Lightweight Chest X-Ray Classification for Pneumonia and Tuberculosis Using MobileNet with Explainable AI

Chest X-rays are a low-radiation, cost-effective alternative for the diagnosis of respiratory diseases and are particularly valuable as a first-line screening tool. Our solution uses deep learning to classify chest radiographs into three categories: pneumonia, tuberculosis, and healthy (without disease). The Mobile Net model has a lightweight architecture perfectly suited for smartphones which in turn is useful as our aim here is to make this tool accessible, easy to use and helpful for health care workers in remote areas. In order to improve trust and reliability, we have incorporated explainable AI (XAI), which provides detailed insights of how the model arrived at a particular conclusion/decision. While our model does maintain a balance between speed and accuracy, we have also tested it against other resource-intensive models such as ResNet and DenseNet. The results show that MobileNet achieves comparable accuracy while significantly reducing computational demands. This was designed as an early diagnostic aid and provides a feasible, affordable alternative to CT scans, especially for regions with minimal access to advanced medical infrastructure. Its aim is to combine innovation and accessibility in one place to bring about timely and effective healthcare delivery.