DECT radiomics model predicts gastric cancer immunotherapy response across multiple centers

A dual-energy CT radiomics model predicted immunotherapy response in gastric cancer with an AUC of 0.827 in training and 0.775–0.803 in external validation sets. High-risk patients had significantly worse progression-free and overall survival.

• Multicenter study of 344 gastric adenocarcinoma patients receiving PD-1/PD-L1 inhibitor combination chemotherapy, split into training, validation, external-validation, and cross-platform validation sets.
• Logistic regression model achieved stable performance across all sets with no significant AUC differences (DeLong P > 0.48); calibration curves and decision curve analysis confirmed clinical utility.
• Retrospective design; the specific dual-energy parameters and machine learning algorithms beyond logistic regression are not detailed in the abstract.

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