B-mode ultrasound for the assessment of hepatic fibrosis: a quantitative multiparametric analysis for a radiomics approach

Abstract

Hepatic fibrosis and cirrhosis are a growing global health problem with increasing mortality rates. Early diagnosis and staging of hepatic fibrosis represent a major challenge. Currently liver biopsy is the gold standard for fibrosis assessment; however, biopsy requires an invasive procedure and is prone to sampling error and reader variability. In the current study we investigate using quantitative analysis of computer-extracted features of B-mode ultrasound as a non-invasive tool to characterize hepatic fibrosis. Twenty-two rats were administered diethylnitrosamine (DEN) orally for 12 weeks to induce hepatic fibrosis. Four control rats did not receive DEN. B-mode ultrasound scans sampling throughout the liver were acquired at baseline, 10, and 13 weeks. Computer extracted quantitative parameters representing brightness (echointensity, hepatorenal index) and variance (heterogeneity, anisotropy) of the liver were studied. DEN rats showed an increase in echointensity from 37.1 ± SD 7.8 to 53.5 ± 5.7 (10 w) to 57.5 ± 6.1 (13 w), while the control group remained unchanged at an average of 34.5 ± 4.5. The three other features studied increased similarly over time in the DEN group. Histologic analysis showed METAVIR fibrosis grades of F2-F4 in DEN rats and F0-F1 in controls. Increasing imaging parameters correlated with increasing METAVIR grades, and anisotropy showed the strongest correlation (ρ = 0.58). Sonographic parameters combined using multiparametric logistic regression were able to differentiate between clinically significant and insignificant fibrosis. Quantitative B-mode ultrasound imaging can be implemented in clinical settings as an accurate non-invasive tool for fibrosis assessment.