Reducing Annotation Burden for Femoral Cartilage Segmentation in Knee MRI via Cross-Sequence Transfer Learning
2026
Read Reducing Annotation Burden for Femoral Cartilage Segmentation in Knee MRI via Cross-Sequence Transfer Learning (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-2605-09067.txt.
Description
Francesco Chiumento and nine co-authors study how to outline thigh cartilage in knee MRI scans with far less hand labeling. Drawing these outlines by hand is slow, and a model trained on one scan type often fails on another. Their idea is to train on one sequence, then adapt the model to a second with only a few labeled cases.
Working with hundreds of scans from a large arthritis study plus smaller sets from a hospital in Bologna, they test transfer in both directions between two scan types. Transfer one way matched training from scratch with just nine subjects. Transfer the other way was weaker, and cartilage damage hurt one scan type but not the other.
Catalog record
| YEAR | 2026 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.09067