RAP: Retrieve, Adapt, and Prompt-Fit for Training-Free Few-Shot Medical Image Segmentation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mao, Zhihao, Chen, Bangpu
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917366286254080
author Mao, Zhihao
Chen, Bangpu
author_facet Mao, Zhihao
Chen, Bangpu
contents Few-shot medical image segmentation (FSMIS) has achieved notable progress, yet most existing methods mainly rely on semantic correspondences from scarce annotations while under-utilizing a key property of medical imagery: anatomical targets exhibit repeatable high-frequency morphology (e.g., boundary geometry and spatial layout) across patients and acquisitions. We propose RAP, a training-free framework that retrieves, adapts, and prompts Segment Anything Model 2 (SAM2) for FSMIS. First, RAP retrieves morphologically compatible supports from an archive using DINOv3 features to reduce brittleness in single-support choice. Second, it adapts the retrieved support mask to the query by fitting boundary-aware structural cues, yielding an anatomy-consistent pre-mask under domain shifts. Third, RAP converts the pre-mask into prompts by sampling positive points via Voronoi partitioning and negative points via sector-based sampling, and feeds them into SAM2 for final refinement without any fine-tuning. Extensive experiments on multiple medical segmentation benchmarks show that RAP consistently surpasses prior FSMIS baselines and achieves state-of-the-art performance. Overall, RAP demonstrates that explicit structural fitting combined with retrieval-augmented prompting offers a simple and effective route to robust training-free few-shot medical segmentation.
format Preprint
id arxiv_https___arxiv_org_abs_2603_27705
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle RAP: Retrieve, Adapt, and Prompt-Fit for Training-Free Few-Shot Medical Image Segmentation
Mao, Zhihao
Chen, Bangpu
Computer Vision and Pattern Recognition
Artificial Intelligence
Few-shot medical image segmentation (FSMIS) has achieved notable progress, yet most existing methods mainly rely on semantic correspondences from scarce annotations while under-utilizing a key property of medical imagery: anatomical targets exhibit repeatable high-frequency morphology (e.g., boundary geometry and spatial layout) across patients and acquisitions. We propose RAP, a training-free framework that retrieves, adapts, and prompts Segment Anything Model 2 (SAM2) for FSMIS. First, RAP retrieves morphologically compatible supports from an archive using DINOv3 features to reduce brittleness in single-support choice. Second, it adapts the retrieved support mask to the query by fitting boundary-aware structural cues, yielding an anatomy-consistent pre-mask under domain shifts. Third, RAP converts the pre-mask into prompts by sampling positive points via Voronoi partitioning and negative points via sector-based sampling, and feeds them into SAM2 for final refinement without any fine-tuning. Extensive experiments on multiple medical segmentation benchmarks show that RAP consistently surpasses prior FSMIS baselines and achieves state-of-the-art performance. Overall, RAP demonstrates that explicit structural fitting combined with retrieval-augmented prompting offers a simple and effective route to robust training-free few-shot medical segmentation.
title RAP: Retrieve, Adapt, and Prompt-Fit for Training-Free Few-Shot Medical Image Segmentation
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2603.27705