Deep Learning Approaches for Multi-Class Lung Disease Classification Utilizing EfficientNet Architectures and CXR Images

Lung diseases such as COVID-19, lung opacity, and viral pneumonia pose great challenges to healthcare organizations globally. It is very important to diagnose these diseases in the early stages. The conventional approaches for diagnosing lung diseases are time-consuming and hence require the use of modern technology. For this purpose, Chest X-ray (CXR) images are often used. In this study, the EfficientNet-based deep learning models have been used to classify lung diseases using the CXR images. In this study, we evaluated the performance of eight EfficientNet models(EfficientNetB0 to EfficientNetB7) to identify normal, opacity, COVID-19, and viral pneumonia. EfficientNetB7 achieved the highest accuracy of 99.42%. The analysis shows that the EfficientNetB7 network has a higher capacity for classifying different lung diseases than other models and can enhance the diagnostic effectiveness and repeatability of the diagnosis. This work underscores the possibility of deep learning frameworks on medical imaging for improved patient and health systems management.