Spatially anchored features improve cross-stain slide alignment accuracy
SAR2Net: Learning Spatially Anchored Representations for Retrieval-Guided Cross-Stain Alignment
Computer Vision and Pattern Recognition
Summary
Aligning images of tissue slides stained in different ways is hard because the tissues can be torn and distorted. The authors propose SAR2Net, which does not try to find a one-size-fits-all adjustment but instead learns how parts of the tissue relate to landmarks nearby. This helps find matching areas between slides even when the overall shapes look very different. Their tests on real biopsy images show SAR2Net does this better than previous methods.
What this means in practice
- •For pathology imaging teams: Match regions across differently stained tissue slides to combine morphological and molecular data for diagnosis.
- •For digital histology platform developers: Integrate SAR2Net for better automated alignment in slide viewers handling multiple staining methods.$Commercial implications: Enables improved slide registration features that can be sold to hospitals and labs for enhanced diagnostic tools.
Authors
Tianle Shen, Fang Yan, Xiaofan Zhang
Abstract
Achieving spatial alignment between whole-slide images(WSIs) across stains is essential for integrating morphology and molecular information, yet it remains challenging due to extreme resolution, tissue fragmentation, and large nonlinear deformations. Conventional registration pipelines depend on global pre-alignment and spatial consistency, which often collapse under such distortions. We present SAR2Net, a framework that learns spatially anchored representations and reformulates cross-stain alignment as a region-level feature retrieval problem. Instead of estimating explicit transformations, SAR2Net learns pointwise representations encoding the relative spatial relationships to tissue landmarks. Given landmarks and arbitrary coordinates, it predicts spatially anchored features that serve as deformation-invariant descriptors of tissue topology. A twostage retrieval framework then establishes correspondences between slides, even when global alignment is infeasible. Experiments on biopsy-oriented HE-IHC datasets show that SAR2Net achieves robust region-level alignment under severe tissue distortions, outperforming previous methods. Code is available at https://github.com/Shentl/SAR2Net.