ABSTRACT
Coronavirus is highly infectious, and millions of individuals are infected by COVID-19 since February 2020. Covid-19 pandemic has caused the demise of approx. 4.5 million people already. Researchers have developed various deep learning-based systems/solutions to diagnose the COVID-19 by using medical images such as Computed Tomography (CT) scans, and X-Ray images. CT scans are standard examination methods employed for COVID-19 detection. Deep learning-based methods or techniques achieve a promising result in medical image analysis and automatic diagnosis of COVID-19. This paper reviews and summarizes the existing deep learning-based techniques designed for COVID-19 diagnostic from CT scans and X-rays and lists the datasets, that can be helpful in future research and automatic detection of Covid-19. Additionally, for fair comparison of these published techniques, we implement them on same dataset. Additional explorations for justifiable and explainable results are obligatory for more transparent, robust, and precise predictions of COVID-19. Despite the promising outcomes in existing techniques, there is crucial need for comprehensive, public, and diverse datasets.
