Explaining COVID-19 and Thoracic Pathology Model Predictions by Identifying Informative Input Features
This work addresses the need for interpretable AI in medical imaging to support clinical decision-making, though it is incremental as it builds on existing feature attribution methods.
The authors tackled the problem of making neural network predictions on chest X-rays interpretable for clinical trust by proposing Inverse IBA to identify all informative input regions, along with extensions for regression and multi-layer analysis, achieving evaluation with human-centric and independent metrics on NIH Chest X-ray8 and BrixIA datasets.
Neural networks have demonstrated remarkable performance in classification and regression tasks on chest X-rays. In order to establish trust in the clinical routine, the networks' prediction mechanism needs to be interpretable. One principal approach to interpretation is feature attribution. Feature attribution methods identify the importance of input features for the output prediction. Building on Information Bottleneck Attribution (IBA) method, for each prediction we identify the chest X-ray regions that have high mutual information with the network's output. Original IBA identifies input regions that have sufficient predictive information. We propose Inverse IBA to identify all informative regions. Thus all predictive cues for pathologies are highlighted on the X-rays, a desirable property for chest X-ray diagnosis. Moreover, we propose Regression IBA for explaining regression models. Using Regression IBA we observe that a model trained on cumulative severity score labels implicitly learns the severity of different X-ray regions. Finally, we propose Multi-layer IBA to generate higher resolution and more detailed attribution/saliency maps. We evaluate our methods using both human-centric (ground-truth-based) interpretability metrics, and human-independent feature importance metrics on NIH Chest X-ray8 and BrixIA datasets. The Code is publicly available.