Leveraging anatomical constraints with uncertainty for pneumothorax segmentation
Han Yuan, Chuan Hong, Nguyen Tuan Anh Tran, Xinxing Xu, Nan Liu · 2024
Read Leveraging anatomical constraints with uncertainty for pneumothorax segmentation by Han Yuan, Chuan Hong, Nguyen Tuan Anh Tran, Xinxing Xu, Nan Liu (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-11671217.txt.
Description
Han Yuan and four co-authors teach AI to find collapsed lung air pockets on chest X-rays using anatomy as a guardrail, in a 2024 paper. Pneumothorax is an emergency where air leaks into the chest cavity. The leak can only occur inside a known region around the lung.
Their model trains with that region as a constraint, computed from a side task of lung outlining. A checker discards constraints that look unreliable for a given image. Across six baseline models, the constraint lifted accuracy by roughly 3 to 5 percent on standard scores. Gains held across two network families. The work argues medical AI should use body knowledge, not ignore it.
Catalog record
| CREATOR | Han Yuan, Chuan Hong, Nguyen Tuan Anh Tran, Xinxing Xu, Nan Liu |
|---|---|
| YEAR | 2024 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |