Enhancing COVID-19 Classification in Chest X-ray Images through CNN-based Model Optimization Techniques

This paper explores the enhancement of COVID-19 classification in chest X-ray images through the integration of advanced model optimization techniques. Utilizing a dataset compiled by researchers from Qatar University, Doha, Qatar, and the University of Dhaka, Bangladesh, the study employs a Convolutional Neural Network (CNN) architecture, specifically leveraging transfer learning with the VGG network. The dataset, named COVID-QU-Ex, contains 33,920 chest X-ray images, including 11,956 COVID-19 positive cases, 11,263 Non-COVID infections (Viral or Bacterial Pneumonia), and 10,701 Normal cases. The model incorporates Global Average Pooling (GAP) to reduce overfitting and improve generalization by replacing fully connected layers. Additionally, Spatial Transformer Networks (STN) are utilized to enhance the network's spatial invariance, allowing it to better handle variations in the input images. The outcomes demonstrate a significant improvement in classification accuracy, with the optimized model achieving superior performance compared to baseline approaches. This research highlights the potential of these techniques in improving the reliability and effectiveness of COVID-19 detection in medical imaging.