Rethinking the Nested U-Net Approach: Enhancing Biomarker Segmentation with Attention Mechanisms and Multiscale Feature Fusion
Saad Wazir, Daeyoung Kim · 2025
Read Rethinking the Nested U-Net Approach: Enhancing Biomarker Segmentation with Attention Mechanisms and Multiscale Feature Fusion by Saad Wazir, Daeyoung Kim (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-2504-06158.txt.
Description
Saad Wazir and Daeyoung Kim present a 2025 image model for marking biomarkers in medical scans. Such marking is key for biotech work, but shifts in shape and staining blunt many models. Data are also scarce, which makes the hand-off of traits between the model's halves a weak point.
Their nested U-Net design fuses traits at many scales and adds attention steps that stress key channels and zones. That mix keeps fine detail while holding the wide scene in view. Tests on four datasets, plus part-removal checks, show the design beats top past methods.
Catalog record
| CREATOR | Saad Wazir, Daeyoung Kim |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 22 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-06158