Mapping of liver fat with triple-echo gradient echo imaging: validation against 3.0-T proton MR spectroscopy.

Abstract

The purpose of this study was to validate a magnetic resonance imaging (MRI) technique for mapping liver fat, using (1)H magnetic resonance spectroscopy ((1)H-MRS) as the reference standard. In 91 patients with type 2 diabetes, 3.0-T single-voxel point-resolved (1)H-MRS was used to calculate the liver fat fraction (LFF) from the water (4.76 ppm) and methylene (1.33 ppm) peaks, corrected for T1 and T2 decays. LFF (corrected for T1 and T2* decays) was also obtained from the mean signal intensity on a map built from a triple-echo (consecutive in-phase, opposed-phase, and in-phase echo times) breath-hold gradient echo sequence, using basic image calculation functions (arithmetic mean, subtraction, division, multiplication by a numerical factor). Mean LFF was 8.9% (range, 0.9-33.5) by MRI and 8.8% (range, 0-34.1) by (1)H-MRS. Pearson's coefficient was 0.976 (P < 0.0001) and Lin's coefficient was 0.975 (P < 0.0001). Liver segment had no significant influence. With Bland-Altman analysis, 95.6% (87/91) of data points were within the limits of agreement. Given its excellent agreement with (1)H-MRS, our mapping technique can be used for visual and quantitative evaluation of liver fat in everyday practice.