MRI radiomics plus clinical AI model predicts early HCC recurrence after RFA

In a retrospective study of 169 patients, an integrated MRI radiomics and clinical random forest model predicted early hepatocellular carcinoma (HCC) recurrence after radiofrequency ablation (RFA) with a test AUC of 0.909 (sensitivity 83.3%, specificity 90.9%).

• The integrated model used 16 MRI radiomics features and three clinical predictors (alpha-fetoprotein, platelet count, tumor location).
• Radiomics-only models significantly outperformed clinical-only models (test AUC 0.826–0.830 vs 0.688–0.724; P<0.05).
• Decision curve analysis confirmed net clinical benefit across threshold probabilities of 15%–70%.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.