Texture Analysis of Hepatocellular Carcinoma on Magnetic Resonance Imaging: Assessment for Performance in Predicting Histopathologic Grade

Abstract

Objective: The aim of the study was to evaluate the performance of texture analysis for discriminating the histopathological grade of hepatocellular carcinoma (HCC) on magnetic resonance imaging.

Methods: Preoperative magnetic resonance imaging data from 101 patients with HCC, including T2-weighted imaging, arterial phase, and apparent diffusion coefficient mapping, were analyzed using texture analysis software (TexRAD). Differences among the histological groups were analyzed using the Mann-Whitney U test. The performance of texture features was evaluated using receiver operating characteristic analysis.

Results: Entropy was the most significantly relevant texture feature for distinguishing each histological grade group of HCC (P < 0.05). In ROC analysis, entropy with spatial scale filter 3 (area under curve the receiver operating characteristic curve [AUC], 0.778), mean with coarse filter (spatial scale filter 5; AUC, 0.670), and skewness without filtration (AUC, 0.760) had the highest AUC value on T2-weighted imaging, arterial phase, and apparent diffusion coefficient maps, respectively.

Conclusions: Magnetic resonance imaging texture analysis demonstrated potential for predicting the histopathological grade of HCCs.