• Roman Guggenberger, Daniel Nanz, Lorenz Bussmann, Avneesh Chhabra, Michael A Fischer, Jürg Hodler, Christian W A Pfirrmann, and Gustav Andreisek.
• Department of Radiology, University Hospital, Raemistrasse 100, Zurich, 8091, Switzerland. roman.guggenberger@usz.ch
• Eur J Radiol. 2013 Oct 1; 82 (10): e590-6.

ObjectiveTo assess the agreement of fractional anisotropy (FA) and apparent diffusion coefficient (ADC) values of the median nerve on 3.0 T MR scanners from different vendors.Materials And MethodsIRB approved study including 16 healthy volunteers (9 women; mean age 30.6 ± 5.3 years). Diffusion tensor imaging (DTI) of the dominant wrist was performed on three 3.0 T MR scanners (GE, Siemens, Philips) using similar imaging protocols and vendor-proprietary hard- and software. Intra-, inter-reader and inter-vendor agreements were assessed.ResultsICCs for intra-/inter-reader agreements ranged from 0.843-0.970/0.846-0.956 for FA, and 0.840-0.940/0.726-0.929 for ADC, respectively. ANOVA analysis identified significant differences for FA/ADC measurements among vendors (p < 0.001/p < 0.01, respectively). Overall mean values for FA were 0.63 (SD ± 0.1) and 0.999 × 10(-3)mm(2)/s (SD ± 0.134 × 10(-3)) for ADC. A significant negative measurement bias was found for FA values from the GE scanner (-0.05 and -0.07) and for ADC values from the Siemens scanner (-0.053 and -0.063 × 10(-3)mm(2)/s) as compared to the remainder vendorsConclusionFA and ADC values of the median nerve obtained on different 3.0 T MR scanners differ significantly, but are in comparison to the standard deviation of absolute values small enough to not have an impact on larger group studies or when substantial diffusion changes can be expected. However, caution is warranted in an individual patient when interpreting diffusion values from different scanner acquisitions.Copyright © 2013 Elsevier Ireland Ltd. All rights reserved.

23 keywords...

hide…