Detection of SARS-COVID-19 Based on CT Images Using Deep Learning-Based Hybrid Particle Swarm and Mayfly Optimization Algorithm

Abstract

More than 25 million people worldwide have contracted COVID-19 due to the SARS-COV-2 disease. Screening many suspected cases for quarantine and treatment is crucial to controlling the disease. Pathogenic laboratory testing is the gold standard but has a high false-negative rate, making alternate diagnostic procedures necessary to fight the disease. Based on COVID-19 radiographic changes in Computed Tomography (CT) scans, this research hypothesized that artificial intelligence (AI) approaches may extract particular graphical features and offer a clinical diagnosis before the pathogenic test, saving time for disease management. 1065 CT images of pathogen-confirmed COVID-19 and typical viral pneumonia patients were obtained. To develop the method, we proposed Deep Learning (DL)-based hybrid Particle Swarm Optimization and Mayfly Optimization (PSO-MFO) algorithm. PSO-MFO as a classifier to detect SARS-COVID-19. Internal validation yielded accuracy, specificity, and sensitivity. The external testing dataset has accuracy, specificity, and sensitivity. These findings show that AI can extract radiological features for COVID-19 diagnosis.