An AI-based System for Pneumonia Detection in Chest X-Rays Using Deep Learning Models

Pneumonia remains a significant global health concern, necessitating rapid and accurate diagnostic tools. This study presents an AI-based system for pneumonia detection in chest X-rays using deep learning models. The research emphasizes enhancing diagnostic accuracy through advanced image processing techniques while maintaining clinical applicability. Deep learning has demonstrated strong potential in medical image analysis, particularly in identifying pulmonary abnormalities in radiographic images. The proposed system incorporates pre-processing techniques, such as multi-CLAHE, to improve image contrast and highlight infection regions. Additionally, synthetic data generation using Conditional CycleGAN mitigates dataset limitations, enhancing the model’s ability to detect early-stage pneumonia. Three deep learning models-VGG16, VGG19, and ResNet50-were fine-tuned and evaluated. Among these, ResNet50 achieved the highest accuracy of 95.2%, while VGG19 provided a favourable balance between performance and computational efficiency. Image enhancement and synthetic data increased recall by 6%, demonstrating improved reliability. These results indicate that AI-assisted diagnosis can enhance pneumonia detection and provide a viable solution for clinical deployment. The system includes a web-based interface to ensure usability in healthcare settings with limited radiological resources. Future work will explore attention mechanisms and larger datasets to further improve accuracy.