Radiomics of Hepatocellular Carcinoma: Identifying Predictors of Microvascular Invasion Using Multi-Phase CT Analysis

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Microvascular invasion (MVI) is a critical prognostic factor in hepatocellular carcinoma (HCC).
• Accurate prediction of MVI is essential for guiding treatment decisions and improving patient outcomes.
• Non-invasive imaging biomarkers are needed to assess MVI risk before surgery.

Purpose of the Study:

• To investigate the potential of radiomic texture features from multi-phase contrast-enhanced computed tomography (CECT) for predicting MVI in HCC.
• To evaluate radiomic features across different CECT phases (arterial, venous, delayed) and tumor margin expansions.
• To identify reliable radiomic predictors of MVI in HCC lesions of varying sizes.

Main Methods:

• Retrospective analysis of 54 HCC lesions from 49 patients who underwent liver resection.
• Extraction of 642 radiomic features from multi-phase CECT using original and 5 mm-expanded tumor margins.
• Statistical analysis to assess the predictive performance of texture features, focusing on lesion size subgroups.

Main Results:

• In the 20-50 mm lesion subgroup (n=37), arterial phase texture homogeneity features achieved an Area Under the Curve (AUC) of 0.772 for MVI prediction.
• Radiomic features from lesions <20 mm showed evidence of overfitting and were excluded from primary analysis.
• Delayed phase features demonstrated preliminary associations (AUC 0.8) in a smaller LR-3/4 subset (n=20).

Conclusions:

• Multi-phase CECT radiomic analysis holds potential for predicting MVI in intermediate-sized HCC.
• Arterial phase texture homogeneity is a promising feature for MVI prediction in this cohort.
• External validation in larger, multi-center cohorts is crucial before clinical implementation.