Automated detection of COVID-19 based on transfer learning

The world has recently experienced unprecedented disruptions due to the COVID-19 pandemic, which have greatly impacted daily life. Deep Learning (DL) is a branch of Artificial Intelligence (AI) that has seen significant growth in recent years, and its features could be useful in the fight against COVID-19. By leveraging these features, public health efforts could be better supported. In this research, we propose a method for detecting COVID-19 positive patients using chest X-ray images. Our method employs pre-trained deep neural networks, specifically the DenseNet-169 and ResNet-50 architectures. For each architecture, we kept the basic model and replaced the border layers with Dense layers. We used 2Dense in the first iteration and 5Dense in the second iteration. Our results show that Transfer Learning (TL) is a useful technique for detecting COVID-19 cases. The DenseNet-169 + 2Dense, DenseNet-169 + 5Dense, and using the ELU function achieved the highest accuracy value of 90.04%.