From noisy labels to intrinsic structure: A geometric-structural dual-guided framework for noise-robust medical image segmentation.
Authors
Affiliations (6)
Affiliations (6)
- College of Physics and Information Engineering, Fuzhou University, Xueyuan Road No. 2, Fuzhou, 350108, China; Bioengineering Department and Imperial-X, Imperial College London, London W12 7SL, UK. Electronic address: [email protected].
- Bioengineering Department and Imperial-X, Imperial College London, London W12 7SL, UK.
- College of Physics and Information Engineering, Fuzhou University, Xueyuan Road No. 2, Fuzhou, 350108, China.
- Faculty of Applied Science, Macao Polytechnic University, Macao Special Administrative Region of China.
- Bioengineering Department and Imperial-X, Imperial College London, London W12 7SL, UK; National Heart and Lung Institute, Imperial College London, London SW7 2AZ, UK; Cardiovascular Research Centre, Royal Brompton Hospital, London SW3 6NP, UK; School of Biomedical Engineering & Imaging Sciences, King's College London, London WC2R 2LS, UK.
- College of Physics and Information Engineering, Fuzhou University, Xueyuan Road No. 2, Fuzhou, 350108, China. Electronic address: [email protected].
Abstract
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 S<sub>R</sub> simulated noise. The code of this study is available at https://github.com/ortonwang/GSD-Net.