Radiomics Analysis of Gadoxetic Acid-Enhanced MRI for Predicting Proliferative Hepatocellular Carcinoma

Motivation: Proliferative HCC exhibits more aggressive biological behavior and has a poorer prognosis. Goal(s): This study aimed to preoperatively diagnose proliferative HCC using diverse radiomics models and a nomogram, while also assessing their prognostic value. Approach: This study is a retrospective research. Results: BCLC stage and combined radscore are independent prognostic factors for proliferative HCC. The nomograms based on these factors had AUCs of 0.838 and 0.801 in the training and validation cohorts, respectively. The delta radiomics model (DR)-predicted proliferative HCC showed the best performance in risk stratification. Impact: The nomogram combining independent clinical predictive factors and the combined radscore demonstrates the best diagnostic efficacy for proliferative HCC, while different radiomic models and the nomogram can provide varying prognostic stratification values.