COMMENTS AND CONTROVERSIES Functional MRI of the Human Amygdala? Klaus-Dietmar Merboldt, Peter Fransson,* Harald Bruhn, and Jens Frahm 1 Biomedizinische NMR Forschungs GmbH am Max-Planck-Institut fu ¨ r biophysikalische Chemie, 37070 Go ¨ttingen, Germany; and *MR Research Center, Karolinska Institutet/Hospital, Stockholm, Sweden Received November 10, 2000; published online May 30, 2001 In view of an increasing number of publications that deal with functional mapping of the human amygdala using blood oxygenation-level-dependent (BOLD) magnetic resonance imaging, we reevaluated the un- derlying image quality of T2*-weighted echoplanar im- aging (EPI) and fast low angle shot (FLASH) sequences at 2.0-T with regard to susceptibility-induced signal losses and geometric distortions. Apart from the tim- ing of the gradient echoes, the degree of susceptibility influences is controlled by the image voxel size. Whereas published amygdala studies report voxel sizes ranging from 22 to 125 l, the present results suggest that reliable imaging of the amygdala with BOLD sensitivity requires voxel sizes of 4 to 8 l or less. Preferentially, acquisitions should be performed with a coronal section orientation. Although high-res- olution BOLD MRI is at the expense of temporal reso- lution and volume coverage, it seems to provide the only solution to this physical problem. © 2001 Academic Press INTRODUCTION Many of the early magnetic resonance functional neuroimaging studies focused on the mapping of pri- mary sensory systems. Prominent reasons included the need for a validation of this new modality and the desire to detail previous neuroscientific insights ob- tained by positron emission tomography in terms of spatiotemporal resolution or paradigm specificity. In addition, simple extensions of the functional MRI (fMRI) approach to higher cognitive processing often failed because of the presence of pronounced magnetic field inhomogeneities in brain regions at focus. For example, this holds true for the anterior portion of the hippocampal formation, relating to studies of memory and learning, and particularly the amygdala, relating to functional mapping of brain responses to stimuli with emotional contents. The purpose of this comment is a reanalysis of our ability to image the human amygdala with blood oxy- genation-level-dependent (BOLD) MRI sensitivity, i.e., by using T2*-weighted gradient-echo sequences. Its primary aim is to put up a warning sign for paying sufficient attention to technical confounds that need to be addressed before the scientific community should trust the colored spots of alleged activations. The data presented here should help to further the tremendous potential of fMRI in both basic and clinical neuro- science by drawing a clear line between sound results and dubious material. THE SUSCEPTIBILITY PROBLEM We specifically draw attention to the unavoidable presence of susceptibility-induced magnetic field inho- mogeneities in the vicinity of the human amygdala. The underlying physical reason is the neighborhood of the air-filled bony cavities at the skull base, which have very different magnetic susceptibilities than brain tissue. In the static magnetic field of an MRI system these structures therefore generate magnetic field gradients in extended transition zones. The re- sulting spatially dependent frequency distribution of the MRI signal leads to a partial dephasing or even a complete randomization of the individual magnetiza- tion components. Integration of the intravoxel compo- nents during data acquisition yields a corresponding signal loss within affected voxels. Despite this fundamental physical problem, a sur- prisingly large number of publications claim successful mapping of emotional processing in the amygdala, for example, see Breiter et al. (1996), Irwin et al. (1996), Schneider et al. (1997), Birbaumer et al. (1998), Bu ¨ chel et al. (1998), LaBar et al. (1998), Phillips et al. (1998), Whalen et al. (1998), Baron-Cohen et al. (1999), Phil- lips et al. (1999), Schneider et al. (1999), Critchley et al. (2000), Rauch et al. (2000), Schneider et al. (2000), and 1 To whom correspondence and reprint requests should be ad- dressed. Fax: +49-551-201-1307. E-mail: jfrahm@gwdg.de. NeuroImage 14, 253–257 (2001) doi:10.1006/nimg.2001.0802, available online at http://www.idealibrary.com on 253 1053-8119/01 $35.00 Copyright © 2001 by Academic Press All rights of reproduction in any form reserved.