For example, the ‘spike-triggered’ fMRI paradigm takes advantage of the lagging nature of the blood oxygenation level dependent hemodynamic response. In this technique, one whole-head echo-planar imaging frame is acquired a few seconds after each epileptic spike (Warach et al., 1996; Seeck et al., 1998; Krakow et al., 1999). It has recently become clear, however, that it is possible to remove this large artifact and see the EEG during an EPI sequence, thus allowing a continuous fMRI recording with visualization of the EEG. The continuous fMRI recording allows the collection of several images for each spike, thereby increasing statistical power. It also made possible the plotting of the actual BOLD response to epileptic events (Lemieux et al., 2001; Be ´nar et al., in press). 1388-2457/02/$ - see front matter q 2002 Elsevier Science Ireland Ltd. All rights reserved. doi:10.1016/S1388-2457(02)00383-8 Clinical Neurophysiology 114 (2003) 569–580 www.elsevier.com/locate/Cinph * Corresponding author. Montreal Neurological Institute, 3801 University Street, Montreal, Quebec, Canada, H3A 2B4. Tel.: þ 1-514- 398-1953; fax: þ 1-514-398-8106. E-mail address: jean.gotman@mcgill.ca (J. Gotman). CLINPH 2001178