@article{Wang:2026:10.1016/j.media.2026.104130,
author = {Wang, T and Zhang, Z and Zhou, Y and Zhang, X and Chen, Y and Tan, T and Yang, G and Tong, T},
doi = {10.1016/j.media.2026.104130},
journal = {Med Image Anal},
title = {From noisy labels to intrinsic structure: A geometric-structural dual-guided framework for noise-robust medical image segmentation.},
url = {http://dx.doi.org/10.1016/j.media.2026.104130},
volume = {112},
year = {2026}
}

Download

TY  - JOUR
AB  - The effectiveness of convolutional neural networks in medical image segmentation relies on large-scale, high-quality annotations, which are costly and time-consuming to obtain. Even expert-labeled datasets inevitably contain noise arising from subjectivity and coarse delineations, which disrupt feature learning and adversely impact model performance. To address these challenges, this study proposes a Geometric-Structural Dual-Guided Network (GSD-Net), which integrates geometric and structural cues to improve robustness against noisy annotations. It incorporates a Geometric Distance-Aware module that dynamically adjusts pixel-level weights using geometric features, thereby strengthening supervision in reliable regions while suppressing noise. A Structure-Guided Label Refinement module further refines labels with structural priors, and a Knowledge Transfer module enriches supervision and improves sensitivity to local details. To comprehensively assess its effectiveness, we evaluated GSD-Net on six publicly available datasets: four containing three types of simulated label noise, and two with multi-expert annotations that reflect real-world subjectivity and labeling inconsistencies. Experimental results demonstrate that GSD-Net achieves state-of-the-art performance under noisy annotations, achieving improvements of 1.58% on Kvasir, 22.76% on Shenzhen, 8.87% on BU_SUC, and 1.77% on BraTS2020 under SR simulated noise. The code of this study is available at https://github.com/ortonwang/GSD-Net.
AU  - Wang,T
AU  - Zhang,Z
AU  - Zhou,Y
AU  - Zhang,X
AU  - Chen,Y
AU  - Tan,T
AU  - Yang,G
AU  - Tong,T
DO  - 10.1016/j.media.2026.104130
PY  - 2026///
TI  - From noisy labels to intrinsic structure: A geometric-structural dual-guided framework for noise-robust medical image segmentation.
T2  - Med Image Anal
UR  - http://dx.doi.org/10.1016/j.media.2026.104130
UR  - https://www.ncbi.nlm.nih.gov/pubmed/42217335
VL  - 112
ER  - 

Download