Quantitative ultrasound-derived estimated fat fraction for detection and grading of metabolic dysfunction-associated steatotic liver disease in a pediatric population

Background Pediatric metabolic dysfunction-associated steatotic liver disease (MASLD) prevalence is approximately 10%, but ultrasound fat quantification lacks validation. This study evaluated the diagnostic performance of quantitative ultrasound-derived estimated fat fraction (USFF) against magnetic resonance imaging proton density fat fraction (MRI-PDFF) in children with suspected MASLD. Methods This prospective single-center study enrolled 65 children aged 6–18 years with suspected MASLD. USFF, tissue attenuation imaging (TAI), tissue scatter distribution imaging (TSI), hepato-renal index (EzHRI), and shear wave elastography (SWE) were measured on a Samsung R10 Prestige platform; 3.0 T MRI-PDFF served as reference. Steatosis was graded using pediatric PDFF thresholds. Diagnostic performance was assessed by ROC analysis with DeLong comparisons and internally validated by bootstrap resampling; agreement by Bland-Altman analysis; inter-operator reproducibility assessed in a randomly selected 30-child subset. Results Among 65 children (48 boys; median age 11.2 years), 20 were classified as S0, 29 as S1, and 16 as S2-3. USFF AUCs were 0.941 (≥S1) and 0.960 (≥S2), higher than EzHRI and SWE, and similar to TAI (0.903, 0.957) and TSI (0.900, 0.880). USFF strongly correlated with MRI-PDFF (ρ = 0.906) but systematically underestimated PDFF (mean bias − 3.13%; non-parametric limits of agreement − 14.51% to + 4.24%). In an exploratory analysis, younger children (<12 years) showed larger bias (−4.00% vs − 1.82%, p = 0.029) with similar AUC point estimates. Conclusions USFF shows high diagnostic performance (AUC > 0.94) for pediatric MASLD, but its 3.13% underestimation and wide non-parametric limits of agreement warrant caution. USFF may serve as a practical tool for categorical steatosis grading in children with suspected MASLD, and the proposed thresholds require external validation before clinical use.