Automatic Diagnosis of COVID-19 Using a tailored Transformer-Like Network
Chengeng Liu, Qingshan Yin
Abstract
Abstract The emergence of the novel coronavirus(COVID-19) has left disastrous effect on global health and individuals. Even though in most areas, the RT-PCR test used as the dominant approach for diagnosis of COVID-19 has shown good accuracy, the test requires equipment, personnel and it is time-consuming. Researches have shown the effectiveness of X-ray images for predicting COVID-19. In this study, we applied a transformer-like deep-learning model on this problem with transfer learning technique, to diagnose X-ray images as COVID-19 or normal. The model outperformed the CNN SOTA. The model achieved a classification accuracy of 99.7% on the targeting dataset.
Topics & Concepts
Coronavirus disease 2019 (COVID-19)TransformerTransfer of learningSevere acute respiratory syndrome coronavirus 2 (SARS-CoV-2)2019-20 coronavirus outbreakArtificial intelligenceComputer scienceDeep learningMachine learningVirologyEngineeringMedicinePathologyElectrical engineeringDiseaseOutbreakVoltageInfectious disease (medical specialty)COVID-19 diagnosis using AIAI in cancer detectionDigital Imaging for Blood Diseases