Quantifying the Resolution Ceiling of Prototype-Based Few-Shot Segmentation on Sub-Centimetre Pulmonary Nodules: A Training-Free Geometric Analysis
Joon-Ho Cho · 2026
Read Quantifying the Resolution Ceiling of Prototype-Based Few-Shot Segmentation on Sub-Centimetre Pulmonary Nodules: A Training-Free Geometric Analysis by Joon-Ho Cho (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/pmc-13605439.txt.
Description
Joon-Ho Cho measures a hard geometric limit on outlining tiny lung nodules with few shot segmentation, in work from twenty twenty-six. Few shot methods outline new targets from a few examples. Most tests use large belly organs, but most lung nodules are under a centimetre and scores drop sharply there.
Cho trains no model here. Instead, a math bound shows the best score any whole cell grid method can reach, set by lesion size versus cell size. On more than twenty-six hundred shared nodule outlines, the bound fell very low for the most common size band. Careless masking could erase the very lesion sought. A training free sweep tracked the ceiling, and a derived tile size did far better than full chest input.
Catalog record
| CREATOR | Joon-Ho Cho |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 95 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |