3D Fractal Analysis of Gd-EOB-DTPA-MRI for Vessels Encapsulating Tumor Clusters Prediction in Hepatocellular Carcinoma.

This study aims to evaluate the potential role of 3-dimensional (3D) fractal analysis derived from Gd-EOB-DTPA-MRI in predicting vessels encapsulating tumor clusters (VETC) pattern in patients with hepatocellular carcinoma (HCC). This retrospective study included 212 HCC patients who underwent preoperative Gd-EOB-DTPA-MRI between January 2018 and August 2024. Fractal dimension (FD) and lacunarity from arterial and hepatobiliary phases were extracted using box-counting method. Variables associated with VETC pattern were analyzed using univariate and multivariate logistic regression. Model performance was assessed using area under the curve (AUC). Recurrence-free survival (RFS) was analyzed using Kaplan-Meier methods. FD and lacunarity from arterial and hepatobiliary phases were higher in VETC-positive HCC than in VETC-negative HCC (P < 0.05). The AUC of the fractal feature model was 0.76 (0.70,0.82). Multivariate analysis identified AFP, capsule, and intratumoral necrosis as independent predictors of VETC pattern. Combining these with fractal parameters yielded a hybrid model with an AUC of 0.80 (0.75,0.86). Cox regression identified AST/ALT, intravascular tumor thrombus, and VETC pattern as risk factors for RFS, with significantly shorter RFS in VETC-positive HCC (P < 0.05). MRI-based 3D fractal parameters show significant correlation with VETC pattern. When combined with clinical radiological features, FD and lacunarity can effectively predict VETC pattern and identify high-risk HCC patients.