Explainable Lung Disease Classification from Chest X-Ray Images Utilizing Deep Learning and XAI
Tanzina Taher Ifty, Saleh Ahmed Shafin, Shoeb Mohammad Shahriar, Tashfia Towhid
TL;DR
The study tackles automated, multi-class lung disease diagnosis from chest X-ray images across five classes using a broad range of deep learning approaches, including CNNs, hybrids, ensembles, transformers, and Big Transfer, augmented by explainable AI techniques. A comprehensive pipeline with data augmentation, transfer learning, and stratified K-fold cross-validation demonstrates that the Xception model, when fine-tuned with 5-fold cross-validation, achieves 96.21% accuracy, while ensembles and transformer-based models provide competitive performance. Grad-CAM and LIME are applied to offer visual explanations of predictions, supporting clinical trust and interpretability. The work suggests practical pathways toward robust, explainable chest X-ray classifiers with potential for real-time clinical deployment and highlights directions for future improvements through multimodal data and SHAP-based explanations.
Abstract
Lung diseases remain a critical global health concern, and it's crucial to have accurate and quick ways to diagnose them. This work focuses on classifying different lung diseases into five groups: viral pneumonia, bacterial pneumonia, COVID, tuberculosis, and normal lungs. Employing advanced deep learning techniques, we explore a diverse range of models including CNN, hybrid models, ensembles, transformers, and Big Transfer. The research encompasses comprehensive methodologies such as hyperparameter tuning, stratified k-fold cross-validation, and transfer learning with fine-tuning.Remarkably, our findings reveal that the Xception model, fine-tuned through 5-fold cross-validation, achieves the highest accuracy of 96.21\%. This success shows that our methods work well in accurately identifying different lung diseases. The exploration of explainable artificial intelligence (XAI) methodologies further enhances our understanding of the decision-making processes employed by these models, contributing to increased trust in their clinical applications.