Karl Gerard Helmer1, Ming-Chung Chou2, Allen Song3, Jessica Turner4, Barjor Gimi5, Susumu Mori6
1Radiology, Massachusetts General Hospital, Charlestown , MA, United States; 2Computer Science & Engineering, National Sun Yat-sen University, Kaohsiung, Taiwan; 3Duke University, Durham, NC, United States; 4The Mind Research Network, Albuquerque, NM, United States; 5Radiology, Dartmouth Medical School, Hanover, NH, United States; 6Radiology, School of Medicine, Johns Hopkins Univerity, Baltimore, MD, United States
Multi-site studies have become ubiquitous when large numbers of subjects are needed. Therefore, the characterization of site and protocol variation effects has become increasingly important. We present data from Siemens, Philips, and GE MR scanners and characterize the behavior of the fractional anisotropy as decreasing amounts of data and/or diffusion-weighted directions are used in the tensor calculation. In addition, the effect of a common protocol variation (TE) is investigated. The methods used in this study are also suitable for the site-reliability characterization before a multi-site study is begun or after upgrades during the study.