Fig. 2From: Performance comparison between multi-center histopathology datasets of a weakly-supervised deep learning model for pancreatic ductal adenocarcinoma detectionClassification performance of the independent models and their generalization. Panel A Confusion matrix and ROC curve obtained in the patient-wise stratified 10-Fold CV performed on the TCGA + GTEx data. Panel B Confusion matrix and ROC curve obtained in the patient-wise stratified 10-Fold CV performed on the CPTAC data. Panel C Confusion matrices obtained from the pretrained models, trained on CPTAC and TCGA + GTEx respectively, over the opposite dataset (CPTAC model predicting TCGA + GTEx data and TCGA + GTEx model predicting CPTAC data)Back to article page