Deep learning-assisted tumor radiomic dynamics on MRI predict pathological complete response in HCC undergoing immune-based therapy followed by hepatectomy - PubMed
3 hours ago
- #Pathological Complete Response
- #Radiomics
- #Hepatocellular Carcinoma
- Deep learning-assisted tumor radiomic dynamics on MRI predict pathological complete response (pCR) in hepatocellular carcinoma (HCC) undergoing immune-based therapy followed by hepatectomy.
- The study developed a model integrating clinicopathological and radiomic features to predict pCR in initially unresectable HCC (uHCC).
- Temporal radiomics features were extracted from baseline, post-treatment, and delta MRIs, with serum AFP response also considered.
- Feature selection involved univariate analysis, collinearity assessment, LASSO, and random forest, with 14 machine learning models benchmarked.
- The delta radiomic model outperformed baseline and preoperative models in predicting lesion-level pCR, with AUCs of 0.835 (test) and 0.783 (validation).
- The combined radiomics-AFP model achieved higher AUCs of 0.920 (test) and 0.857 (validation) in predicting lesion-level pCR.
- Dynamic radiomic changes effectively predict pCR in uHCC after conversion therapy, offering a non-invasive method for assessing pCR and guiding personalized treatment decisions.