TriAlign UDA for hybrid domain adaptation of frozen histopathology foundation model representations

Foundation models produce strong patch-level representations in histopathological image analysis; however, domain shifts across different centers, staining protocols, and scanner conditions can limit the direct generalizability of these representations. In this study, a hybrid feature-level unsupervised domain adaptation (UDA) framework called TriAlign-UDA, which operates on frozen histopathology foundation model representations, is proposed. TriAlign-UDA aims to reduce the mismatch between the labeled source domain and the completely unlabeled target domain without requiring encoder…

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give... [797 chars]