Fig. 2

Workflow for developing the radiomics nomogram models. CT image segmentation was performed using manual semiautomatic segmentation using radiomics prototype software (Radiomics, Frontier, Siemens). The radiomic features from the volumes of interest were then computed using the CT images on the prototype. A predictive model was constructed on the basis of the CT-derived radiomic features using the random forest (RF) method to output a radiomics score (Rad-score) for each patient. The Rad-score was combined with significant clinicopathological factors for multivariate logistic regression analysis to develop radiomics nomogram model 1 to predict the durable clinical benefit (DCB). Radiomics nomogram model 2 was established to predict the progression-free survival (PFS) and was developed via multivariate logistic regression analysis of the Rad-score and significant risk factors combined