![[2602.19412] Redefining the Down-Sampling Scheme of U-Net for Precision Biomedical Image Segmentation](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.19412] Redefining the Down-Sampling Scheme of U-Net for Precision Biomedical Image Segmentation
Summary
This paper presents a novel down-sampling strategy called Stair Pooling for U-Net architectures, aimed at enhancing precision in biomedical image segmentation by reducing information loss during down-sampling.
Why It Matters
Biomedical image segmentation is crucial for accurate medical diagnostics and treatment planning. This research addresses limitations in existing U-Net models, proposing a method that improves segmentation accuracy, which could lead to better patient outcomes and advancements in medical imaging technologies.
Key Takeaways
- Stair Pooling moderates down-sampling pace, preserving more information.
- The method enhances U-Net's ability to capture long-range dependencies.
- Experimental results show an average improvement of 3.8% in Dice scores.
- The approach can be adapted for both 2D and 3D biomedical image segmentation.
- Transfer entropy is used to optimize down-sampling paths effectively.
Computer Science > Computer Vision and Pattern Recognition arXiv:2602.19412 (cs) [Submitted on 23 Feb 2026] Title:Redefining the Down-Sampling Scheme of U-Net for Precision Biomedical Image Segmentation Authors:Mingjie Li, Yizheng Chen, Md Tauhidul Islam, Lei Xing View a PDF of the paper titled Redefining the Down-Sampling Scheme of U-Net for Precision Biomedical Image Segmentation, by Mingjie Li and 3 other authors View PDF HTML (experimental) Abstract:U-Net architectures have been instrumental in advancing biomedical image segmentation (BIS) but often struggle with capturing long-range information. One reason is the conventional down-sampling techniques that prioritize computational efficiency at the expense of information retention. This paper introduces a simple but effective strategy, we call it Stair Pooling, which moderates the pace of down-sampling and reduces information loss by leveraging a sequence of concatenated small and narrow pooling operations in varied orientations. Specifically, our method modifies the reduction in dimensionality within each 2D pooling step from $\frac{1}{4}$ to $\frac{1}{2}$. This approach can also be adapted for 3D pooling to preserve even more information. Such preservation aids the U-Net in more effectively reconstructing spatial details during the up-sampling phase, thereby enhancing its ability to capture long-range information and improving segmentation accuracy. Extensive experiments on three BIS benchmarks demonstrate that the p...