Tonsillitis based Early Diagnosis of COVID-19 for Mass-Screening using One-Shot Learning Framework

Abstract

The outbreak of COVID-19 received substantial international attention due to its life-threatening repercussions. Automatic and non-invasive mass screening is pivotal to ensure that the suspected cases can be quarantined, and the spread of the disease can be controlled during the incubation period of the coronavirus. Currently, thermal screening is the only non-invasive mass-screening technique being used. However, with the consumption of paracetamol, the symptoms of fever can be suppressed. Throat inflammation is one of the perceivable symptoms of COVID-19. This paper proposes a novel approach of analyzing the throat images using a Siamese Network based One-Shot learning framework for the identification of Tonsillitis and Pharyngitis for the automatic mass-screening and early-detection of COVID-19 that can be administered without direct contact with the infectious patients, thereby reducing the burden on the medical and paramedical fraternity. Quantitative and qualitative evaluation exhibit promising results and indicate the reliability of the proposed method.