Pneumonia Image Classification Using DenseNet Architecture

Abstract

Pulmonary diseases, including pneumonia, represent a significant health challenge and are often diagnosed using X-rays. This study investigates the effectiveness of artificial intelligence (AI) in enhancing the diagnostic capabilities of X-ray imaging. Using Python and the PyTorch framework, we developed and trained several deep learning models based on DenseNet architectures (DenseNet121, DenseNet169, and DenseNet201) on a dataset comprising 5856 annotated X-ray images classified into two categories: Normal (Healthy) and Pneumonia. Each model was evaluated on its ability to classify images with metrics including binary accuracy, sensitivity, and specificity. The results demonstrated accuracy rates of 92% for Normal and 97% for Pneumonia. The models also showed significant improvements in diagnostic accuracy and reduced time for disease detection compared to traditional methods. This study underscores the potential of integrating convolutional neural networks (CNNs) with medical imaging to enhance diagnostic precision and support clinical decision-making in the management of pulmonary diseases. Further research is encouraged to refine these models and explore their application in other medical imaging domains.