ProPL: Universal Semi-Supervised Ultrasound Image Segmentation via Prompt-Guided Pseudo-Labeling

2025-11-19

AI Summary

Existing ultrasound image segmentation methods suffer from poor generalizability and limited cross-organ and multi-task transferability. Method: This paper proposes the first general-purpose semi-supervised ultrasound segmentation framework, featuring a shared visual encoder and a prompt-driven dual-decoder architecture for flexible task adaptation, along with a novel Uncertainty-guided Pseudo-Label Calibration (UPLC) module to enhance the reliability of unlabeled data utilization. Contribution/Results: Trained jointly on multi-source ultrasound data spanning five organs and eight segmentation tasks, the framework significantly outperforms state-of-the-art supervised and semi-supervised methods across multiple metrics. It establishes a new benchmark for generalizable ultrasound segmentation and provides a practical pathway for broad clinical deployment under low-labeling-cost constraints.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Semi-Supervised Learning
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Existing approaches for the problem of ultrasound image segmentation, whether supervised or semi-supervised, are typically specialized for specific anatomical structures or tasks, limiting their practical utility in clinical settings. In this paper, we pioneer the task of universal semi-supervised ultrasound image segmentation and propose ProPL, a framework that can handle multiple organs and segmentation tasks while leveraging both labeled and unlabeled data. At its core, ProPL employs a shared vision encoder coupled with prompt-guided dual decoders, enabling flexible task adaptation through a prompting-upon-decoding mechanism and reliable self-training via an uncertainty-driven pseudo-label calibration (UPLC) module. To facilitate research in this direction, we introduce a comprehensive ultrasound dataset spanning 5 organs and 8 segmentation tasks. Extensive experiments demonstrate that ProPL outperforms state-of-the-art methods across various metrics, establishing a new benchmark for universal ultrasound image segmentation.

Problem

• Universal semi-supervised ultrasound image segmentation for multiple organs
• Overcoming task-specific limitations in clinical ultrasound segmentation
• Leveraging labeled and unlabeled data across diverse anatomical structures

Innovation

• Prompt-guided dual decoders enable flexible task adaptation
• Uncertainty-driven pseudo-label calibration ensures reliable self-training
• Shared vision encoder processes multiple organs and tasks