id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-336178-k8za0doe	Afshar, Parnian	COVID-CAPS: A Capsule Network-based Framework for Identification of COVID-19 cases from X-ray Images	2020-09-16		.txt	text/plain	4233	235	54	To potentially and further improve diagnosis capabilities of the COVID-CAPS, pre-training and transfer learning are utilized based on a new dataset constructed from an external dataset of X-ray images. In a study by Wang and Wong [9] , a CNN model is first pre-trained on the ImageNet dataset [10] , followed by finetuning using a dataset of X-ray images to classify subjects as normal, bacterial, non-COVID-19 viral, and COVID-19 viral infection, achieving an overall accuracy of 83.5%. In a similar study by Sethy and Behera [11] , different CNN models are trained on X-ray images, followed by a Support Vector Machine (SVM) classifier to identify positive COVID-19 cases, reaching an accuracy of 95.38%. In summary, pre-training with an external dataset of X-ray images further improved accuracy of COVID-CAPS to 98.3%, specificity to 98.6%, and AUC to 0.97, however, with a lower sensitivity of 80%.	./cache/cord-336178-k8za0doe.txt	./txt/cord-336178-k8za0doe.txt