Ensemble of Weak Spectral Total Variation Learners: a PET-CT Case Study
Anna Rosenberg 1, John Kennedy 2, Zohar Keidar 3, Yehoshua Y. Zeevi 4 and Guy Gilboa 5 · 2025
Read Ensemble of Weak Spectral Total Variation Learners: a PET-CT Case Study by Anna Rosenberg 1, John Kennedy 2, Zohar Keidar 3, Yehoshua Y. Zeevi 4 and Guy Gilboa 5 (2025) 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/arxiv-2507-08735.txt.
Description
Anna Rosenberg, John Kennedy, Zohar Keidar, Yehoshua Zeevi, and Guy Gilboa build in a 2025 paper ensembles of weak image learners for a PET CT task. Training data scarcity dogs many medical vision problems in practice. Their features come from spectral total variation, which reads texture at many scales.
The clinical task predicts hot uptake on PET from CT alone in suspected bone spread cases. Data hold 457 scans with 1,524 matched slice pairs across the set. Weak learners on those features vote as an ensemble for the call. The ensemble scored 0.87 on the curve test, beating neural nets at 0.75. Radiomics features landed between at 0.79 in the same trials.
Catalog record
| CREATOR | Anna Rosenberg 1, John Kennedy 2, Zohar Keidar 3, Yehoshua Y. Zeevi 4 and Guy Gilboa 5 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-08735