Lung Disease Detection with Explainable AI and Deep Learning: A Comprehensive Study

Lung diseases, which include COVID-19, pneumonia, and tuberculosis, are significant global public health concerns and have the potential for serious health outcomes if not identified in time. This work proposes an explainable AI model that uses deep learning techniques to identify and predict different types of lung diseases from medical image data, such as chest X-rays and CT scans. This shall further allow a proper classification model by using convolutional neural networks to classify properly and make the proper diagnosis of the input based on the medical image while using the techniques of explanation like Grad-CAM, SHAP, and LIME to make the prediction understandable, thereby developing confidence of healthcare professionals in applying such models into the medical practices. It thus explores improvement of diagnostics with this improved model against classic issues facing the domain, data imbalance, and demand for explainability in AI-driven diagnosis. Results of this study show improvement in the detection of lung diseases while emphasizing the importance of transparency, which could result in improved patient outcomes and diagnostic reliability.