Review Acoustic Noise and Functional Magnetic Resonance Imaging: Current Strategies and Future Prospects Edson Amaro, Jr., MD, PhD, 1,2 * Steve C.R. Williams, PhD, 1 Sukhi S. Shergill, MD, PhD, 1 Cynthia H.Y. Fu, MD, 1 Mairead MacSweeney, PhD, 1 Marco M. Picchioni, MD, 1 Michael J. Brammer, PhD, 1 and Philip K. McGuire, MD, PhD 1 Functional magnetic resonance imaging (fMRI) has become the method of choice for studying the neural correlates of cognitive tasks. Nevertheless, the scanner produces acous- tic noise during the image acquisition process, which is a problem in the study of auditory pathway and language generally. The scanner acoustic noise not only produces activation in brain regions involved in auditory processing, but also interferes with the stimulus presentation. Several strategies can be used to address this problem, including modifications of hardware and software. Although reduc- tion of the source of the acoustic noise would be ideal, substantial hardware modifications to the current base of installed MRI systems would be required. Therefore, the most common strategy employed to minimize the problem involves software modifications. In this work we consider three main types of acquisitions: compressed, partially si- lent, and silent. For each implementation, paradigms using block and event-related designs are assessed. We also pro- vide new data, using a silent event-related (SER) design, which demonstrate higher blood oxygen level-dependent (BOLD) response to a simple auditory cue when compared to a conventional image acquisition. Key Words: task design; auditory cortex; acoustic noise; gradient noise; functional magnetic resonance imaging J. Magn. Reson. Imaging 2002;16:497–510. © 2002 Wiley-Liss, Inc. OVER THE PAST DECADE, magnetic resonance imag- ing (MRI) has superseded positron emission tomogra- phy (PET) as the technique of choice for mapping brain function. Functional MRI (fMRI) provides better tempo- ral and spatial resolution and does not involve the use of radioisotopes; therefore, the number of images ac- quired per subject is not restricted by the risk of radi- ation exposure (1). However, the process of image ac- quisition for fMRI generates a very loud acoustic noise (2–5) that can interfere with auditory processing (6 –10), and furthermore generates blood oxygen level-depen- dent (BOLD) contrast in the auditory cortex similar to that produced by vocalized words (11). Several methods of dealing with this problem have been described. Some focus on the source of the acoustic noise directly (12– 19), while others involve the insertion of silent gaps between acquisition periods (8,20 –23). There are al- ready many variations on the latter approach, depen- dent upon the questions and requirements posed. The aims of this work are to 1) review the literature regarding scanner acoustic noise, and 2) provide fur- ther, novel experimental data and a constructive cri- tique of these different acquisition approaches. NOISE SOURCE Scanner acoustic noise is derived from the mechanical oscillation of the gradient coils placed in a magnetic field (24,25). The Lorentzian forces induced when an electrical current is applied to these coils make them physically move. This displacement is dependent on the strength of the static magnetic field, the amplitude of the voltage applied, and the frequency and waveform of the switching (26). Echo-planar imaging (EPI) is currently the most widely used pulse sequence for acquiring dynamic physiological information, including fMRI data. It re- quires high-performance magnetic field gradients to al- low the collection of all necessary imaging data pertain- ing to one slice after a single excitation. EPI typically generates high-frequency noise, peaking at around 1 kHz. The amplitude of the acoustic noise typically var- ies from 94 to 135 dB SPL (specific pressure level) and depends upon the following parameters: the character- istics of the scanner materials and their assembly, the extent of vibration of the system, the type of pulse se- quence applied, the number of slices acquired, the re- ceiver bandwidth, the RF pulse envelope, the TR and TE, the FOV, the static magnetic field, and the rate and amplitude of gradient switching (13,16,27–29). Indus- trial guidelines in the U.K. and the U.S. stipulate that 1 Institute of Psychiatry, King’s College, University College, London, UK. 2 Institute of Radiology, Faculdade de Medicina, Universidade de Sa ˜o Paulo, Sa ˜o Paulo, Brazil. Contract grant sponsor: Wellcome Trust; Contract grant number: 061739/Z/00/Z. *Address reprint requests to: E.A., Department of Psychological Medi- cine, Institute of Psychiatry, University College, Box 067, DeCrespigny Park, Denmark Hill, SE5 8AF London, UK. E-mail: e.amaro@iop.kcl.ac.uk Received March 26, 2002; Accepted June 27, 2002. DOI 10.1002/jmri.10186 Published online in Wiley InterScience (www.interscience.wiley.com). JOURNAL OF MAGNETIC RESONANCE IMAGING 16:497–510 (2002) © 2002 Wiley-Liss, Inc. 497