Strategy to Improve Diagnostic Performance of Hepatic Steatosis Grading Using Ultrasound: Combination of Qualitative...

Background and Objectives: To propose an integrated qualitative and quantitative US strategy for improving the diagnostic accuracy of hepatic steatosis (HS) grading. Materials and Methods: We retrospectively identified patients at a tertiary hospital who underwent B-mode US examination, attenuation imaging (ATI), hepatorenal index (HRI), and liver biopsy on the same day between 2023 and 2024. B-mode liver US images were independently reviewed by three radiologists with varying levels of experience. Each radiologist qualitatively assessed the degree of HS using a three-tier grading system: (1) no HS, (2) mild HS, and (3) moderate/severe HS. Median attenuation coefficient (AC) and HRI values were recorded. Liver biopsy pathology reports were used as the reference standard. We evaluated four strategies integrating qualitative assessment with ATI or HRI and compared their diagnostic performance. Results: Based on B-mode US, accuracies for detecting HS by the three radiologists were 87.0%, 87.0%, and 86.0%, respectively; for moderate/severe HS, the values were 79.0%, 76.0%, and 71.0%. ATI and HRI alone yielded diagnostic accuracies of 83.0% and 81.0%, respectively, for any HS and 67.0% and 53.0%, respectively, for moderate/severe HS. The highest accuracy was achieved using a strategy that retained the qualitative grade for one-grade discrepancies and assigning the intermediate grade only for two-grade discrepancies between the AC-based grade and the qualitative grade (93.0%, 87.0%, and 90.0% for any HS; 80.0%, 76.0%, and 74.0% for moderate/severe HS). Conclusions: The proposed integration strategy, which retained the qualitative grade for one-grade discrepancies and assigned the intermediate grade only for two-grade discrepancies, improved or maintained diagnostic performance, regardless of the radiologist's level of experience.