Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Utilizing Weak-to-Strong Consistency for Semi-Supervised Glomeruli Segmentation

Irina Zhang, Jim Denholm, Azam Hamidinekoo, Oskar Ålund, Christopher Bagnall, Joana Palés Huix, Michal Sulikowski, Ortensia Vito, Arthur Lewis, Robert Unwin, Magnus Soderberg, Nikolay Burlutskiy, Talha Qaiser

TL;DR

The authors' experimental results indicate superior performance of the semi-supervised learning approach for glomeruli segmentation based on the weak-to-strong consistency framework validated on multiple real-world datasets as compared with existing supervised baseline models such as U-Net and SegFormer.

Abstract

Accurate segmentation of glomerulus instances attains high clinical significance in the automated analysis of renal biopsies to aid in diagnosing and monitoring kidney disease. Analyzing real-world histopathology images often encompasses inter-observer variability and requires a labor-intensive process of data annotation. Therefore, conventional supervised learning approaches generally achieve sub-optimal performance when applied to external datasets. Considering these challenges, we present a semi-supervised learning approach for glomeruli segmentation based on the weak-to-strong consistency framework validated on multiple real-world datasets. Our experimental results on 3 independent datasets indicate superior performance of our approach as compared with existing supervised baseline models such as U-Net and SegFormer.

Utilizing Weak-to-Strong Consistency for Semi-Supervised Glomeruli Segmentation

TL;DR

The authors' experimental results indicate superior performance of the semi-supervised learning approach for glomeruli segmentation based on the weak-to-strong consistency framework validated on multiple real-world datasets as compared with existing supervised baseline models such as U-Net and SegFormer.

Abstract

Accurate segmentation of glomerulus instances attains high clinical significance in the automated analysis of renal biopsies to aid in diagnosing and monitoring kidney disease. Analyzing real-world histopathology images often encompasses inter-observer variability and requires a labor-intensive process of data annotation. Therefore, conventional supervised learning approaches generally achieve sub-optimal performance when applied to external datasets. Considering these challenges, we present a semi-supervised learning approach for glomeruli segmentation based on the weak-to-strong consistency framework validated on multiple real-world datasets. Our experimental results on 3 independent datasets indicate superior performance of our approach as compared with existing supervised baseline models such as U-Net and SegFormer.

Paper Structure

This paper contains 30 sections, 4 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Glomeruli localisation
  4. Semi-supervised learning
  5. Methods
  6. Supervised baseline models
  7. Weak-to-strong augmentation
  8. Semi-supervised glomeruli segmentation
  9. Experimental Settings
  10. Data
  11. HuBMAP - Hacking the Kidney
  12. HuBMAP - Hacking the Human Vasculature
  13. Kidney Precision Medicine Project
  14. NURTuRE
  15. Implementation Details
  16. ...and 15 more sections

Figures (4)

  • Figure 1: Model Diagram. Can you mark directly on the figure where the weak and strong augmentation is applied, and possibly include examples of the augmented patches?
  • Figure 2: tiles to show heterogeneity across these datasets
  • Figure 3: segmentation results of different models
  • Figure 4: Model Diagram. Can you mark directly on the figure where the weak and strong augmentation is applied, and possibly include examples of the augmented patches?