Background: The COVID-19 pandemic continues to have a devastating impact on the worldwide population’s health and welfare. A key measure that is taken in combating COVID-19 is effectively screening infected patients. A vital screening process is examination through chest radiology. Initial studies have shown irregularities in the chest radiography images of patients specific to those suffering from the COVID-19 infection. Motivated and inspired by the research community’s open source efforts, this study introduces a dilated bi-branched convoluted neural network (CNN) architecture called VGG-COVIDNet, tailored to detect COVID-19 cases from chest X-ray (CXR) images.
Results: The simulation results show that the proposed architecture yielded the highest accuracy and produced the highest sensitivity compared to state-of-the-art architectures. The proposed architecture’s accuracy and sensitivity are 96.5% and 96%, respectively, for each of the infection types.
Conclusions: We applied VGG-CovidNet, a VGG-16, and dilated convolution-based bi-branched architecture for classifying different types of infectious images into corresponding classes. The front end of the VGG-CovidNet is integrated with the first 10 layers of the VGG-16 model. In these layers, the convolution is reduced to two instead of three to alleviate the model’s computation complexity. Furthermore, the back end of the VGG-CovidNet consists of six parallel convolutional layers having different strides of convolutional kernel. Using the dilated convolution, we can capture the spatial information of the feature maps and help reduce the complexity of the model.