Chest X-Ray Pneumonia Detection Using Federated Learning : ^*Privacy-Preserving Deep Learning for Medical Image Classification

Pneumonia is a serious and potentially life-threatening infection that remains a leading cause of morbidity and mortality worldwide [26]. Early diagnosis is critical for improving patient outcomes, and chest X-rays are the most widely used imaging modality for pneumonia detection. Recent advances in deep learning (DL), particularly convolutional neural networks (CNNs), have demonstrated radiologist-level performance in automating pneumonia diagnosis from chest radiographs [5], [26], [27]. However, conventional deep learning methods depend on centralized data aggregation, which raises significant challenges in the healthcare domain due to strict data privacy regulations such as HIPAA and GDPR [6], [10]. Centralized approaches also introduce vulnerabilities like single points of failure and data breaches, further limiting scalability in real-world multiinstitutional settings. To overcome these limitations, this study investigates the application of Federated Learning (FL) for pneumonia detection using chest X-ray images. FL enables multiple healthcare institutions to collaboratively train machine learning models without sharing sensitive patient data [6], [7], [10]. By training CNNbased models locally on institution-specific data and aggregating only model parameters using the Federated Averaging (FedAvg) algorithm [28], our framework ensures patient confidentiality while maintaining diagnostic performance. The final federated model achieved high accuracy (88.5