Metraipolis › archive › dc_47be2d635a95

Medical AI Encodes a "Feeling of Error": Verifying Cancer Segmentation via Internal Concepts — arXiv:2609.08879

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.08879
License (as archived)
CC0-1.0
Archived
2026-10-16T15:12:27.299Z
SHA-256 of the archived text
ffb1b486894ab30cecd1e98e7695ab28a5a40b5b99fd1aca4777a9f99760c6b9
Stable URL of this page
https://metraipolis.com/documento/dc_47be2d635a95

How to cite

Medical AI Encodes a "Feeling of Error": Verifying Cancer Segmentation via Internal Concepts — arXiv:2609.08879 — arXiv (en), https://arxiv.org/abs/2609.08879 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_47be2d635a95 (Metraipolis archive dc_47be2d635a95, sha256 ffb1b486894ab30cecd1e98e7695ab28a5a40b5b99fd1aca4777a9f99760c6b9)
BibTeX
@misc{dc47be2d635a95,
  title = {Medical AI Encodes a "Feeling of Error": Verifying Cancer Segmentation via Internal Concepts — arXiv:2609.08879},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.08879},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_47be2d635a95; sha256 ffb1b486894ab30cecd1e98e7695ab28a5a40b5b99fd1aca4777a9f99760c6b9},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.08879. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Cancer segmentation models can fail silently, generating plausible but incorrect masks that risk missed findings or unnecessary biopsies. A critical question arises: Do AI models "know" when they are wrong, and if so, can we use the signal to predict their own failures? Humans do have a "Feeling of Error" (FOE): a spontaneous sense of unease that flags a potential error during thinking. We investigate whether cancer segmentation models exhibit an analogous internal signal. Unlike output-level cues (e.g., prediction confidence or uncertainty), which offer no insight into why a failure occurs and suffer from a sensitivity-quality tradeoff where high detection sensitivity could degrade overall segmentation quality. We instead propose to capture the model's FOE from its inner workings. Using mechanistic interpretability tools, specifically Sparse Autoencoders, we decompose internal neural activations into a dictionary of human-interpretable concepts and show that failure cases exhibit a distinct latent signature: fewer active concepts with lower activation magnitudes compared to successful segmentation. By training a classifier on these concept activations, we achieve accurate failure detection along with explanations for the model's mistakes. Experiments on prostate, pancreatic, and brain cancer segmentation demonstrate that our approach outperforms output-based methods in failure detection while preserving segmentation quality.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.08879 — 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 2026-10-16T14:49:57.361Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).