Prediction power of radiomics in early recurrence of hepatocellular carcinoma: A systematic review and meta-analysis.

ABSTRACT

BACKGROUND: Raiomics is an emerging auxiliary diagnostic tool, but there are still differences in whether it can be applied to predict early recurrence of hepatocellular carcinoma (HCC). The purpose of this meta-analysis was to systematically evaluate the predictive power of radiomics in the early recurrence (ER) of HCC. METHODS: Comprehensive studies on the application of radiomics to predict ER in HCC patients after hepatectomy or curative ablation were systematically screened in Embase, PubMed, and Web of Science. RESULTS: Ten studies which is involving a total of 1929 patients were reviewed. The overall estimates of radiomic models for sensitivity and specificity in predicting the ER of HCC were 0.79 (95% confidence interval [CI] 0.68-0.87) and 0.83 (95% CI 0.73-0.90), respectively. The area under the summary receiver operating characteristic curve (SROC) was 0.88 (95% CI 0.85-0.91). CONCLUSIONS: The imaging method is a reliable method for diagnosing HCC. Radiomics, which is based on medical imaging, has excellent power in predicting the ER of HCC. With the help of radiomics, we can predict the recurrence of HCC after surgery more effectively and provide a useful reference for clinical practice.