A Comprehensive Study on Classification of COVID-19 on Computed
Tomography with Pretrained Convolutional Neural Networks
Abstract
This study presents an investigation on sixteen pretrained CNNs for
classification of COVID-19 using a large public database of CT scans
collected from COVID-19 patients and non-COVID-19 subjects. The results
update CNNs that achieve very high performance on the classification
task and discover that implementation of transfer learning with direct
input of whole image slices and without the use of data augmentation
provide better classification results than the use of data augmentation.
{\it Conclusions:} The findings alleviate the task of
data augmentation and manual extraction of regions of interest on CT
images, which are adopted by current implementation of deep-learning
models, and can facilitate the rapid deployment of AI tools to contain
the spread of the coronavirus disease.