Decision Support System for Lung Health Assessment and Pneumonia Detection from Chest X-Ray Images

Pneumonia remains one of the most critical global health challenges, where the most commonly used diagnostic method is analyzing chest X-ray images. In this study, we analyzed three separate datasets comprising more than 20000 chest X-ray images. While state-of-the-art works remained limited to model development using machine learning and deep learning algorithms, we have developed a web-based decision support system (DSS) that can predict the input chest X-ray image in a resource-limited setting. This attribute is especially important where access to expert radiologists is scarce and timely diagnosis is critical, and where the number of doctors in clinics is limited, but patients need early diagnosis. The DSS is a two-step deep learning-based system that leverages transfer learning with the pre-trained VGG convolutional neural networks for automated lung health assessment and pneumonia detection from chest Xray images. Initial model classification and performance analysis were done using the VGG-16 and VGG-19 models. The VGG16 achieved 97.73% accuracy in detecting pneumonia, while the VGG-19 achieved 94.48% in detecting healthy lungs. © 2026 IEEE.