Anatomical Token Uncertainty for Transformer-Guided Active MRI Acquisition
Lev Ayzenberg, Shady Abu-Hussein, Raja Giryes, Hayit Greenspan · 2026
Read Anatomical Token Uncertainty for Transformer-Guided Active MRI Acquisition by Lev Ayzenberg, Shady Abu-Hussein, Raja Giryes, Hayit Greenspan (2026) 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-2603-21806.txt.
Description
Ayzenberg and co-authors speed up MRI scans by letting the scanner choose what to measure next. Full MRI sampling is slow and hard on patients, so methods rebuild images from partial data. Their system uses a pretrained image tokenizer that turns anatomy into tokens with known odds. Token entropy then marks where the model is least sure. One strategy maps that doubt into sampling lines, and another follows its gradient. On fastMRI knee and brain data, their active policies beat strong baselines in perceptual quality at high speedups.
Catalog record
| CREATOR | Lev Ayzenberg, Shady Abu-Hussein, Raja Giryes, Hayit Greenspan |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20660 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.21806