Covid CT Net: A Transfer Learning Approach for Identifying Corona Virus from CT Scans


Smaranjit Ghose and Suhrid Datta, SRM Institute of Science and Technology, India


The pandemic of COVID-19 has been rapidly spreading across the globe since it first surfaced in the Wuhan province of China. Several governments are forced to have nationwide lockdowns due to the progressive increase in a daily number of cases. The hospitals and other medical facilities are facing difficulties to cope with the overwhelming number of patients they can provide support due to the shortage in the number of required medical professionals and resources for meeting this demand. While the vaccine to cure this disease is still on the way, early diagnosis of patients and putting them in quarantine has become a cumbersome task too. In this study, we propose to build an artificial intelligence-based system for classifying patients as COVID-19 positive or negative within a few seconds by using their chest CT Scans. We use a transfer learning approach to build our classifier model using a dataset obtained from openly available sources. This work is meant to assist medical professionals in saving hours of their time for the diagnosis of the Coronavirus using chest radiographs and not intended to be the sole way of diagnosis.


COVID-19, Deep Learning, CT Scans, Deep Convolutional Neural Networks, computer tomography scans.

Full Text  Volume 10, Number 11