Comparison of Chemical Shift-Encoded Water-Fat MRI and MR Spectroscopy in Quantification of Marrow Fat in Postmenopausal Females

Purpose: To validate a chemical shift-encoded (CSE) water-fat imaging for quantifying marrow fat fraction (FF), using proton magnetic resonance spectroscopy (MRS) as reference. Materials and Methods: Multiecho T-2-corrected MRS and CSE imaging with eight-echo gradient-echo acquisitions at 3T were performed to calculate marrow FF in 83 subjects, including 41 with normal bone mineral density (BMD), 26 with osteopenia, and 16 with osteoporosis (based on DXA). Eight participants were scanned three times with repositioning to assess the repeatability of CSE FF map measurements. Pearson correlation coefficient, Bland-Altman 95% limit of agreement, and Lin's concordance correlation coefficient were calculated. Results: The Pearson correlation coefficient was 0.979 and Lin's concordance correlation coefficient was 0.962 between CSE-based FF and MRS-based FF. All data points, calculated using the Bland-Altman method, were within the limits of agreement. The intra-and interrater agreement for average CSE-based FF was excellent (intrarater, intraclass correlation coefficient [ICC] = 0.993; interrater, ICC = 0.976-0.982 for different BMD groups). In the subgroups of varying BMD, inverse correlations were observed to be very similar between BMD (r = -0.560 to -0.710), T-score (r = -0.526 to -0.747), and CSE-based FF, and between BMD (r = -0.539 to -0.706), T-score (r = 0.501 to -0.742), and MRS-based FF even controlling for age, years since menopause, and body mass index. The repeatability for CSE FF map measurements expressed as absolute precision error was 1.45%. Conclusion: CSE imaging is equally accurate in characterizing marrow fat content as MRS. Given its excellent correlation and concordance with MRS, the CSE sequence could be used as a potential replacement technique for marrow fat quantification.