Chest X-ray data have been proven to be extremely promising for screening COVID-19 patients, particularly for addressing overcapacity in emergency departments and urgent care centers. Deep-learning (DL) approaches in artificial intelligence (AI) have a leading role as high-performance classifiers in illness diagnosis utilizing chest X-rays.
Given that many new DL models have been established for this purpose, the goal of this study is to look into the fine-tuning of pre-trained convolutional neural networks (CNNs) for COVID-19 classification using chest X-rays. If fine-tuned pre-trained CNNs can produce comparable or better classification results than other more advanced CNNs, then AI-based systems for identifying COVID-19 using chest X-ray data can be deployed more quickly and cost-effectively. Three pre-trained CNNs, AlexNet, GoogleNet, and SqueezeNet, were chosen and fine-tuned without data augmentation to perform two and three-class classification tasks on three public chest X-ray databases.