LAUSR

The CT scan is an important diagnostic procedure for COVID-19. CT images tend to reveal similar features in most COVID-19 cases, including ground-glass opacity in the early stages and pulmonary consolidation in the later stages. The work presented here demonstrates the feasibility of developing a classification system for Covid-19 CT lung scans, which would assist doctors in distinguishing COVID-19 from healthy cases. Three architec-tures, ResN50, Inception-V3, and Xception, were trained, validated, and tested to achieve the highest accuracy. Furthermore, the influence of hyper-parameters on the accuracy of each model was evaluated, searching for minimal loss values. With this system, classification of CT scans can be performed. © 2022 CAD Solutions, LLC.