Deep learning-based medical image analysis with explainable transfer learning

Abstract

Along with the remarkable progress of deep learning-based medical image analysis (DLB-MIA), deep learning models are widely deployed for computer-aided diagnosis (CAD). However, Data scarcity and model interpretability pose noteworthy challenges to DLB-MIA application. Explainable artificial intelligence (XAI) can be applied in transfer learning to address the aforementioned problems, which makes explainable transfer learning a promising methodology. The utilization of transfer learning combined with XAI techniques is therefore surveyed. The current status of explainable transfer learning is summarized. The application of explainable transfer learning in DLB-MIA is investigated respectively on convolutional neural networks (CNNs) and transformers.