id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-317016-codk0by1	Trivizakis, Eleftherios	Advancing COVID-19 differentiation with a robust preprocessing and integration of multi-institutional open-repository computer tomography datasets for deep learning analysis	2020-09-11		.txt	text/plain	3269	167	45	Tsiknakis et al (17) proposed an Interpretable Convolutional Neural Network (CNN) based on transfer learning for predicting COVID-19 against viral and bacterial pneumonia and normal cases based on more than 400 X-ray images, achieving an area under curve (AUC) of 100%. Some key innovations of this study can be summarized in the integration of multi-institutional and open-access data from a variety of scanners and imaging protocols retrieved from online repositories in formats such as DICOM or Portable network Graphics (png), the development of a deep learning lung segmentation model for multiple CT window settings and finally a deep learning model for differentiating COVID-19 from CAP. This study introduces a state-of-the-art deep learning model for lung segmentation on slices with a variety of CT window settings (DSC 99.6%) and an image analysis deep model trained with multi-institutional data for differentiating COVID-19 from CAP (AUC 96.1%).	./cache/cord-317016-codk0by1.txt	./txt/cord-317016-codk0by1.txt