COVID-SEGNET: Diagnosis of Covid-19 Cases on Radiological Images using Mask R-CNN

摘要

The novel coronavirus (COVID-2019) pandemic has caused a catastrophic effect on health and global economy. Early screening and diagnosis of COVID-19 pneumonia are the critical steps to stop the further spread of the virus. The most common standard for confirming the virus relies on RT-PCR tests. This method generates false-negative results if there is limited viral load. Recent radiological findings suggest that the distinct distribution of ground-glass opacities (GGOs), which are found on certain parts of lungs, can determine the status of the infection among patients. As a complement to RT-PCR, Computed tomography (CT) can be used for diagnosing COVID-19. In this study, the authors have described a Mask R-CNN (region-based convolution neural network) approach for the detection of the ground glass opacities (GGOs) in chest CT images of COVID-19 infected persons. The proposed approach provides an accuracy of 98.25% during instance segmentation. Therefore, the authors believe this proposed method will aid health professionals to fasten the screening and validation of the initial assessment towards COVID-19 patients.