Fig. 2

ROC curves of the DLCT parameters and prediction models, and the combined prediction nomogram. a Predictive performance of DLCT parameters in predicting MSI status of GC in the training set. b The combined nomogram for discriminating MSI status of gastric cancer in the training set. The prediction nomogram was built based on the multivariate logistic model integrated with the variables of clinical features and C_DLCT. c ROC curves of the clinical, DLCT, and combined models for the prediction of microsatellite instability status in the training set. d ROC curves of the clinical, DLCT, and combined models for the prediction of microsatellite instability status in the training set. Both in the training set and the validation set, the combined model showed the best prediction performance. C_DLCT combined dual-layer spectral detector CT parameters; DLCT dual-layer spectral detector CT; GC, gastric cancer; ID iodine density; λHU the slope of the spectral curve; MSI microsatellite instability; MSI-H microsatellite instability high; NID normalized iodine density; VP venous phase; Zeff effective atomic number