News/Events

[Frontiers in immunology, 2026] [안혁원] Deep learning–assisted, pathogenesis-informed lung histopathology scoring in mice

작성자

안혁원

작성일

2026-09-10

조회

Deep Learning Helps Refine Lung Pathology Assessment in Viral Pneumonia

In our study published in Frontiers in Immunology, we combined deep learning with expert histopathology to characterize distinct patterns of lung injury and recovery in mouse models of SARS-CoV-2 and influenza A infection. Using H&E whole-slide images and class activation mapping, we identified differences in lesion distribution and tissue architecture that conventional inflammation-burden scoring could overlook.

We found that SARS-CoV-2 predominantly affected alveolar and interstitial regions at peak injury, whereas influenza A showed bronchocentric inflammation. During the resolution phase,

influenza A exhibited more prominent epithelial regeneration and remodeling, while SARS-CoV-2 retained localized inflammatory foci. Based on these findings, we proposed a pathogen- and phase-specific histopathology scoring guide to support more biologically meaningful assessment

of lung injury and repair in preclinical viral pneumonia studies.