XAI-Guided Analysis of Residual Networks for Interpretable Pneumonia Detection in Paediatric Chest X-rays

TL;DR

This paper presents an interpretable deep learning approach using Residual Networks and Bayesian Grad-CAM for accurate and explainable pediatric pneumonia detection in chest X-rays, emphasizing clinical applicability.

Contribution

It introduces an interpretable ResNet-based model with uncertainty quantification for pneumonia diagnosis, advancing clinical AI transparency and reliability.

Findings

Achieved 95.94% accuracy on pediatric X-ray dataset

Provided spatial explanations with uncertainty measures

Demonstrated high interpretability alongside strong performance

Abstract

Pneumonia remains one of the leading causes of death among children worldwide, underscoring a critical need for fast and accurate diagnostic tools. In this paper, we propose an interpretable deep learning model on Residual Networks (ResNets) for automatically diagnosing paediatric pneumonia on chest X-rays. We enhance interpretability through Bayesian Gradient-weighted Class Activation Mapping (BayesGrad-CAM), which quantifies uncertainty in visual explanations, and which offers spatial locations accountable for the decision-making process of the model. Our ResNet-50 model, trained on a large paediatric chest X-rays dataset, achieves high classification accuracy (95.94%), AUC-ROC (98.91%), and Cohen's Kappa (0.913), accompanied by clinically meaningful visual explanations. Our findings demonstrate that high performance and interpretability are not only achievable but critical for clinical AI deployment.