MRI Radiomics for Distinguishing HBV-Related Early Hepatocellular Carcinoma from Precancerous Nodules.

Objective: To investigate the predictive value of intratumoral and peritumoral radiomic features from multiparametric MRI in differentiating HBV-associated early hepatocellular carcinoma (eHCC) from precancerous nodules. Methods: Between January 2019 and October 2024, a total of 133 patients with eHCC and 89 patients with pre-HCC lesions who underwent preoperative MRI were enrolled in the study. Multivariate analysis identified independent predictors for eHCC used to build a clinical prediction model. Radiomics models based on multiphase MRI across multiple tumor regions (entire tumor, Peri_4mm, Peri_6mm, Peri_8mm) were developed. Features from the intratumoral and optimal peritumoral region were combined into an IntraPeri model. A nomogram was developed by combining clinical and IntraPeri model variables. Model performance was evaluated using the area under the curve (AUC). Results: 222 patients (mean age, 57.2 years ± 11.0; 164 men) were evaluated. DWI hyperintensity, T1WI hypointensity, and a low globulin-to-lymphocyte ratio (GLR) were identified as risk factors for eHCC. 1,409 radiomic features were extracted from each ROI; 27 were retained for analysis and model development. The clinical model achieved AUCs of 0.826 and 0.798 in the training and internal validation cohorts, respectively, while the intratumoral model yielded AUC values of 0.884 and 0.838. The Peri_6mm model outperformed the Peri_8mm and Peri_4mm models. The IntraPeri model demonstrated excellent performance, achieving AUCs of 0.947 and 0.935 in the training and internal validation cohorts, respectively. The comprehensive nomogram demonstrated the highest performance with AUCs of 0.958 and 0.950. Conclusion: Our study highlights the potential of a multiparametric MRI-based radiomics nomogram that integrates both intratumoral and peritumoral features as an effective tool for differentiating eHCC.