CEMRI Habitat and Deep Learning Fusion Model Predicts VETC Pattern in HCC with High Accuracy

A fusion model combining CEMRI habitat and DL features predicted VETC in HCC with AUCs 0.901 (training) and 0.870 (validation). High-risk group showed reduced resting dendritic cells (P<0.05). Retrospective study.

• Retrospective analysis of 336 patients with HCC; model development and internal validation. Radiogenomic analysis in 47 patients from TCIA/TCGA.
• Fusion model outperformed individual ITH (AUC 0.806) and DL (0.745) models; CIs not reported.
• No external validation; single-center retrospective design limits generalizability.

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