Performance of different Dixon-based methods for MR liver iron assessment in comparison to a biopsy-validated R2* relaxometry method.

Abstract

To prospectively evaluate a 3D-multiecho-Dixon sequence with inline calculation of proton density fat fraction (PDFF) and R2* (qDixon), and an improved version of it (qDixon-WIP), for the MR-quantification of hepatic iron in a clinical setting. Patients with increased serum ferritin underwent 1.5-T MRI of the liver for the evaluation of hepatic iron overload. The imaging protocol for R2* quantification included as follows: (1) a validated, 2D multigradient-echo sequence (initial TE 0.99 ms, R2*-ME-GRE), (2) a 3D-multiecho-Dixon sequence with inline calculation of PDFF and R2* (initial TE 2.38 ms, R2*-qDixon), and optionally (3) a prototype (works-in-progress, WIP) version of the latter (initial TE 1.04 ms, R2*-qDixon-WIP) with improved water/fat separation and noise-corrected parameter fitting. For all sequences, three manually co-registered regions of interest (ROIs) were placed in the liver. R2* values were compared and linear regression analysis and Bland-Altman plots calculated. Forty-six out of 415 patients showed fat-water (F/W) swap with qDixon and were excluded. A total of 369 patients (mean age 52 years) were included; in 203/369, the optional qDixon-WIP was acquired, which showed no F/W swaps. A strong correlation was found between R2*-ME-GRE and R2*-qDixon (r² = 0.92, p < 0.001) with Bland-Altman revealing a mean difference of - 3.82 1/s (SD = 21.26 1/s). Correlation between R2*-GRE-ME and R2*-qDixon-WIP was r² = 0.95 (p < 0.001) with Bland-Altman showing a mean difference of - 0.125 1/s (SD = 30.667 1/s). The 3D-multiecho-Dixon sequence is a reliable tool to quantify hepatic iron. Results are comparable with established relaxometry methods. Improvements to the original implementation eliminate occasional F/W swaps and limitations regarding maximum R2* values. • The 3D-multiecho-Dixon sequence for 1.5 T is a reliable tool to quantify hepatic iron. • Results of the 3D-multiecho-Dixon sequence are comparable with established relaxometry methods. • An improved version of the 3D-multiecho-Dixon sequence eliminates minor drawbacks.