Interpreting Deep Ensemble Learning through Radiologist Annotations for COVID-19 Detection in Chest Radiographs

Data-driven deep learning (DL) methods using convolutional neural networks (CNNs) demonstrate promising performance in natural image computer vision tasks. However, using these models in medical computer vision tasks suffers from several limitations, viz., (i) adapting to visual characteristics that are unlike natural images; (ii) modeling random noise during training due to stochastic optimization and backpropagation-based learning strategy; (iii) challenges in explaining DL black-box behavior to support clinical decision-making; and (iv) inter-reader variability in the ground truth (GT) annotations affecting learning and evaluation. This study proposes a systematic approach to address these limitations for COVID-19 detection using chest X-rays (CXRs). Specifically, our contribution benefits from (i) pretraining specific to CXRs in transferring and fine-tuning the learned knowledge toward improving COVID-19 detection performance; (ii) using ensembles of the fine-tuned models to further improve performance compared to individual constituent models; (iii) performing statistical analyses at various learning stages to validate our claims; (iv) interpreting learned individual and ensemble model behavior through class-selective relevance mapping (CRM)-based region of interest (ROI) localization; (v) analyzing inter-reader variability and ensemble localization performance using Simultaneous Truth and Performance Level Estimation (STAPLE) methods. We observe that: (i) ensemble approaches improved classification and localization performance; and, (ii) inter-reader variability and performance level assessment helped guide algorithm design and parameter optimization. To the best of our knowledge, this is the first study to construct ensembles, perform ensemble-based disease ROI localization, and analyze inter-reader variability and algorithm performance for COVID-19 detection in CXRs.