. Unsupervised Domain Adaptation for Cell Detection Across Histopathological Stains. In 21st European Congress on Digital Pathology. Barcelona: The European Society of Digital and Integrative Pathology; 2025. 
(601.11 KB)
(601.11 KB) Abstract
Deep learning models trained on hematoxylin and eosin (H&E) stained images often fail to generalize to immunohistochemistry (IHC) stains due to domain shifts. The scarcity of annotated IHC datasets limits the applicability of supervised learning. To address this challenge, we propose an unsupervised domain adaptation (UDA) approach that enables CellNuc-DETR to generalize from H&E to IHC without requiring additional manual annotations.
