MR elastography of the liver: defining thresholds for detecting viscoelastic changes.

PURPOSE: To define thresholds for detecting significant change in liver viscoelasticity with magnetic resonance (MR) elastography, both for whole-liver measurements and for voxel-wise measurements in relation to spatial resolution.
MATERIALS AND METHODS: This prospective study was approved by the institutional review board, and all participants provided written informed consent. Thirty participants (16 volunteers and 14 patients with hepatitis B or C; 18 men; median age, 30.4 years; age range, 18.9-58.6 years) underwent imaging twice while in the same position (intraimage reproducibility), after repositioning (within-day reproducibility), and 1-4 weeks later (between-weeks reproducibility). MR elastography parameters comprised elasticity, viscosity, attenuation parameter α, and propagation parameter β. Bland-Altman analysis was used to calculate repeatability indexes for each parameter. Analyses were performed in a region-of-interest and a voxel-by-voxel level. Voxel-wise results were calculated in relation to spatial resolution by applying Gaussian filtering to establish the optimal trade-off point between resolution and reproducibility.
RESULTS: For elasticity, α, and β, within-day and between-weeks results were significantly lower than intraimage results (P ≤ .018 for all). Within-day and between-weeks results did not differ significantly. Over-time changes of more than 22.2% for elasticity, 26.3% for viscosity, 26.8% for α, and 10.1% for β represented thresholds for significant change. The optimal trade-off between spatial resolution and reproducibility was found at a filter size of 8-mm full width at half maximum (FWHM) for elasticity and propagation parameter β and at 16-mm FWHM for viscosity and attenuation parameter α.
CONCLUSION: Repositioning causes a significant decrease in the reproducibility of MR elastography. The propagation parameter β is the most reliable parameter, with an over-time threshold for significant change of 10.1% and the ability to reproduce viscoelasticity up to a resolution of 8-mm FWHM. Online supplemental material is available for this article. ©RSNA, 2013.