Detecting Covid-19 in Chest X-Rays using Transfer Learning with VGG16

Abstract

Covid-19 is a novel epidemic that has hugely impacted countries worldwide [13]; and for which there is a need for quick and accurate screening methods. Current testing methods include the reverse transcription-polymerase chain reaction test and medical diagnosis using computed tomography scans. Both of these require expensive technologies as well as highly-trained practitioners and thus are in short supply [18]. Less developed countries and overloaded hospitals have increased the demand for cheap, easy and accurate screening methods [4]. X-ray devices are now cheap, portable and easy to use; there are few professionals, however, who are capable of manually identifying Covid-19 from a chest x-ray. We suggest implementing a machine learning model that incorporates transfer learning to automatically detect Covid-19 from chest x-ray images. The suggested model is built on top of the VGG16 architecture and pre-trained ImageNet weights. Compared with the VGG19, Inception-V3, Inception-ResNet, Xception, RestNet152-V2, and DenseNet201 models, the VGG16 model achieved the highest testing accuracy of 98% on 10 epochs as well as high positive-class accuracy. Gradient-weighted class activation mapping (Grad-CAM) was also applied to detect the regions that have a greater impact on the model classification decision.