Dynamic radiomic features from MRI can predict complete response in hepatocellular carcinoma (HCC) post-immune therapy, critical for surgical planning.

• A delta radiomic model achieved an AUC of 0.835 in predicting lesion-level complete response compared to 0.483 for baseline models.
• Combining radiomics with serum AFP response boosted predictive accuracy, yielding AUCs of 0.920 in the test cohort.

Utilizing this model could refine patient selection and improve surgical outcomes for those undergoing hepatectomy.

• Validation across multiple cohorts demonstrates strong generalizability of the findings.