Diagnostic Value and Influencing Factors of Ultrasound-Derived Fat Fraction for Quantifying MASLD in Children

To evaluate the diagnostic performance of ultrasound-derived fat fraction (UDFF) for quantifying Metabolic dysfunction-associated steatotic liver disease (MASLD) in children, using Magnetic resonance imaging-derived proton density fat fraction (MRI-PDFF) as the reference standard.

This prospective study enrolled 70 children with a median age of 12.23 years [IQR, 10.47-14.29] from December 2025 to April 2026. All participants underwent UDFF, MRI-PDFF, and point shear wave elastography (pSWE) examinations. Steatosis was defined as MRI-PDFF ≥ 6.4%. Correlations, agreement, diagnostic accuracy, and bias predictors were assessed using Pearson/Spearman correlation, Bland-Altman, ROC curves, and multivariable regression.

Of 70 participants, 42 had MASLD. UDFF correlated strongly with MRI-PDFF (r = 0.84, p < 0.001). Overall mean bias was -2.19% (95% concordance limits: -15.07% to 10.68%), with a more pronounced systematic numerical bias observed in moderate-to-severe steatosis (mean bias -9.23%). No significant bias predictors were found. For detecting steatosis (≥ 6.4%), UDFF had an AUC of 0.98, optimal cutoff 5.5% (sensitivity 97.6%, specificity 92.9%, accuracy 95.7%); for ≥ S2 steatosis, the cutoff was 12.5% (AUC 0.96). Intraclass correlation for five repeated UDFF measurements was 0.98 (95% CI: 0.98-0.99). Visceral fat thickness (β = 2.004, p = 0.003) and measurement depth (β = 4.716, p = 0.033) independently predicted UDFF values. UDFF showed a weak negative correlation with pSWE Young's modulus (r = -0.258, p = 0.033).

UDFF is a non-invasive, accurate, and reproducible tool for quantifying paediatric hepatic steatosis. Although a systematic numerical bias exists relative to MRI-PDFF in moderate-to-severe cases, UDFF retains excellent discriminative ability and is suitable for screening, staging, and longitudinal follow-up of paediatric MASLD.