Hepatic Fat Quantification Using Chemical Shift MR Imaging and MR Spectroscopy in the Presence of Hepatic Iron Deposition: Validation in Phantoms and in Patients With Chronic Liver Disease

Purpose: To compare the accuracy of four chemical shift magnetic resonance imaging (MRI) (CS-MRI) analysis methods and MR spectroscopy (MRS) with and without T2-correction in fat quantification in the presence of excess iron. Materials and Methods: CS-MRI with six opposed-and in-phase acquisitions and MRS with five-echo acquisitions (TEs of 20, 30, 40, 50, 60 msec) were performed at 1.5 T on phantoms containing various fat fractions (FFs), on phantoms containing various iron concentrations, and in 18 patients with chronic liver disease. For CS-MRI, FFs were estimated with the dual-echo method, with two T2*-correction methods (triple-and multiecho), and with multi-interference methods that corrected for both T2* and spectral interference effects. For MRS, FF was estimated without T2-correction (single-echo MRS) and with T2-correction (multiecho MRS). Results: In the phantoms, T2*- or T2-correction methods for CS-MRI and MRS provided unbiased estimations of FFs (mean bias, -1.1% to 0.5%) regardless of iron concentration, whereas the dual-echo method (-5.5% to -8.4%) and single-echo MRS (12.1% to 37.3%) resulted in large biases in FFs. In patients, the FFs estimated with triple-echo (R = 0.98), multiecho (R = 0.99), and multi-interference (R 0.99) methods had stronger correlations with multiecho MRS FFs than with the dual-echo method (R = 0.86; P <= 0.011). The FFs estimated with multi-interference method showed the closest agreement with multiecho MRS FFs (the 95% limit-of-agreement, -0.2 +/- 1.1). Conclusion: T2*- or T2-correction methods are effective in correcting the confounding effects of iron, enabling an accurate fat quantification throughout a wide range of iron concentrations. Spectral modeling of fat may further improve the accuracy of CS-MRI in fat quantification.