Ethical issues in the clinical application of fMRI: Factors affecting the validity and interpretation of activations John E. Desmond * and S.H. Annabel Chen Department of Radiology, Lucas MRS Center, MC: 5488, Stanford University, Stanford, CA 94305-5488, USA Accepted 5 August 2002 Abstract The ability of functional magnetic resonance imaging (fMRI) to localize activations in a singlepatient,alongwiththesafetyandwidespreadavailabilityofthismethodology,haslead to an increasing use of fMRI for clinical purposes such as pre-surgical planning. As meth- odology continues to improve and more experience with fMRI in the clinical setting is ac- quired, clinical functional neuroimaging will likely have an increasing influence over patient care.Therefore,ethicaluseoffMRI,aswithothermedicaltechniques,requiresunderstanding the factors impacting the interpretation of the methodology. Issues affecting the validity and interpretation of clinical functional neuroimaging, including effects of altered hemodynamic response function, head motion, and structural changes in the brain, are reviewed. The dis- tinctionbetweencorrelatedandnecessaryactivationinaclinicalcontextisdiscussed.Different typesofstatisticalerrorsinfMRIanalysisaredescribed,alongwiththeirconsequencestothe patient. Finally, for the future of clinical fMRI development, the need for normative patient data, as well as standardized tasks, scan protocols, and data analyses, is discussed. Ó 2002 Elsevier Science (USA). All rights reserved. Keywords: Clinical; Neuroimaging; fMRI; Methods; Hemodynamic; Validation; Statistics; Standardiza- tion 1. Introduction Functional MRI (fMRI) is a neuroimaging method that relies on regional levels ofoxygenatedbloodtoproducechangesinsignal,andthereforerequiresnoionizing radiation or other invasive procedures to image brain function. FMRI has spatial resolution,temporalresolution,andsignal-to-noiseratiothataresuperiortothatof positronemissiontomography(PET).Althoughelectroencephalography(EEG)and magnetoencephalography (MEG) have superior temporal resolution compared to fMRI—becausethesetechniquesaremoredirectlylinkedtotheelectricalactivityof neurons than is fMRI—neither EEG nor MEG are true 3D imaging modalities. Brain and Cognition 50 (2002) 482–497 www.academicpress.com * Corresponding author. Fax: +1-650-723-5795. E-mail address: jdesmond@stanford.edu (J.E. Desmond). 0278-2626/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII:S0278-2626(02)00531-6