ORIGINAL RESEARCH Measurement of Magnetization Transfer Ratio (MTR) From Cervical Spinal Cord: Multicenter Reproducibility and Variability Benoit Combès, PhD, 1 * Laureline Monteau, MD, 1,2 Elise Bannier, PhD, 1,2 Virginie Callot, PhD, 3,4 Pierre Labauge, MD, PhD, 5 Xavier Ayrignac, MD, 5 Clarisse Carra Dallière, MD, 5 Jean Pelletier, MD, 3,6 Adil Maarouf, MD, 3,6 Jerome de Seze, MD, PhD, 7 Nicolas Collongues, MD, PhD, 7 Christian Barillot, PhD, 1 Gilles Edan, MD, 1,8 Jean Christophe Ferré, MD, 1,2 and Anne Kerbrat, MD, 1,8 and the EMISEP study group Background: Assessing the multicenter variability of magnetization transfer ratio (MTR) measurements in the spinal cord of healthy controls is the first step toward investigating its clinical use as a biomarker. Purpose: To analyze the between-session, between-participant, and between-scanner variability of MTR measurements in automatically extracted regions of interest in the cervical cord of healthy controls. Study Type: Control study. Population: Forty-four participants, distributed across five MRI scanners (all from the same manufacturer). Ten participants were scanned twice in the same scanner, and 10 others were scanned twice in two different scanners. Field Strength/Sequence: 3D-gradient echo images, centered on C5, without and with magnetization transfer prepulse at 3T. Assessment: We calculated the mean MTR for different vertebral levels in the whole cord (WC), as well as in the white matter and gray matter, and determined the between-session, between-participant, and between-scanner variabilities. Statistical Tests: Coefficients of variation and intraclass correlations (ICCs) for the different variabilities and their associated confidence intervals. Results: The MTR measurements for Levels C4-C6 (near the slab center) exhibited a mean value in WC of 34.6 pu and a pooled standard deviation of 0.9 pu. The between-session coefficient of variation was estimated as 2.3% (ICC = 0.63), the between-participant coefficient as 1.6% (ICC = 0.32), and the between-scanner coefficient as 0.7% (ICC = 0.05). The resulting aggregate coefficient of variation was 2.9%, which was sufficiently low to detect an MTR reduction of 1 pu between groups of about 45 participants (Type-I error rate: 0.05; Type-II error rate: 0.10). Data Conclusion: The good between-scanner reproducibility and low overall variability in cervical spinal cord MTR mea- surements in a control population might pave the way for multicenter analyses in various neurological diseases with mod- erate cohort sizes. Level of Evidence: 2 Technical Efficacy: Stage 2 J. MAGN. RESON. IMAGING 2018. S pinal cord pathology is a major cause of disability in a range of neurological diseases, including multiple sclerosis (MS), cervical spondylotic myelopathy, amyotrophic lateral sclerosis, and acute spinal cord injury. 1 T 1 -weighted and T 2 - weighted magnetic resonance imaging (MRI) are the tools of choice for diagnosing cord abnormalities. However, these View this article online at wileyonlinelibrary.com. DOI: 10.1002/jmri.26537 Received Mar 16, 2018, Accepted for publication Sep 20, 2018. * Address reprint requests to: B.C., IRISA - Campus de Beaulieu, 263 avenue du Général Leclerc, 35042 Rennes Cedex, France. E-mail: benoit.combes@inria.fr The first two authors contributed equally to this work. From the 1 Univ Rennes, Inria, CNRS, Inserm, IRISA UMR 6074, Visages, U1128, France; 2 CHU Rennes, Radiology Department, F-35033, Rennes, France; 3 AP-HM, Pôle d’imagerie médicale, Hôpital de la Timone, CEMEREM, Marseille, France; 4 Aix-Marseille Université, CNRS, UMR 7339, CRMBM, Marseille, France; 5 Montpellier University Hospital, France; 6 AP-HM, CHU Timone, Pole de Neurosciences Cliniques, Department of Neurology, Marseille, France; 7 Strasbourg University Hospital, France; CIC Strasbourg INSERM 1434, Strasbourg, France; and 8 Neurology Department, Rennes University Hospital, France Additional supporting information may be found in the online version of this article. © 2018 International Society for Magnetic Resonance in Medicine 1