Bi-regional quantitative DCE-MRI for prediction of microvascular invasion in hepatocellular carcinoma and its significance for treatment

Motivation: Accurately predicting microvascular invasion (MVI) risk in hepatocellular carcinoma before surgery could aid clinicians in selecting appropriate surgical approaches to improve the patient’s prognosis. Goal(s): To construct DCE-MRI based nomogram for predicting MVI, and to assess its ability for stratifying the risk of recurrence after hepatectomy and guiding surgical approaches. Approach: Quantitative DCE-MRI parameters from both intra-tumoral region (ITR) and peritumoral region (PTR), along with clinical-radiological (CR) features, were utilized to establish the nomogram. Results: The nomogram presented AUC values of 0.966 in the training and 0.937 in the validation set for predicting MVI. High-risk patients could obtain survival benefit from anatomical resection. Impact: We constructed and evaluated the performance of the bi-regional quantitative DCE-MRI based nomogram for predicting MVI risk in HCC. Our predictive model effectively predicts MVI risk and assists clinicians in selecting appropriate therapeutic strategies for patients.