Robust CNN-Based COVID-19 Detection from Chest X-rays Using ResNet-50 with Grad-CAM Visualization

The COVID-19 pandemic has shown the urgent need for quick and reliable diagnostic methods to support standard RT-PCR testing. In this study, we introduce a deep learning framework to detect COVID-19 from chest X-ray images, based on the ResNet-50 architecture and improved with Gradient-weighted Class Activation Mapping (Grad-CAM) for better interpretability. The model was trained on a balanced dataset of 798 chest X-rays and achieved an accuracy of 97.5% on internal test data. External validation using three independent datasets showed strong generalization, with accuracy between 81.00% and 94.40%. Grad-CAM visualizations often highlighted lung areas connected to known radiological patterns of COVID-19, closely matching clinical observations. This combination of high accuracy and clear decision-making supports the promise of the proposed approach as a useful tool to aid radiological diagnosis, especially in healthcare settings with limited access to expert interpretation.