Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Fast, Unsupervised Framework for Registration Quality Assessment of Multi-stain Histological Whole Slide Pairs

Shikha Dubey, Patricia Raciti, Kristopher Standish, Albert Juan Ramon, Erik Ames Burlingame

TL;DR

This work tackles the challenge of assessing registration quality between multi-stain histological whole-slide images without ground-truth annotations. It introduces the Unsupervised Registration Quality Assessment (URQA) framework, which jointly uses masks-based geometric alignment and deformations-based regularity to compute a unified quality score Q. The method comprises a Masks-based RQA (MRQA) module, a Deformations-Based RQA (DRQA) module, and a fusion rule that yields an interpretable Pass/Fail outcome suitable for large-scale quality control; validated on 300 H&E–IHC WSI pairs with expert input. The approach enables GT-free, real-time quality control with scalable computational performance, and shows promise for extension to other registration models and downstream analyses such as cell-type mapping and multiplex staining.

Abstract

High-fidelity registration of histopathological whole slide images (WSIs), such as hematoxylin & eosin (H&E) and immunohistochemistry (IHC), is vital for integrated molecular analysis but challenging to evaluate without ground-truth (GT) annotations. Existing WSI-level assessments -- using annotated landmarks or intensity-based similarity metrics -- are often time-consuming, unreliable, and computationally intensive, limiting large-scale applicability. This study proposes a fast, unsupervised framework that jointly employs down-sampled tissue masks- and deformations-based metrics for registration quality assessment (RQA) of registered H&E and IHC WSI pairs. The masks-based metrics measure global structural correspondence, while the deformations-based metrics evaluate local smoothness, continuity, and transformation realism. Validation across multiple IHC markers and multi-expert assessments demonstrate a strong correlation between automated metrics and human evaluations. In the absence of GT, this framework offers reliable, real-time RQA with high fidelity and minimal computational resources, making it suitable for large-scale quality control in digital pathology.

Fast, Unsupervised Framework for Registration Quality Assessment of Multi-stain Histological Whole Slide Pairs

TL;DR

This work tackles the challenge of assessing registration quality between multi-stain histological whole-slide images without ground-truth annotations. It introduces the Unsupervised Registration Quality Assessment (URQA) framework, which jointly uses masks-based geometric alignment and deformations-based regularity to compute a unified quality score Q. The method comprises a Masks-based RQA (MRQA) module, a Deformations-Based RQA (DRQA) module, and a fusion rule that yields an interpretable Pass/Fail outcome suitable for large-scale quality control; validated on 300 H&E–IHC WSI pairs with expert input. The approach enables GT-free, real-time quality control with scalable computational performance, and shows promise for extension to other registration models and downstream analyses such as cell-type mapping and multiplex staining.

Abstract

High-fidelity registration of histopathological whole slide images (WSIs), such as hematoxylin & eosin (H&E) and immunohistochemistry (IHC), is vital for integrated molecular analysis but challenging to evaluate without ground-truth (GT) annotations. Existing WSI-level assessments -- using annotated landmarks or intensity-based similarity metrics -- are often time-consuming, unreliable, and computationally intensive, limiting large-scale applicability. This study proposes a fast, unsupervised framework that jointly employs down-sampled tissue masks- and deformations-based metrics for registration quality assessment (RQA) of registered H&E and IHC WSI pairs. The masks-based metrics measure global structural correspondence, while the deformations-based metrics evaluate local smoothness, continuity, and transformation realism. Validation across multiple IHC markers and multi-expert assessments demonstrate a strong correlation between automated metrics and human evaluations. In the absence of GT, this framework offers reliable, real-time RQA with high fidelity and minimal computational resources, making it suitable for large-scale quality control in digital pathology.
Paper Structure (12 sections, 10 equations, 4 figures, 3 tables)

This paper contains 12 sections, 10 equations, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Masks-based RQA (MRQA) Module
  4. Deformations-Based RQA (DRQA) Module
  5. Unified Quality Score by URQA
  6. Experimentations and Results
  7. Dataset and Preprocessing
  8. Results and Discussions
  9. Conclusion and Future Work
  10. Acknowledgments
  11. Compliance with Ethical Standards
  12. Disclosure of Interest

Figures (4)

  • Figure 1: URQA Framework. Mag.: magnitude, Hist: histogram.
  • Figure 2: (A) and (B) show both modules of URQA. Reg.: Registration; Hist. Corr.: Histogram Correlation; Lower-res.: Lower-resolution; Smooth. Residual: Smoothness residual.
  • Figure 3: Confusion matrices: (a) Binary RQA (0:Fail, Pass:1); (b) Cardinal RQA (0:Poor, 1:Fair, 2:Good, 3:Excellent); (c) Comparison of binary and cardinal assessments between experts.
  • Figure 4: Qualitative RQA examples: (a–c) align with both experts; (d–e) show partial agreement; (f) no agreement with experts. Each displays tissue masks, deformation directions, and Jacobian maps; (e) highlights poor deformations at 20x.