Using infrared imaging and deep learning in fit-checking of respiratory protective devices among healthcare professionals

Abstract

Aims: This study aimed to investigate the application of infrared thermal imaging and adopt deep learning to detect air leakage for determining the fitness of respirators during fit-checks. Background: The outbreak of Covid-19 virus constitutes a public health crisis with substantial resultant morbidities and mortalities; has exerted profound impacts. Methods: This was a prospective observational study, employing a non-probability sampling method on a convenience sample to recruit the participants and followed the Strengthening the Reporting of Observational Studies in Epidemiology statement guidelines. Results: The use of infrared thermal imaging identified air leakage points as a disruption to the facial thermal pattern distribution at (a) front of face; (b) right lateral of the face; (c) left lateral of the face; (d) top of the facemask with the head facing down; and (e) bottom of the facemask with the head facing up. Results also indicated that artificial intelligence tools and the proliferation of deep learning have the potential to detect the location of air leakage locations. Conclusion: The use of infrared thermal imaging provides evidence of the feasibility and applicability of infrared thermal imaging techniques in detecting air leakage for individuals wearing respirators. Clinical relevance: The use of infrared thermal technology can serve a potential role in complement fit-checking of respiratory protective devices and offers promising practical utility in determining the fitness of respirators for nurses at the frontline to protect against the air-borne viruses.