Papers/2608.18193
🧪 Test?View on arXiv

Bound-Aware Per-Organ Recall Risk Control for Multi-Organ CT Segmentation under Clinical Domain Shift

Not provided in the content

risk controlsegmentationclinical AIdomain adaptation
2608.18193
Builder Relevance
70%
1h ago

Abstract

This paper presents a method for adding organ-specific recall guarantees to segmentation models under clinical domain shifts.

Reality Card

Core Claim

The study demonstrates that while the AMOS control passes, 7 out of 12 organs exceed the recall threshold after transfer, indicating a need for better calibration methods.

Method / Result

The Waudby--Smith--Ramdas (WSR) betting bound re-certifies six Tier-1 organs with only 25 local cases.

Limitations

Smaller calibration sets can mask exceedances, leading to conservative or vacuous thresholds which affect reproducibility.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.
← Back to all papers