Exploring the Efficacy of Deep CNN Algorithms for Pulmonary Infections Detection using X-rays: Case Study COVID-19

This paper explores the efficacy of various deep convolutional neural network methods for detecting and diagnosing COVID-19 induced pulmonary infections using X-ray images. The study involves training and testing different deep learning models on a consistent dataset. Model performance is evaluated based on accuracy, training time, and complexity, providing valuable insights into their suitability for this critical medical application. To ensure the robustness of the trained models, typical challenges arising from improper dataset handling, such as overfitting and underfitting, are addressed and mitigated through appropriate dataset splitting techniques. This approach enhances the reliability of the results and improves the generalizability of the models. The results show a 96% accuracy using ResNet50V2 model. Furthermore, a diagnosis App based on ResNet50V2 model is developed. This application serves as a practical diagnosis tool, facilitating the early detection and diagnosis of COVID-19 induced pulmonary infections.