Automated Detection of Covid-19Waves with Computerized Tomography Scan Using Deep Learning

ABSTRACT

Covid-19 is a newly found corona virus that causes an infectious disease. An accurate diagnosis of several waves in Covid-19 is still a tremendous confront due to the difficulties of marking infection areas, and it is an emergency and important for worldwide in 2020 and still now. There is almost no difference between common pneumonia and other viral pneumonia using CT scanned images, so false-negative images may be obtained. An ensemble of deep multi-instance learning (DMIL), train a blotch-level classifier and view the chest CT images as a bag of samples to avoid false negative. Mask R-CNN is used to train an image-level classifier that labels input image as common pneumonia or Covid pneumonia. These Ensemble models of DMIL with Mask R-CNN show an accuracy of 98.96%. These advantages make our model an efficient tool in the screening of Covid-19. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023.