Fig. 4From: Feasibility and effectiveness of automatic deep learning network and radiomics models for differentiating tumor stroma ratio in pancreatic ductal adenocarcinomaThe ROC curves, calibration curves, decision curves among radiomics and deep learning groups, respectively. a, c, e ROC curves, calibration curves, and decision curves among radiomics models. b, d, f ROC curves, calibration curves, decision curves among deep learning models. The RF model and Resnet18 achieved the optimal efficiency in radiomics models and deep learning models, respectively. The calibration curves presented a good consistency between predicted and actual TSR in radiomics and deep learning models. The graphs show that the SVM model and ResNet18 have the greatest net benefit in radiomics models and deep learning models, respectivelyBack to article page