CTCovid19: Automatic Covid-19 model for Computed Tomography Scans Using Deep Learning

COVID-19 is an extremely contagious respiratory sickness instigated by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Common symptoms encompass fever, cough, fatigue, and breathing difficulties, often leading to hospitalization and fatalities in severe cases. CTCovid19 is a novel model tailored for COVID-19 detection, specifically honing in on a distinct deep learning structure, ResNet-50 trained with ImageNet serves as the foundational framework for our model. To enhance its capability to capture pertinent features related to COVID-19 patterns in Computed Tomography scans, the network underwent fine-tuning through layer adjustments and the addition of new ones. The model achieved accuracy rates that went from 97.0% to 99.8% across three widely recognized and documented datasets dedicated to COVID-19 detection. • Objective : The paper presents a novel deep learning model, CTCovid19, designed specifically for detecting COVID-19 in Computed Tomography (CT) scans. The model is based on the ResNet-50 architecture, trained with ImageNet, and fine-tuned to capture COVID-19 patterns in CT scans. • Background : COVID-19, caused by the SARS-CoV-2 virus, is a highly contagious respiratory disease. Traditional detection methods can be laborious, time-consuming, and sometimes fail, leading to detection failures at the initial stages of the disease. CT scans of the thorax provide an alternative method for detecting COVID-19. • Model Development : The CTCovid19 model is an improvement over a previous model developed by the authors, CPAlexNet, which was focused on the AlexNet architecture. The new model uses the ResNet-50 architecture as its backbone, with layers adjusted and new ones added to better capture COVID-19 patterns in CT scans. • E xplainable AI : The paper also discusses the use of Explainable Artificial Intelligence (XAI) in CT scans to help explain to doctors how the detection was achieved. This is illustrated by two methods: Local Interpretable Model-agnostic Explanations (LIME) and Gradient-weighted Class Activation Mapping (Grad-Cam). • Testing and Results : The model was tested on three different public datasets of CT scans with and without COVID-19. The model achieved accuracy rates ranging from 97.0% to 99.8% across these datasets. • Contribution : The main contribution of the paper is to provide a novel, accurate, and practical mechanism for detecting COVID-19 on CT scans based on a deep learning model. • Keywords : Deep Learning, CT scans, Covid-19, Convolutional Neural Network (CNN), and Explainable Artificial Intelligence (XAI).