Automatic Prognosis of COVID-19 from CT Scan using Super-convergence CNN Algorithm

Abstract

—Due to the high incident rate of COVID-19, the number of suspected patients needing diagnosis presents over whelming pressure on hospital and health management systems such that the disease outbreak elapsed into a global pandemic. More so, the infected patients present a higher risk of being infected to the health workers because once a patient is positive of the virus, the progress of recovery or deterioration needs to be monitored by medical experts and other health workers, which eventually exposes them to the infection. In this paper, we present an automatic prognosis of COVID-19 from a CT scan using deep CNN. The models were trained using a super-convergence discriminative fine-tuning algorithm, which uses a layer-specific learning rate to fine-tune a deep CNN model; this learning rate is increased or decreased per iteration to avoid the saddle-point problem and achieve the best performance within few training epochs. The best performance results were obtained as 98.57% accuracy, 98.59% precision and 98.55% recall rate.