CoDiR: Confidence-Guided Diffusion Refinement for Semi-Supervised Histopathology Segmentation

2026-08-12

AI Summary

This work addresses the challenges of scarce annotations and unreliable pseudo-labels in ambiguous gland regions for semi-supervised histopathology image segmentation. To this end, we propose a confidence-guided diffusion refinement mechanism built upon the Mean Teacher framework. In regions where the teacher model exhibits low prediction confidence, a conditional diffusion model is introduced to perform structure-aware refinement, while high-confidence predictions are leveraged to formulate a weighted consistency loss for training the student model. The proposed approach substantially enhances pseudo-label quality, achieving mDice scores of 88.09%/89.83% on the GlaS dataset and 89.19%/90.29% on the CRAG dataset using only 10% and 20% labeled data, respectively. Notably, the diffusion refinement module alone contributes a +6.36% mDice improvement, consistently outperforming current state-of-the-art methods.

Abstract

Semi-supervised histopathology segmentation is challenging due to scarce annotations and unreliable pseudo-labels in ambiguous gland regions. To address this problem, we propose Confidence-Guided Diffusion Refinement (CoDiR), a semi-supervised framework that combines a Mean Teacher segmentation model with diffusion-based pseudo-label refinement. Given an unlabeled image, the teacher first produces a soft prediction, and only low-confidence regions are refined by a conditional diffusion model trained to capture plausible mask structures from labeled data. The refined mask is then fused with reliable teacher predictions and used to train the student with confidence weighting and consistency regularization. On the GlaS and CRAG datasets CoDiR reaches 88.09\% and 89.83\% mDice with 10\% labeled data, and 89.19\% and 90.29\% mDice with 20\%, matching or exceeding the strongest published method on seven of the eight benchmark metrics. Ablations attribute the largest single contribution to the refinement module, which adds +6.36\% mDice over the Mean Teacher baseline. The implementation code is publicly available at: https://github.com/vongla345/codir

Problem

• semi-supervised learning
• histopathology segmentation
• pseudo-labeling
• ambiguous regions
• scarce annotations

Innovation

• semi-supervised learning
• diffusion model
• histopathology segmentation
• pseudo-label refinement
• confidence-guided