Beyond Visual Cues: CoT-Enhanced Reasoning for Semi-supervised Medical Image Segmentation

2026-06-16

AI Summary

This work addresses the limitations of existing semi-supervised medical image segmentation methods, which often rely excessively on visual similarity and consequently suffer from semantic inconsistency or ambiguous boundaries in clinically relevant scenarios where visual and semantic cues mismatch. To overcome this challenge, the study introduces Chain-of-Thought (CoT) reasoning into the task for the first time, proposing the CERS framework. CERS leverages large language models to generate pathology-aware reasoning descriptions, establishes a semantic-aware reference selection mechanism, and incorporates a multi-scale coordinate attention module (MCAM) to integrate reasoning context into the decoding process. The proposed approach effectively mitigates visual-semantic misalignment, achieving significant performance gains over state-of-the-art methods across multiple benchmarks, particularly demonstrating superior accuracy and robustness in regions with ambiguous boundaries or complex semantic content.

Abstract

Semi-supervised medical image segmentation has emerged as a dominant research problem in medical image analysis, mitigating annotation scarcity by leveraging consistency regularization on unlabeled data. However, existing approaches operate predominantly via visual pattern matching, relying heavily on pixel-level similarities. This visual-centric dependency often falters in clinical scenarios characterized by the visual-semantic mismatch, where visually similar lesions warrant distinct diagnostic conclusions, thus failing to capture the underlying diagnostic logic used by experts. To address this, we move beyond visual cues and propose CERS (CoT-Enhanced Reasoning Segmentation), a framework that integrates Chain-of-Thought (CoT) reasoning to distinguish pathologically distinct cases. Specifically, we construct a knowledge pool enriched with linguistic reasoning descriptions generated by large language models (LLMs). A semantic-aware reference selection strategy is introduced to identify historical evidence, filtering candidates first by morphology, and then refining them via CoT consistency to eliminate hard negatives. Furthermore, a multi-scale coordinate attention module (MCAM) is designed to effectively fuse this reasoning-derived context into the decoding process. Extensive experiments demonstrate the superiority of CERS against state-of-the-art approaches, particularly in resolving boundary ambiguities and semantic inconsistencies. The code is available at https://github.com/cymasuna/CERS.

Problem

• semi-supervised medical image segmentation
• visual-semantic mismatch
• diagnostic reasoning
• lesion ambiguity
• annotation scarcity

Innovation

• Chain-of-Thought reasoning
• semi-supervised segmentation
• medical image analysis
• semantic-aware reference selection
• multi-scale coordinate attention