Deep learning assisted detection of chronic lung allograft dysfunction using pulmonary DCE-MRI

Motivation: Chronic Lung Allograft Dysfunction (CLAD) is a significant cause of mortality among lung transplant recipients, making early detection crucial for timely intervention. Goal(s): This study aimed to evaluate a deep learning-based method for distinguishing patients with CLAD from non-CLAD using 3D DCE-MRI of the lungs. Approach: We developed a model using transfer learning from pre-trained VGG-16 model weights and evaluated model performance by 8-fold cross-validation. Results: The model achieved an average AUROC of 98.5%, indicating high accuracy in distinguishing between CLAD and non-CLAD cases. Impact: This deep learning approach effectively combines spatial, depth, and temporal information from 3D DCE-MRI, offering a promising tool for enhancing CLAD diagnostic precision.