A Revolutionizing Pneumonia Diagnostics: A Systematic Review of Deep Learning Applications in Chest Radiography

Revolutionizing Pneumonia Diagnostics: A Systematic Review of Deep Learning Applications in Chest Radiography

Abstract

Pneumonia remains a leading cause of global morbidity and mortality, particularly in vulnerable populations such as children and the elderly. Early and accurate diagnosis is crucial, but is often hampered by the scarcity of trained radiologists and the inherent subjectivity in chest X-ray (CXR) image interpretation. In recent years, the rapid advancement of deep learning (DL), especially Convolutional Neural Networks (CNNs), has presented transformative solutions to these challenges. This systematic literature review (SLR) thoroughly analyzes twelve primary studies exploring the application of DL models for pneumonia detection from CXR images. Through a rigorous SLR methodology, these studies were identified, evaluated, and synthesized. Results indicate that DL models, whether built from scratch or based on transfer learning, are capable of achieving very high diagnostic accuracy, often surpassing or matching the performance of human radiologists. The discussion highlights key findings, addresses the formulated research questions, and identifies existing study limitations. The conclusion affirms the significant potential of DL in revolutionizing pneumonia diagnostics, and outlines future research directions to enhance the generalization, interpretability, and implementation of DL models in daily clinical practice.