Assessment of hepatic steatosis in humans using multivariable quantitative ultrasound Free

The objective of soft-tissue quantitative ultrasound (QUS) is to improve diagnostic ultrasound imaging via quantitative outcomes. Over the past three or so decades, there have been an increasing number of QUS successes. In a UIUC-UCSD collaboration, nonalcoholic fatty liver disease (NAFLD) assessed from seven QUS biomarkers [AC, BSC, three Lizzi-Feleppa markers (slope, intercept, midband), two envelope parameters (k and mu)] derived from ultrasound RF data shows dependencies with the liver fat content in human subjects. 102 participants underwent QUS exams on the right liver lobe with an Acuson S3000 scanner (4C1 and 6C1HD transducers). Two multivariable models have been developed based on QUS biomarkers: (1) generalized linear regression model to predict hepatic PDFF using stepwise regression for biomarker selection and (2) regularized logistic regression model to classify normal (MRI-PDFF <5%, n = 26) versus NAFLD (MRI-PDFF ≥ 5%) using LASSO regularization for biomarker selection. Leave-one-out cross-validation was used for both models. The regression model selected the midband and k-parameter (R² = 0.59, Spearman ρ = 0.84, and Pearson’s r = 0.77). The classifier model selected the midband and μ-parameter (AUC of 0.89). Multivariable QUS provides higher quantification and classification accuracy than univariate QUS approaches. [R01DK106419 and Siemens Healthineers.]