Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Feature-prompting GBMSeg: One-Shot Reference Guided Training-Free Prompt Engineering for Glomerular Basement Membrane Segmentation

Xueyu Liu, Guangze Shi, Rui Wang, Yexin Lai, Jianan Zhang, Lele Sun, Quan Yang, Yongfei Wu, MIng Li, Weixia Han, Wen Zheng

TL;DR

GBMSeg addresses the challenge of segmenting the glomerular basement membrane (GBM) in TEM images without training by using a one-shot reference to guide a training-free framework. It combines patch-level feature matching with automatic prompt engineering to generate high-quality positive and negative prompt points for the Segment Anything Model (SAM), leveraging DINOv2 features for robust cross-image correspondence. On a dataset of 2538 TEM images, GBMSeg achieves a Dice Similarity Coefficient of 87.27% with only one labeled reference, outperforming both one-shot and few-shot methods as well as other training-free approaches. This work demonstrates a practical, domain-independent segmentation pipeline with potential clinical impact, and suggests future work on automated GBM thickness measurement and pathological indicator quantification.

Abstract

Assessment of the glomerular basement membrane (GBM) in transmission electron microscopy (TEM) is crucial for diagnosing chronic kidney disease (CKD). The lack of domain-independent automatic segmentation tools for the GBM necessitates an AI-based solution to automate the process. In this study, we introduce GBMSeg, a training-free framework designed to automatically segment the GBM in TEM images guided only by a one-shot annotated reference. Specifically, GBMSeg first exploits the robust feature matching capabilities of the pretrained foundation model to generate initial prompt points, then introduces a series of novel automatic prompt engineering techniques across the feature and physical space to optimize the prompt scheme. Finally, GBMSeg employs a class-agnostic foundation segmentation model with the generated prompt scheme to obtain accurate segmentation results. Experimental results on our collected 2538 TEM images confirm that GBMSeg achieves superior segmentation performance with a Dice similarity coefficient (DSC) of 87.27% using only one labeled reference image in a training-free manner, outperforming recently proposed one-shot or few-shot methods. In summary, GBMSeg introduces a distinctive automatic prompt framework that facilitates robust domain-independent segmentation performance without training, particularly advancing the automatic prompting of foundation segmentation models for medical images. Future work involves automating the thickness measurement of segmented GBM and quantifying pathological indicators, holding significant potential for advancing pathology assessments in clinical applications. The source code is available on https://github.com/SnowRain510/GBMSeg

Feature-prompting GBMSeg: One-Shot Reference Guided Training-Free Prompt Engineering for Glomerular Basement Membrane Segmentation

TL;DR

GBMSeg addresses the challenge of segmenting the glomerular basement membrane (GBM) in TEM images without training by using a one-shot reference to guide a training-free framework. It combines patch-level feature matching with automatic prompt engineering to generate high-quality positive and negative prompt points for the Segment Anything Model (SAM), leveraging DINOv2 features for robust cross-image correspondence. On a dataset of 2538 TEM images, GBMSeg achieves a Dice Similarity Coefficient of 87.27% with only one labeled reference, outperforming both one-shot and few-shot methods as well as other training-free approaches. This work demonstrates a practical, domain-independent segmentation pipeline with potential clinical impact, and suggests future work on automated GBM thickness measurement and pathological indicator quantification.

Abstract

Assessment of the glomerular basement membrane (GBM) in transmission electron microscopy (TEM) is crucial for diagnosing chronic kidney disease (CKD). The lack of domain-independent automatic segmentation tools for the GBM necessitates an AI-based solution to automate the process. In this study, we introduce GBMSeg, a training-free framework designed to automatically segment the GBM in TEM images guided only by a one-shot annotated reference. Specifically, GBMSeg first exploits the robust feature matching capabilities of the pretrained foundation model to generate initial prompt points, then introduces a series of novel automatic prompt engineering techniques across the feature and physical space to optimize the prompt scheme. Finally, GBMSeg employs a class-agnostic foundation segmentation model with the generated prompt scheme to obtain accurate segmentation results. Experimental results on our collected 2538 TEM images confirm that GBMSeg achieves superior segmentation performance with a Dice similarity coefficient (DSC) of 87.27% using only one labeled reference image in a training-free manner, outperforming recently proposed one-shot or few-shot methods. In summary, GBMSeg introduces a distinctive automatic prompt framework that facilitates robust domain-independent segmentation performance without training, particularly advancing the automatic prompting of foundation segmentation models for medical images. Future work involves automating the thickness measurement of segmented GBM and quantifying pathological indicators, holding significant potential for advancing pathology assessments in clinical applications. The source code is available on https://github.com/SnowRain510/GBMSeg
Paper Structure (11 sections, 1 equation, 3 figures, 2 tables)

This paper contains 11 sections, 1 equation, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Patch-level feature extraction
  4. Automatic prompt engineering
  5. Experimental Results
  6. Dataset preparation
  7. Implementation Detail
  8. Ablation study
  9. Performance comparison with few-shot and one-shot methods
  10. Conclusion
  11. Disclosure of Interests.

Figures (3)

  • Figure 1: The workflow of GBMSeg, the one-shot reference guided training-free framework, automates the segmentation of the GBM through three components: Patch-level feature extraction, automatic prompt engineering, and GBM segmentation.
  • Figure 2: A process for automatic prompt engineering. As the prompt engineering is refined, the prompt scheme is gradually optimized.
  • Figure 3: Different stages of the automatic prompt engineering yield positive and negative prompt schemes, alongside corresponding GBM segmentation results. Notably, as the components of automatic prompt engineering are refined, the segmentation results steadily converge toward the ground truth.