Screening and Classification of Covid-19 from Lung Images using Deep Learning Models

Abstract

Abstract: The world is facing human mankind issues which made people quarantine. Real-Time Reverse Transcription Polymerase Chain Reaction (RT-PCR) detection of viral RNA from nasopharyngeal swab has a relatively low positive rate for the early-stage detection of COVID-19. The Computed Tomography is the first imaging technique that plays an important role in the early diagnosis of COVID-19 and their manifestations had their characteristics that are different from other non-COVID diseases like Influenza-A viral pneumonia. So this survey is trained to determine an early screening model to distinguish COVID-19 from healthy cases and Influenza-A viral pneumonia using 2D and 3D deep learning techniques algorithms like Inception-V3, Xception, ResNet, ResNeXt, etc., Further proceeding into the design, the candidate regions were first segmented using these deep learning models from the CT images by applying filter layers such as convolutional layer pooling layer other dense layers depending on the model selected and activation functions such as a sigmoid function or ReLU function and then these were categorized into COVID-19, influenza-A viral pneumonia and healthy groups with the best accuracy results and the promising screening.

Index Terms: COVID-19, Deep Learning, CT images, Convolutional Neural Networks, Screening.