Metraipolis › archive › dc_52cb6ebb2920

Look What the Probes Dragged In! Real-World Chest X-ray Shortcuts in MedCLIP — arXiv:2608.12086

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.12086
License (as archived)
CC0-1.0
Archived
2026-10-19T05:40:28.389Z
SHA-256 of the archived text
36ec60beeb2c946231e2bb787d49c6c0bf41eff9ca324cea3095e4c4552d57bd
Stable URL of this page
https://metraipolis.com/documento/dc_52cb6ebb2920

How to cite

Look What the Probes Dragged In! Real-World Chest X-ray Shortcuts in MedCLIP — arXiv:2608.12086 — arXiv (en), https://arxiv.org/abs/2608.12086 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_52cb6ebb2920 (Metraipolis archive dc_52cb6ebb2920, sha256 36ec60beeb2c946231e2bb787d49c6c0bf41eff9ca324cea3095e4c4552d57bd)
BibTeX
@misc{dc52cb6ebb2920,
  title = {Look What the Probes Dragged In! Real-World Chest X-ray Shortcuts in MedCLIP — arXiv:2608.12086},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.12086},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_52cb6ebb2920; sha256 36ec60beeb2c946231e2bb787d49c6c0bf41eff9ca324cea3095e4c4552d57bd},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.12086. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Vision-language models, such as contrastive language-image pre-training (CLIP)-based approaches, have reached state-of-the-art (SOTA) results in medical artificial intelligence. However, recent work reveals that CLIP-based models remain vulnerable to shortcuts. We investigate how real-world shortcuts manifest across different layers of the medical CLIP-based model, MedCLIP, and its vision encoder, a frozen ResNet-50. We attach 17 linear classification probes to the intermediate layers of the ResNet-50 and train them on three different dataset configurations and targets: NIH-CXR14 (pneumothorax) and PadChest (cardiomegaly and pneumothorax). This setup allows us to observe model behaviour during evaluation using subgroup-based calibration and layer-wise confidence curves. We find that the final linear probes achieve a high AUROC but poor calibration in the models. The layer-wise confidence analyses suggest that shortcuts emerge at different depths. Patterns consistent with localised shortcuts, such as drains, appear at later layers, while patterns consistent with diffuse shortcuts, such as scanner-specific noise patterns, emerge earlier, aligning with previous work. Finally, we conduct a manual analysis of the images, which reveals data quality issues in both NIH-CXR14 and PadChest. Our findings underscore that even SOTA models remain vulnerable to shortcuts, and the need for high-quality and well-annotated datasets to draw solid conclusions. Code can be found on our GitHub: this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-13 in cs.LG).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2608.12086 — 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 2026-10-19T05:18:32.554Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-08-13?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).