Ultrasonographic assessment of liver fibrosis with computer-assisted analysis of liver surface irregularities

Purpose: The goal of this study was to evaluate the diagnostic accuracy of a software program that automatically analyzes the liver surface to diagnose significant fibrosis, by comparing it to the subjective analysis of a radiologist and to transient elastography (Fibroscan (R)). Patients and methods: One hundred fourteen patients with chronic liver disease were included in the study. They underwent liver biopsy, FibroScan (R) and ultrasonographic examination of the liver surface. The liver surface was analyzed by a software program that gave a score of surface irregularities. This evaluation was compared to subjective analysis by a radiologist expert in liver imaging and by two general radiologists. Results: Fifty percent of the patients had significant fibrosis according to the METAVIR score. The AUROC for the diagnosis of significant fibrosis by the software program was 0.80 (95% CI: 0.71-0.87), which was equivalent (P = 0.86) to that of FibroScan (R) (0.81; 95% CI: 0.71-0.89). Results of the subjective analysis by the expert radiologist were poorer than those of those of the software analysis (P = 0.02) (AUROC = 0.66; 95% CI: 0.56-0.75). Interobserver agreement among radiologists was poor (0.25 < kappa < 0.37). Conclusion: Computer-assisted liver surface analysis was better than subjective analysis, and similar to that of the FibroScan (R). This method could be useful for the diagnosis of significant fibrosis in patients with chronic hepatitis and complementary to the other non-invasivediagnostic tests. (C) 2015 Editions francaises de radiologie. Published by Elsevier Masson SAS. All rights reserved.