%0 Journal Article
%A Fischer, Maximilian
%A Muckenhuber, Alexander
%A Peretzke, Robin
%A Farah, Luay
%A Ulrich Harsy, Constantin
%A Ziegler, Sebastian
%A Schader, Philipp
%A Feineis, Lorenz
%A Gao, Hanno
%A Xiao, Shuhan
%A Götz, Michael
%A Nolden, Marco
%A Steiger, Katja
%A Sieveke, Jens T
%A Endrös, Lukas
%A Braren, Rickmer
%A Kleesiek, Jens
%A Schüffler, Peter
%A Neher, Peter
%A Maier-Hein, Klaus
%T Contrastive virtual staining enhances deep learning-based PDAC subtyping from H</td><td width="150">
%T amp;E-stained tissue cores.
%J The journal of pathology
%V 268
%N 1
%@ 0368-3494
%C Bognor Regis [u.a.]
%I Wiley
%M DKFZ-2025-02277
%P 89-98
%D 2026
%Z #EA:E230#LA:E230# / 2026 Jan;268(1):89-98
%X Pancreatic ductal adenocarcinoma (PDAC) subtyping typically relies on immunohistochemistry (IHC) staining for critical markers like HNF1A and KRT81, a labor-intensive manual staining process that introduces variability. Virtual staining methods offer promising alternatives by generating synthetic IHC images from routine hematoxylin and eosin (H</td><td width="150">
%X E) slides. However, most current approaches evaluate success by image quality measures rather than assessing diagnostically relevant features. Here, we introduce a novel cycleGAN framework utilizing a contrastive-inspired approach trained on semipaired datasets derived from consecutive tissue sections. Our method significantly enhances PDAC subtyping accuracy based on synthetic IHC images generated from standard H</td><td width="150">
%X E inputs, improving the classification F1-score from 0.66 to 0.77 for KRT81 and from 0.61 to 0.73 for HNF1A, compared with classification directly on H</td><td width="150">
%X E images. This approach also substantially outperforms baseline CycleGAN models. These results underscore the clinical potential of contrastive virtual staining to streamline PDAC diagnostics and improve their robustness. © 2025 The Author(s). The Journal of Pathology published by John Wiley </td><td width="150">
%X  Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.
%K CycleGAN (Other)
%K H&E (Other)
%K IHC (Other)
%K PDAC subtyping (Other)
%K contrastive learning (Other)
%K digital pathology (Other)
%K virtual staining (Other)
%F PUB:(DE-HGF)16
%9 Journal Article
%$ pmid:41188199
%R 10.1002/path.6491
%U https://inrepo02.dkfz.de/record/305635