Automatic X-ray COVID-19 Lung Image Classification System based on Multi-Level Thresholding and Support Vector Machine

Aboul Ella Hassanien; Lamia Nabil Mahdy; Kadry Ali Ezzat; Haytham H. Elmousalami; Hassan Aboul Ella

As a result of this surveying, Although the diagnosis of COVID-19 is based firstly and mainly on the presence of clinical signs related to respiratory systems and more specifically an pneumonia related signs (e.g., dry cough, fatigue, myalgia, fever, and dyspnea) as well as history of recent (two weeks apart) exposure to a known confirmed cases, a broad range of means of COVID-19 rapid bed side, in field or point of care (POCT) medical diagnostics to help in this pandemic outbreak screening already have been developed in the few past weeks or still under developing for few upcoming days such as IgM/IgG based lateral flow (lateral immunochromatographic assay), Nucleic acid lateral flow (NALF), CRISPR-cas13 lateral flow (CASLFA) and radiological imaging. Although, CT scans and Xray imaging are time consuming and exhaustive even for expert radiologists, it is characterized by that the highest sensitivity diagnostic method in comparison to the gold standard (qRT-PCR) and other already developed or will be developed rapid diagnostics. So, relying on radiological imaging in screening and diagnosis of COVID-19 is very medically meaningful and therefore a need to automated system based on Artificial Intelligence (AI) tools could be accurately provide automated, less exhaustive for medical imaging workers, early detection, follow up method of COVID-19 cases and at same time method of differentiation of the radiological images of affected lung due to COVID-19 from other causes of lung affections especially MERS which caused by (MERS-CoV) and SARS which caused by (SARS-CoV-1) which are another viruses belonging to Coronavirdeae viruses family, the same family to which the COVID-19 causative agent (SARS-CoV-2) is belonging and all three viruses have an overlapping clinical signs and affected lung radiological image patterns but a differences and uniqueness still present.