Meeting Banner
Abstract #2873

Reproducibility of Brain Morphometry Results Derived at 3T: A Multi-Center European Study Comparing the Longitudinal and Cross-Sectional Freesurfer Segmentation Analyses

Jorge Jovicich1, Moira Marizzoni2, Roser Sala-Llonch3, Nuria Bargall3, David Bartrs-Faz3, Jennifer Arnold4, Jens Benninghoff4, Jens Wiltfang4, Luca Roccatagliata5, Flavio Nobili5, Christian Zeeh6, Peter Schonknecht6, Giada Zoccatelli7, Franco Alessandrini7, Alberto Beltramello7, Hlne Gros-Dagnac8, Pierre Payoux8, Valrie Chanoine9, 10, Jean-Philippe Ranjeva9, 10, Mira Didic9, 10, Melanie Leroy11, Regis Bordet11, Giovanni Frisoni2

1University of Trento, Center for Mind/Brain Sciences, Mattarello, Trento, Italy; 2IRCCS San Giovanni di Dio Fatebenefratelli, LENITEM Lab of Epidem., Neuroim. & Telem, Brescia, Lombardia, Italy; 3Dept. Psychiatry and Clinical Psychobiology, Faculty of Medicine University of Barcelona, Barcelona, Spain; 4Department of Psychiatry and nuclear medicine, Universitaet Duisburg-Essen, Essen, Germany; 5Dept of Neuroscience, Ophthalmology and Genetics, University of Genoa, Genoa, Lombardia, Italy; 6Dept of Psychiatry and Dept of Neuroradiology, University of Leipzig, Leipzig, Germany; 7Dept of Neuroradiology, Verona General Hospital, Verona, Italy; 8U825 - Plateau Technique IRM, INSERM / Universit Paul Sabatier, Toulouse, France; 9Hpital La Timone CIC UPCET, Marseille, France; 10Centre de Resonance Magnetique Biologique et Medicale , Aix Marseille Universit , Marseille, France; 11Universit Lille UL2, Lille, France


The success in finding clinically useful MRI-derived biomarkers is highly dependent on data acquisition and analysis strategies. In this brain morphometry study we show for the first time the across-session test-retest reproducibility advantages of the fully automated longitudinal FreeSurfer segmentation protocol relative to the cross-sectional analysis, when tested in a consortia using different 3T MRI scanners (Siemens, Philips, GE) acquiring standard 3D MPRAGE data, with 7 out of 8 sites using parallel imaging acquisition (about 5 min acquisition per volume) and no data averaging.