Background & aims: Despite increasing use of various imaging examinations for non-invasive assessment of hepatic steatosis (HS), their relative accuracy is unknown. The objective of this study is to prospectively compare the accuracy of computed tomography (CT), dual gradient echo magnetic resonance imaging (DGE-MRI), proton magnetic resonance spectroscopy ((1)H-MRS), and ultrasonography (US) for the diagnosis and quantitative estimation of HS.
Methods: A total of 161 consecutive potential living liver donors underwent US (performed by two independent radiologists, US1 and US2), CT, DGE-MRI, (1)H-MRS, and liver biopsy on the same day. Using the histologic degree of HS as the reference standard, we compared the diagnostic performance of US1, US2, CT, DGE-MRI, and (1)H-MRS for diagnosing HS >or= 5% and HS >or= 30% and compared the accuracy of CT, DGE-MRI, and (1)H-MRS in the quantitative estimation of HS.
Results: DGE-MRI and (1)H-MRS significantly outperformed CT and US for the diagnosis of HS5%. DGE-MRI showed a tendency of higher accuracy than the other examinations for diagnosing HS >or= 30%. The cross-validated sensitivity and specificity of DGE-MRI at the optimal cut-off were 76.7% and 87.1%, respectively, for diagnosing HS >or= 5% and 90.9% and 94%, respectively, for diagnosing HS >or= 30%. The cross-validated Bland-Altman 95% limits of agreement between the estimated degree of HS on imaging examinations and the histologic degree of HS, were the narrowest with DGE-MRI, yielding -12.7% to 12.7%.
Conclusions: Among CT, DGE-MRI, (1)H-MRS, and US, DGE-MRI is the most accurate method for the diagnosis and quantitative estimation of HS. Therefore, DGE-MRI may be the preferred imaging examination for the non-invasive assessment of HS.