Authors: Princi Saini, Mridula
Abstract: Pneumonia is a serious respiratory disease that affects millions of people worldwide and remains a major cause of mortality, particularly among children and elderly individuals. Early and accurate diagnosis is essential for effective treatment; however, manual interpretation of chest X-ray images can be time-consuming and subject to human error. Recent advances in deep learning, especially Convolutional Neural Networks (CNNs), have demonstrated significant potential in automating medical image analysis and supporting clinical decision-making. This study proposes a compact and computationally efficient CNN-based framework for the automatic detection of pneumonia from chest X-ray images. The proposed model is designed to achieve a balance between diagnostic accuracy and computational efficiency, making it suitable for deployment in resource-constrained healthcare environments. A publicly available dataset consisting of 5,863 chest X-ray images, including both pneumonia and normal cases, was utilized for model development and evaluation. To improve data quality and model generalization, preprocessing techniques such as image resizing, normalization, denoising, contrast enhancement, and data augmentation were applied. The CNN architecture comprises four convolutional layers with ReLU activation, max-pooling operations, dropout regularization, and fully connected layers for binary classification. The model was trained using the Adam optimization algorithm and evaluated using multiple performance metrics, including accuracy, precision, recall, specificity, F1-score, and Area Under the Receiver Operating Characteristic Curve (AUC-ROC). Experimental results demonstrate that the proposed framework achieves an accuracy of 95.8%, precision of 96.5%, recall of 94.9%, specificity of 96.2%, F1-score of 95.7%, and an AUC of 0.97. These results indicate strong classification capability and reliable discrimination between pneumonia and normal chest X-ray images. A comparative analysis with established deep learning architectures such as VGG16, ResNet50, InceptionV3, and EfficientNet-B0 shows that the proposed model delivers competitive performance while requiring significantly lower computational resources. Its lightweight architecture enables faster training and inference, making it well-suited for real-time clinical applications, telemedicine platforms, and healthcare facilities with limited computational infrastructure.
