Development and Evaluation of an AI-Driven Convolutional Neural Network (CNN) Model for Automated Pneumonia Detection Using Chest X-Ray Image

Pneumonia remains a leading cause of mortality globally, especially in low-resource settings where timely and accurate diagnosis is often not readily available. Chest X-ray (CXR) imaging is a critical tool for pneumonia diagnosis; however, its interpretation depends heavily on the expertise of radiologists, which can introduce variability. This study presents a computer-aided diagnosis (CAD) system based on a convolutional neural network (CNN) designed to automatically classify CXR images as either normal or indicative of pneumonia. The CNN model was evaluated across 10, 20, 50, and 100 epochs, achieving a peak accuracy of 92%, a precision-recall score of 95%, and an AUC of 96%. Notably, the model demonstrated strong performance without the need for advanced preprocessing techniques such as Dynamic Histogram Equalization (DHE) or Histogram Equalization (HE). Despite the simplicity of the approach, the model maintained competitive performance, with a precision of 93% observed at 50 epochs. Further analysis through precision-recall and ROC curves confirmed the model's consistent and reliable performance across all evaluated epochs. This CNN-based approach provides an efficient solution for pneumonia detection in clinical settings, particularly in regions with limited access to radiologists. It holds significant potential to improve diagnostic accuracy, particularly in under-resourced areas.