Interpretable artificial intelligence framework for COVID‑19 screening on chest X‑rays

TLDR

COVID‑19 has caused a global healthcare crisis, and limited testing supplies have spurred research into alternative high‑sensitivity diagnostics such as AI‑assisted chest X‑ray interpretation. This study introduces an interpretable AI framework evaluated by expert radiologists based on the focus of its attention maps on diagnostically relevant regions. The framework employs transfer‑learning to adapt deep models to scarce COVID‑19 X‑ray data, enabling specialized classifiers to converge with limited samples. The transfer‑learning approach attains a perfect area‑under‑curve of 1.0 on a 5‑fold cross‑validated binary classification of COVID‑19 versus non‑COVID‑19 chest X‑rays.

Abstract

COVID-19 has led to an unprecedented healthcare crisis with millions of infected people across the globe often pushing infrastructures, healthcare workers and entire economies beyond their limits. The scarcity of testing kits, even in developed countries, has led to extensive research efforts towards alternative solutions with high sensitivity. Chest radiological imaging paired with artificial intelligence (AI) can offer significant advantages in diagnosis of novel coronavirus infected patients. To this end, transfer learning techniques are used for overcoming the limitations emanating from the lack of relevant big datasets, enabling specialized models to converge on limited data, as in the case of X‑rays of COVID‑19 patients. In this study, we present an interpretable AI framework assessed by expert radiologists on the basis on how well the attention maps focus on the diagnostically‑relevant image regions. The proposed transfer learning methodology achieves an overall area under the curve of 1 for a binary classification problem across a 5‑fold training/testing dataset.