Explainable Lung Disease Detection on Chest X-Ray Images Using Gradient-Weighted Class Activation Mapping

Early detection of respiratory diseases such as COVID-19 and pneumonia is essential to speed up treatment and prevent severe complications. This study develops a chest X-ray image classification system using a Convolutional Neural Network (CNN), specifically a VGG16-based architecture, to distinguish between COVID-19, pneumonia, and normal lungs. The model is trained on a balanced dataset of 3,000 images (1,000 per class), with preprocessing steps including resizing, normalization, train–validation splitting, and data augmentation. Hyperparameter exploration identified an optimal configuration with a learning rate of 0.001, 50 epochs, and a batch size of 32, achieving an accuracy of 96.33%. Performance was evaluated using accuracy, precision, recall, F1-score, and a confusion matrix. To enhance transparency and interpretability in a clinical context, Gradient-Weighted Class Activation Mapping (Grad-CAM) was applied to visualize the regions of the CXR images that contributed most to the model’s decisions. The main contribution lies in integrating Grad-CAM with multi-class CXR classification to provide explainable AI support for lung disease diagnosis.