Cardiac image segmentation improves with region-specific confidence roles
Reliability-Role Collaborative Learning for Semi-Supervised Cardiac Image Segmentation
Computer Vision and Pattern Recognition
Summary
Segmenting images of the heart can be tricky, especially when there are few labeled examples to learn from. The authors show that treating different parts of the image based on how confident the model is can improve learning. They use two networks that assign roles to each region depending on agreement and confidence, which helps focus learning on clearer areas and handle uncertain boundaries better. This approach leads to more accurate heart image segmentation, especially in challenging cases with fewer heart pixels.
What this means in practice
- •For medical imaging engineers: Improve automated heart image segmentation by using confidence-based learning to reduce errors in uncertain image regions.
- •For clinical imaging software developers: Enhance segmentation tools for cardiac CT scans by integrating region-role collaborative learning for better boundary and foreground detection.$Commercial implications: This paper enables more accurate and reliable automated heart segmentation features that can be sold as improved modules to healthcare software vendors.
Authors
Zhongrui Xie, Wenfeng Zhang, Wei Hu, Ruitao Xie, Wenjie Wang, Zhouyun Xie, Yi Zhang, Tao Jia
Abstract
Semi-supervised cardiac image segmentation aims to exploit abundant unlabeled images together with limited annotations, but the reliability of pseudo-label supervision is highly uneven across spatial regions. In a dual-network setting, interior regions are usually more stable, whereas boundary regions, low-confidence areas, and disagreements between networks can introduce noisy supervision. However, existing methods typically impose uniform supervision on all unlabeled pixels, overlooking reliability differences and limiting discriminative learning in challenging areas. To address this issue, we propose R2CL (Reliability-Role Collaborative Learning), a framework for semi-supervised cardiac image segmentation. R2CL models confidence relationships between two networks to assign roles to unlabeled regions: mutually uncertain regions are used for entropy regularization, confidence-discrepant regions for directional knowledge transfer, and mutually confident regions as reliable supervision anchors. Reliability-Role Mixed Cross Learning (RR-MCL) implements these roles through region-mixed training, directional cross-network supervision, and entropy regularization over mutually uncertain regions. Reliability-Transition Prototype Contrastive Learning (RT-PCL) constructs query features from foreground-adjacent reliability-transition regions and learns network-specific class prototypes from mutually confident regions, improving feature discrimination around ambiguous boundaries. High-confidence Context Refinement (HCR) exploits reliable interior pixels and trusted dense blocks within mutually confident regions to complement stable contextual supervision. R2CL achieves stable improvements in region-overlap and boundary-related metrics across multiple datasets, with more evident advantages on the foreground-sparse MVAA-CT dataset.