id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-285872-rnayrws3	Elgendi, Mohamed	The Performance of Deep Neural Networks in Differentiating Chest X-Rays of COVID-19 Patients From Other Bacterial and Viral Pneumonias	2020-08-18		.txt	text/plain	3453	188	51	Our results show that DarkNet-19 is the optimal pre-trained neural network for the detection of radiographic features of COVID-19 pneumonia, scoring an overall accuracy of 94.28% over 5,854 X-ray images. Sethy and Behera (8) explored 10 different pre-trained neural networks, reporting an accuracy of 93% on a balanced dataset, for detecting COVID-19 on X-ray images. Our study aims to determine the optimal learning method, by investigating different types of pre-trained networks on a balanced dataset, for COVID-19 testing. To determine the optimal existing pre-trained neural network for the detection of COVID-19, we used the CoronaHack-Chest X-Ray-Dataset. Inception-v3 and ShuffleNet achieved an overall validation accuracy below 90% suggesting that these neural networks are not robust enough for detecting COVID-19 compared to, for example, ResNet-50 and DarkNet-19. After investigating 17 different pre-trained neural networks, our results showed that DarkNet-19 is the optimal pre-trained deep learning network for detection of imaging patterns of COVID-19 pneumonia on chest radiographs.	./cache/cord-285872-rnayrws3.txt	./txt/cord-285872-rnayrws3.txt