Comparative analysis of deep learning models for chest X-ray image classification

This study examines the efficacy of deep learning models in classifying chest X-ray images, particularly enhancing diagnostic precision for thoracic conditions. The aim is to evaluate and compare the performance of several advanced deep learning architectures—ResNet50, DenseNet121, Efficient Net, and Mobile Net—leveraging the NIH Chest X-ray dataset. The methodology employs a rigorous evaluation framework using metrics including precision, recall, F1-score, and accuracy, along- side interpretability methods such as Grad-CAM to elucidate decision-making processes in model predictions. The primary contribution of this work lies in determining the optimal model for clinical deployment and offering approaches to tackle issues like computational demands and dataset imbalances. By addressing these challenges, the research advances toward integrating artificial intelligence into medical workflows, contributing to the progression of AI-enhanced diagnostics to address global healthcare disparities and improve patient care outcomes.