650 bilateral EVLP cases (2008-2022) at Toronto General Hospital, with 1300 radiographs, were split 80:20 into training and validation sets. CNN architectures (ResNet-50, ResNeXt-50, RexNet-100, EfficientNet-B2/B3, DenseNet-121) were pretrained on ChestX-ray14 datasets and finetuned on EVLP radiographs for lung transplant outcome classification. Model performance was evaluated using accuracy and AUROC. GradCAM was used for CNN interpretation, and XGBoost models analyzed EVLP data and latent CNN features.