Fully Differentiable Bidirectional Dual-Task Synergistic Learning for Semi-Supervised 3D Medical Image Segmentation

2026-02-01

Expert systems with applications

AI Summary

This work addresses the scarcity of annotated data in medical image segmentation by proposing a fully differentiable bidirectional co-learning framework. For the first time, it enables online, bidirectional, and differentiable interaction between segmentation and regression tasks, overcoming the limitations of existing unidirectional co-learning approaches. The method seamlessly integrates supervised learning, consistency regularization, pseudo-labeling, and uncertainty estimation within a unified architecture, effectively leveraging unlabeled data through dual-task synergy. Evaluated on two benchmark 3D medical image datasets, the approach achieves state-of-the-art performance, establishing a novel and generalizable paradigm for semi-supervised medical image segmentation and multi-task learning.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Semi-Supervised Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Problem

• semi-supervised learning
• 3D medical image segmentation
• dual-task collaboration
• bidirectional interaction
• label scarcity

Innovation

• bidirectional synergistic learning
• fully differentiable framework
• dual-task collaboration
• semi-supervised segmentation
• consistency regularization