Fig. 6From: CT-based deep learning radiomics nomogram for the prediction of pathological grade in bladder cancer: a multicenter studyActivation maps of the deep convolutional neural networks for high versus low grade bladder cancer, reflected the weights corresponding to different pathological grades of tumors. These maps were constructed using data from the corticomedullary phase, nephrographic phase, and excretory phase. Red areas indicate higher correlation with tumor gradeBack to article page