Abstract: Objective To explore the application of radiomics-based analysis in predicting microvascular invasion (MVI) in hepatocellular carcinoma (HCC) patients. Methods This retrospective study included 210 patients with surgically confirmed diagnosis of HCC, all of whom underwent contrast-enhanced CT scans and obtained pathological specimens. Multiple radiomics features were extracted from the contrast-enhanced CT images. The clinical data of the patients were also collected. Univariate analysis was performed to identify clinical and radiomics features that were significantly associated with MVI. Subsequently, a multivariate regression model was used to select the predictive factors for MVI. Results Of the 210 patients, 67 were confirmed with MVI through postoperative pathology, resulting in an incidence rate of 31.9%. By univariate and logistic multivariate regression analyses, serum AFP (P=0.039), maximum tumor length (P=0.021), and intra-tumoral arteries (P=0.023) were identified as the influencing factors for MVI. The area under the ROC curve (AUC) for assessing the model's performance was 0.895. Conclusion This study demonstrates a crucial role of radiomics features in predicting MVI in HCC, with promising predictive performance. The radiomics-based model may provide clinicians with comprehensive and accurate information for individualized treatment decisions, thus potentially improve the survival rate and quality of life for HCC patients.

Key words: Hepatocellular carcinoma, Microvascular invasion, Influencing factors, Enhanced computed tomography