From Cold Start to Active Learning: Embedding-Based Scan Selection for Medical Image Segmentation
Devon Levy, Bar Assayag, Laura Gaspar, Ilan Shimshoni, Bella Specktor-Fadida · 2026
Read From Cold Start to Active Learning: Embedding-Based Scan Selection for Medical Image Segmentation by Devon Levy, Bar Assayag, Laura Gaspar, Ilan Shimshoni, Bella Specktor-Fadida (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-2601-18532.txt.
Description
Devon Levy and four colleagues cut the hand work in a key step of health care: marking zones in scan shots. A marked scan lets staff track a disease and plan care. But each mark takes skill and time, so labs can mark just a few scans from a large set. The team use a smart pick plan: the tool asks for marks on the scans that teach it most.
The plan starts cold, with no marks at all. It picks a first set by how spread the scans are in a learned space of form. Then it turns to doubt: each new round asks for marks where the tool is least sure. A good pick plan can reach top grades with far fewer hand marks. That could free scan staff for care, not just for marks.
Catalog record
| CREATOR | Devon Levy, Bar Assayag, Laura Gaspar, Ilan Shimshoni, Bella Specktor-Fadida |
|---|---|
| YEAR | 2026 |
| 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/2601-18532