DEEP LEARNING BASED PNEUMONIA DETECTION SYSTEM USING CONVOLUTIONAL NEUTRAL NETWORK

ABSTRACT Pneumonia remains one of the leading causes of death worldwide, particularly in developing countries where access to timely and accurate diagnosis is limited. Traditional diagnostic approaches such as chest X-ray interpretation are often time-consuming and heavily dependent on the expertise of radiologists. This study presents the development and evaluation of a Convolutional Neural Network (CNN)-based system for the automatic detection and classification of pneumonia from chest X-ray images. The model was trained and tested on a publicly available dataset containing two classes, Normal and Pneumonia. The dataset underwent preprocessing, including image resizing, normalization, and augmentation, to enhance learning efficiency and reduce overfitting. The developed CNN model achieved an accuracy of 84.62% and aROC-AUC score of 0.918, demonstrating strong classification performance. The recall for pneumonia (0.8872) was higher than that for normal cases (0.7778), indicating the model’s effectiveness in detecting pneumonia cases while minimizing false negatives, an important aspect in clinical practice. Precision and F1-scores for pneumonia were 0.87 and 0.88, respectively, reflecting the model’s robustness and balance between sensitivity and specificity. Overall, the results show that the proposed CNN model performs competitively compared to existing methods in the literature and can serve as areliable diagnostic support tool for pneumonia detection. The study contributes to the growing field of AI-assisted medical imaging, highlighting the potential of deep learning to enhance diagnostic accuracy and reduce the workload on healthcare professionals, particularly in resource-limited environments.