Performance analysis of liver segmentation using nn-UNet TotalSegmentator: Focus on atypical livers, pathologies, and variants
TS performs excellently on healthy and well on most pathological livers. Despite high Dice scores in many pathological cases, clinical ratings reveal limitations. Clinical evaluation remains essential. Inclusion of PLD and Beavertail cases in training data may reduce bias and improve performance.
Authors
- J. Kleiß
- M May
- Maximilian F. Schmidt (ORCID: https://orcid.org/0000-0002-7055-2439)
- Teresa Graetz
- L Stepansky
- Lisa Sommerfeld
- K Türkan
- Florian Putz
- Simon Mayr
- Michael Uder
- Sebastian Arndt
Institutions
- Friedrich-Alexander-Universität Erlangen-Nürnberg (DE)
- Universitätsklinikum Erlangen (DE)
Publication Details
- Journal
- European Journal of Radiology
- Published
- 2026-01-14
- DOI
- https://doi.org/10.1016/j.ejrad.2026.112674
- Primary Topic
- AI in cancer detection
- Type
- article
- Field-Weighted Citation Impact
- 0.00