Related Experiment Video
Updated: May 28, 2026
A Three-Dimensional Digital Model for Early Diagnosis of Hepatic Fibrosis Based on Magnetic Resonance Elastography
Published on: July 21, 2023
CECT Radiomics and Liver Fibrosis Markers for Predicting Early Recurrence in Older Patients with Hepatocellular
Fang Liu1, Zhi Zou1, Liuchang Zheng2
1Department of Medical Imaging, Henan Provincial People's Hospital, Zhengzhou, People's Republic of China.
Purpose:
Hepatocellular carcinoma (HCC) carries a high risk of early postoperative recurrence in older patients, and traditional prediction methods are limited by invasiveness and subjectivity. Radiomics studies lack older adult-specific designs and adequate integration of liver fibrosis factors, limiting applicability. Accordingly, we aimed to develop and validate a non-invasive model integrating contrast-enhanced computed tomography (CECT) radiomics and liver fibrosis indicators for predicting early postoperative recurrence in older patients with HCC and identify key risk factors.
Patients And Methods:
In total, 169 older patients with HCC (≥60 years) who underwent radical hepatectomy were retrospectively enrolled and randomly assigned to training (n = 131) and validation (n = 38) sets. Radiomics features were derived from preoperative CECT scans acquired during the portal venous phase. The radiomics score (Rad-score) was constructed using the least absolute shrinkage and selection operator algorithm for optimal feature selection. A radiologic‒radiomics (RR) nomogram was developed using multivariable logistic regression to identify independent predictors of early recurrence. Model performance was assessed using the area under the curve *AUC), calibration plots, the Hosmer-Lemeshow goodness-of-fit test, and decision curve analysis (DCA).
Results:
Sixty-three patients (37.3%) experienced early postoperative recurrence. Tumor diameter, total bilirubin, albumin, type IV collagen, and Rad-score (derived from five optimal radiomics features) were identified as independent predictors. The RR nomogram model achieved AUCs of 0.871 and 0.722 in the training and validation sets, respectively, for predicting early recurrence. Calibration curves and Hosmer-Lemeshow test demonstrated good model calibration (P > 0.05); DCA showed a higher net clinical benefit than "treat-all" or "treat-none" strategies across multiple threshold probabilities.
Conclusion:
The integrated RR nomogram shows promising discriminatory and calibration ability for predicting early recurrence in older patients with HCC. While this nomogram may serve as a non-invasive preoperative risk-stratification tool, prospective multicenter validation is warranted before clinical adoption.
