VCDP: Variation-Conditioned Distributional Proxy Learning for Semi-Supervised Medical Image Segmentation

2026-07-08

AI Summary

This work addresses the challenge of semi-supervised medical image segmentation in anatomically complex regions, particularly small organs and areas with ambiguous boundaries, where large intra-class variation hinders effective feature modeling. The authors propose a plug-and-play regularization module for the training phase that, for the first time, integrates intra-class variation modeling with distributional proxy learning. Specifically, semantic class representations are formulated as learnable Gaussian distributions, while multiple variation prototypes capture fine-grained intra-class diversity. A variation-conditioned compatibility mechanism—incurred without additional inference cost—guides voxel embeddings to align simultaneously with global organ identity and local anatomical variations. Evaluated on multiple abdominal multi-organ segmentation benchmarks, the method significantly outperforms state-of-the-art semi-supervised approaches, achieving notably improved performance especially on small structures, boundary-ambiguous regions, and areas exhibiting high intra-class variability.

Abstract

Semi-supervised 3D medical image segmentation reduces the need for dense voxel-level annotations by exploiting unlabeled volumes. Although existing methods such as consistency regularization, pseudo-labeling, and co-training improve prediction-level robustness, they often provide insufficient feature-space organization for anatomically complex structures, especially small organs and ambiguous boundary regions with large intra-class variations. To address this issue, we propose Variation-Conditioned Distributional Proxy Learning (VCDP), a plug-and-play training-only regularization module for semi-supervised 3D medical image segmentation. VCDP represents each class with a learnable Gaussian distribution for shared class semantics and multiple variation prototypes for fine-grained intra-class patterns. A unified variation-conditioned compatibility score is further formulated to fuse distributional similarity and soft variation aggregation, guiding voxel embeddings to align with both global organ identity and local anatomical variations. VCDP is attached to decoder features during training and removed during inference, introducing no additional inference cost. Experiments on multi-organ segmentation benchmarks show that VCDP improves most evaluated baselines, particularly for small, ambiguous, and highly variable organs. Our anonymous code is released at https://anonymous.4open.science/r/VCDP_code-41ED.

Problem

• semi-supervised segmentation
• medical image segmentation
• intra-class variation
• anatomical ambiguity
• small organs

Innovation

• semi-supervised segmentation
• distributional proxy learning
• variation-conditioned compatibility
• 3D medical image segmentation
• feature-space regularization