Automated Pneumonia Detection from Chest X-Ray Images Using Deep Convolutional Neural Networks
Orestis Papadimitriou, Athanasios Kanavos, Manolis Μaragoudakis
Abstract
Recent advancements in deep learning have shown promising results in various clinical image analysis tasks. Among the most commonly performed radiological examinations, chest radiographs play a crucial role and have been extensively investigated for various applications. The availability of large, publicly accessible chest X-ray datasets in recent years has sparked research interest. Pneumonia diagnosis typically relies on the analysis of X-ray radiographs by highly trained experts, a process that can be time-consuming and prone to disagreements among radiologists. In this study, we utilized a dataset of 5,856 high-resolution frontal-view chest X-ray images for training, validation, and testing of our model. The proposed model achieved an impressive accuracy of 96.6%, sensitivity of 98.1%, specificity of 92.4% and precision of 97.2%. These results demonstrate the excellent performance of the model in accurately identifying pneumonia cases and normal cases.