Related Experiment Video
Updated: Apr 30, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.6K
YoloSeg: You only label once for medical image segmentation
Mingen Zhang1, Yuanyuan Gu2, Meng Wang3
1Ningbo Key Laboratory of Biomedical Imaging Probe Materials and Technology, Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo, China; University of Chinese Academy of Sciences, Beijing, China.
Medical Image Analysis
|April 28, 2026
Summary
YoloSeg enables accurate medical image segmentation with only one labeled image by using label propagation and robust pseudo-label learning. This significantly reduces annotation burden, making AI development more feasible and cost-effective.
Area of Science:
- Artificial Intelligence
- Medical Imaging
- Computer Vision
Background:
- Medical image segmentation is crucial for diagnosis and treatment planning.
- Acquiring pixel-level annotations is labor-intensive, hindering AI model development.
- Existing semi-supervised methods still require substantial labeled data (10-30%).
Purpose of the Study:
- To introduce YoloSeg, a novel framework for medical image segmentation under extreme label scarcity.
- To enable effective model training with only a single labeled image.
- To reduce the annotation cost and time for medical image segmentation tasks.
Main Methods:
- Leveraging Segment Anything Model 2 for label propagation from a single image.
- Employing multi-view label propagation to decompose pseudo-labels into consensus and divergence regions.
- Utilizing a dual-component loss function and cross-patch data augmentation for robust learning and generalization.
Main Results:
- YoloSeg achieved performance comparable to fully-supervised methods across ten diverse medical imaging datasets.
- An average Dice score difference of only 3.08% was observed compared to fully-supervised baselines.
- Significantly outperformed existing state-of-the-art semi-supervised and one-shot segmentation methods.
Conclusions:
- YoloSeg significantly improves the feasibility and cost-effectiveness of deep learning for medical image segmentation.
- The framework facilitates rapid development and deployment of custom segmentation models in resource-limited settings.
- This approach supports the broader adoption of intelligent medical technologies by reducing annotation barriers.
