Artificial intelligence for chest imaging in hantavirus pulmonary syndrome: A review of practical considerations for rare-disease imaging.
Authors
Affiliations (2)
Affiliations (2)
- Temerty Faculty of Medicine, University of Toronto, 1 King's College Circle, Toronto, Ontario, M5S 1A8, Canada. Electronic address: [email protected].
- Schulich School of Medicine & Dentistry, Western University, 1151 Richmond St, London, Ontario, N6A 5C1, Canada.
Abstract
Hantavirus pulmonary syndrome (HPS) is an uncommon zoonotic respiratory illness with a high case fatality rate. Recent travel-associated Andes orthohantavirus clusters have renewed interest in early radiographic recognition of the disease outside its usual geographic range. Artificial intelligence (AI) tools developed during the COVID-19 pandemic now inform several radiology workflows, but the application of AI to HPS chest imaging remains largely unexplored. This narrative review summarises the imaging features of HPS across modalities, considers lessons from AI applications in COVID-19 and viral pneumonia imaging, and discusses practical development considerations for HPS-aware imaging AI, including dataset development, label standardisation, realistic comparator inclusion, candidate modelling approaches, external validation, and clinical workflow integration. The review also considers challenges to HPS-aware imaging AI, including data scarcity, low pretest probability in non-endemic settings, overlapping imaging findings, and translational pitfalls from recent AI literature. As HPS is unlikely to support conventional disease-specific model development without coordinated data collection, future work will likely require multi-institutional datasets, prevalence-aware evaluation, and multimodal workflow integration that combines imaging with clinical, laboratory, and epidemiologic context.