Predicting the histological grade of hepatocellular carcinoma based on dynamic radiomics
Year: 2025
Type: article
Study design: Observational Study
Abstract: Background Radiomics studies for preoperative prediction of histological grade of hepatocellular carcinoma (HCC) have primarily concentrated on static features, thereby overlooking the dynamic changes in radiomics features across different phases. Objective To investigate the efficacy of dynamic radiomics based on contrast-enhanced (CE) MRI for the preoperative prediction of histological grade (Edmondson–Steiner grade I–II vs. III–IV) of HCC. Methods Data from 244 HCC patients who underwent preoperative CE-MRI were retrospectively collected from two centres (184 and 60 patients, respectively). We used precontrast, arterial, portal venous, and delayed phase images to extract and construct static and dynamic radiomics features. LASSO was used to select radiomics features, and logistic regression was used for modelling. Predicting performance for histological grade was compared among three models: static, dynamic, and dynamic–static radiomics models. To provide visual explanations, Shapley Additive Explanation (SHAP) values were depicted as beeswarm plots. Lastly, the optimal radiomics score was combined with clinicopathological factors for the radiomics–clinical model construction. Results The areas under the curve (AUCs) of the static, dynamic, and dynamic–static radiomics models were respectively 0.770, 0.826, and 0.841 in the internal validation set and 0.712, 0.774, and 0.776 in the external validation set. When AFP, liver cirrhosis, hepatitis B, and dynamic–static radiomics score were combined, the performance of the radiomics–clinical model was comparable to that of the dynamic–static radiomics model (internal validation cohort: AUC = 0.849; external validation cohort: AUC = 0.785). Conclusion Dynamic radiomics could improve the ability to discriminate high-grade and low-grade HCC cases and inform treatment decisions.
Source: Asian Journal of Surgery
Language: English
Subfield: Radiology, Nuclear Medicine and Imaging
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $1,200
OpenAlex ID: W4413186742
