AI-Driven Radiomics for Rapid and Accurate COVID-19 Detection Using Chest CT Scans

Abstract

This study article seeks to present a useful tool for doctors to use with deep learning architectures in order to aid in the diagnosis of COVID-19, which has recently surged on a global scale. The examination of medical pictures, namely chest CT images, is the main basis for the automated diagnosis of COVID-19. The small air sacs called alveoli are damaged and the regular functioning of the lungs is affected by this COVID-19. Chest computed tomography (CT) scans are more important for diagnosing patients with severe infections and for screening for COVID-19 immediately prior to specific emergency surgeries and treatments. Radiologists' visual interpretation of chest CT scans has been the gold standard for diagnosis up until now, however it can be inaccurate. To start, diagnosing with a chest CT, which holds hundreds of slices, takes time. The next example is COVID-19, a lung illness that causes a wide range of pneumonia types. Based on the radiomics features of the pre-processed chest CT images, this research aims to detect anomalies and diagnose the severity of COVID-19. This information is useful for differentiating between the usual opaque area, GGOs, and the high intensity area, which contains things like blood vessels and other types of consolidations. In comparison to current qualitative evaluation methods, our methodology (classification of chest CT images utilising radiomic characteristics for COVID-19 diagnosis using neural network) can reduce subjective variability and increase diagnostic efficiency.