CPS4: Class Prompt driven Semi-Supervised Spine Segmentation with Class-specific Consistency Constraint

2026-06-14

AI Summary

This work addresses the limitation of existing vision-language models in semi-supervised spinal segmentation, where the absence of explicit semantic alignment between textual class prompts and vertebral regions leads to poor multi-class pseudo-label quality. To overcome this, the authors propose CPS4, the first approach to integrate class-prompt-driven vision-language modeling into this task. During pretraining, token-level and pixel-level attention losses are introduced to enhance semantic consistency between prompts and corresponding image regions. The pretrained encoder is then leveraged to generate class-specific binary segmentation maps for unlabeled images, which are subsequently fused into a unified multi-class segmentation map. Evaluated on a public dataset using only 5% labeled data, the method achieves a Dice coefficient of 80.44%, substantially outperforming current semi-supervised and vision-language baselines.

Abstract

Vision Language Model (VLM) has great potential to enhance the quality of pseudo labels in semi-supervised spine segmentation by leveraging textual class prompts to generate segmentation map, but no one has studied it yet. Although promising, it lacks explicit constraints to ensure consistency between spine class prompts and spine unit region, resulting in unsatisfactory performance in multi-class segmentation map generation. In this paper, we propose CPS4, the first text-guided semi-supervised spine segmentation network using class prompts to enhance the quality of spine pseudo labels. Specifically, CPS4 is implemented through two training stages. (i) Class-specific consistency constrained VLM pretraining stage: we propose token- and pixel-level attention loss to optimize the consistency between class prompts and spine units, forcing the textual class prompt to be closely coupled with the target spine unit in the semantic space. (ii) Class Prompt driven semi-supervised spine segmentation stage: using the pretrained vision-text encoder, we derive each class-specific binary segmentation map for the unlabeled spine image and integrate them into an unified multi-class segmentation map, improving the quality of the spine pseudo label generated by the semi-supervised spine segmentation network. Experimental results show that our CPS4 achieves superior spine segmentation performance with Dice of 80.44%, only using 5% labeled data on the public spine segmentation dataset, surpassing popular semi-supervised learning and VLM methods. Our code will be available.

Problem

• semi-supervised segmentation
• spine segmentation
• vision language model
• class prompt
• pseudo label

Innovation

• Class Prompt
• Semi-Supervised Segmentation
• Vision Language Model
• Consistency Constraint
• Spine Segmentation