Metraipolis › archive › dc_80451a85ec3f

Sparse Autoencoders for Expert-Assisted Interpretation of Histopathology Foundation Model Embeddings — arXiv:2610.40905

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.40905
License (as archived)
CC0-1.0
Archived
2026-11-03T05:46:11.614Z
SHA-256 of the archived text
f2681160d32d3f5ef4e8240a0daf9c196923d6adf98d3b6a867b60f1d3826ed1
Stable URL of this page
https://metraipolis.com/documento/dc_80451a85ec3f

How to cite

Sparse Autoencoders for Expert-Assisted Interpretation of Histopathology Foundation Model Embeddings — arXiv:2610.40905 — arXiv (en), https://arxiv.org/abs/2610.40905 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_80451a85ec3f (Metraipolis archive dc_80451a85ec3f, sha256 f2681160d32d3f5ef4e8240a0daf9c196923d6adf98d3b6a867b60f1d3826ed1)
BibTeX
@misc{dc80451a85ec3f,
  title = {Sparse Autoencoders for Expert-Assisted Interpretation of Histopathology Foundation Model Embeddings — arXiv:2610.40905},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.40905},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_80451a85ec3f; sha256 f2681160d32d3f5ef4e8240a0daf9c196923d6adf98d3b6a867b60f1d3826ed1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40905. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Histopathology foundation models learn rich embedding representations, but their internal structure remains difficult to inspect. Concept-level interpretability is particularly relevant in computational pathology because pathologists reason through recurring morphological and tissue-level patterns, including tissue architecture, cellular morphology, staining variation, and artifacts. In this work, we investigate whether sparse autoencoders (SAEs) can decompose embeddings from histopathology foundation models into sparse latent features suitable for expert-assisted interpretation. We train vanilla and Batch-TopK SAE variants on NCT-CRC embeddings, using CONCH as the primary foundation model and UNI2-h as an additional robustness evaluation, and characterize the resulting latent features through reconstruction fidelity, sparsity, coherence, label enrichment, and downstream class-signal preservation. The learned sparse representations preserve most downstream tissue-class information, and their latent features retrieve more coherent and label-concentrated patch groups than raw embedding dimensions or PCA components. Discussions with pathology experts further motivated a WSI-level latent visualization workflow that places feature activations into larger tissue context. Overall, the additional UNI2-h evaluation suggests that the observed behavior is not specific to CONCH, supporting the use of sparse autoencoders for expert-assisted analysis of histopathology foundation-model embeddings, while emphasizing that latent features should be treated as candidate concepts requiring pathological assessment rather than automatically validated histopathological concepts.

Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2610.40905 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-03T05:33:32.262Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).