One Polyp Identifies All: One-Shot Polyp Segmentation with SAM via Cascaded Priors and Iterative Prompt Evolution

2025-07-22

AI Summary

Colorectal polyp segmentation faces challenges including high morphological variability, severe domain shift, and prohibitively high annotation costs. Method: This paper proposes the first SAM adaptation framework leveraging automatically generated single-sample prompts. It introduces a novel correlation-prior-guided interactive prompt generation mechanism, integrated with scale-cascaded prior fusion and Euclidean-space prompt evolution—enabling iterative prompt optimization without additional annotations. Semantic transfer and noise filtering are further incorporated to enhance robustness. Contribution/Results: Evaluated on five public datasets, the method achieves an IoU of 76.93% on Kvasir—surpassing the state-of-the-art by 11.44%. It significantly improves generalization across domains and clinical deployability, establishing a new benchmark for efficient, low-annotation polyp segmentation.

Technology Category

• Computer Vision: Segmentation
• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Security and Privacy: Data transparency and provenance
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Polyp segmentation is vital for early colorectal cancer detection, yet traditional fully supervised methods struggle with morphological variability and domain shifts, requiring frequent retraining. Additionally, reliance on large-scale annotations is a major bottleneck due to the time-consuming and error-prone nature of polyp boundary labeling. Recently, vision foundation models like Segment Anything Model (SAM) have demonstrated strong generalizability and fine-grained boundary detection with sparse prompts, effectively addressing key polyp segmentation challenges. However, SAM's prompt-dependent nature limits automation in medical applications, since manually inputting prompts for each image is labor-intensive and time-consuming. We propose OP-SAM, a One-shot Polyp segmentation framework based on SAM that automatically generates prompts from a single annotated image, ensuring accurate and generalizable segmentation without additional annotation burdens. Our method introduces Correlation-based Prior Generation (CPG) for semantic label transfer and Scale-cascaded Prior Fusion (SPF) to adapt to polyp size variations as well as filter out noisy transfers. Instead of dumping all prompts at once, we devise Euclidean Prompt Evolution (EPE) for iterative prompt refinement, progressively enhancing segmentation quality. Extensive evaluations across five datasets validate OP-SAM's effectiveness. Notably, on Kvasir, it achieves 76.93% IoU, surpassing the state-of-the-art by 11.44%.

Problem

• Automates polyp segmentation without manual prompts
• Reduces reliance on large-scale annotated datasets
• Adapts to polyp size variations and noisy data

Innovation

• Automated prompt generation from single annotated image
• Cascaded priors for size adaptation and noise filtering
• Iterative prompt refinement for enhanced segmentation quality