Mammographic Lesion Segmentation with Lightweight Models: A Comparative Study
Helder Oliveira · 2026
Read Mammographic Lesion Segmentation with Lightweight Models: A Comparative Study by Helder Oliveira (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-2604-23899.txt.
Description
Oliveira compares light models for breast lesion segmentation against a U-Net baseline on the INbreast dataset. MobileNetV2 with squeeze-and-excitation achieves the best Dice score while using about three-quarters fewer weights than U-Net. Cross-dataset tests show lower scores from domain shift but preserved recall. The results support light models as a practical balance of quality and speed for usable screening systems.
Catalog record
| CREATOR | Helder Oliveira |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39598 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.23899