Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DeReStainer: H&E to IHC Pathological Image Translation via Decoupled Staining Channels

Linda Wei, Shengyi Hua, Shaoting Zhang, Xiaofan Zhang

TL;DR

The paper addresses generating IHC images from H&E by introducing DeReStainer, a destain-restain framework that exploits the shared Hematoxylin channel between H&E and IHC. It decouples staining channels via a staining-separation module, aligns $H$-channel representations, and uses a ModConv-based feature fusion along with a DAB-focused loss to preserve HER2 semantic information. Coupled with a GAN-based ReStainer and auxiliary comparator/semantic losses, the method achieves superior SSIM and HER2-level accuracy on the BCI dataset, outperforming prior I2IT approaches. The results suggest practical potential for cost-efficient, semantically faithful IHC generation with clinically relevant HER2 predictions from standard H&E slides.

Abstract

Breast cancer is a highly fatal disease among cancers in women, and early detection is crucial for treatment. HER2 status, a valuable diagnostic marker based on Immunohistochemistry (IHC) staining, is instrumental in determining breast cancer status. The high cost of IHC staining and the ubiquity of Hematoxylin and Eosin (H&E) staining make the conversion from H&E to IHC staining essential. In this article, we propose a destain-restain framework for converting H&E staining to IHC staining, leveraging the characteristic that H&E staining and IHC staining of the same tissue sections share the Hematoxylin channel. We further design loss functions specifically for Hematoxylin and Diaminobenzidin (DAB) channels to generate IHC images exploiting insights from separated staining channels. Beyond the benchmark metrics on BCI contest, we have developed semantic information metrics for the HER2 level. The experimental results demonstrated that our method outperforms previous open-sourced methods in terms of image intrinsic property and semantic information.

DeReStainer: H&E to IHC Pathological Image Translation via Decoupled Staining Channels

TL;DR

The paper addresses generating IHC images from H&E by introducing DeReStainer, a destain-restain framework that exploits the shared Hematoxylin channel between H&E and IHC. It decouples staining channels via a staining-separation module, aligns -channel representations, and uses a ModConv-based feature fusion along with a DAB-focused loss to preserve HER2 semantic information. Coupled with a GAN-based ReStainer and auxiliary comparator/semantic losses, the method achieves superior SSIM and HER2-level accuracy on the BCI dataset, outperforming prior I2IT approaches. The results suggest practical potential for cost-efficient, semantically faithful IHC generation with clinically relevant HER2 predictions from standard H&E slides.

Abstract

Breast cancer is a highly fatal disease among cancers in women, and early detection is crucial for treatment. HER2 status, a valuable diagnostic marker based on Immunohistochemistry (IHC) staining, is instrumental in determining breast cancer status. The high cost of IHC staining and the ubiquity of Hematoxylin and Eosin (H&E) staining make the conversion from H&E to IHC staining essential. In this article, we propose a destain-restain framework for converting H&E staining to IHC staining, leveraging the characteristic that H&E staining and IHC staining of the same tissue sections share the Hematoxylin channel. We further design loss functions specifically for Hematoxylin and Diaminobenzidin (DAB) channels to generate IHC images exploiting insights from separated staining channels. Beyond the benchmark metrics on BCI contest, we have developed semantic information metrics for the HER2 level. The experimental results demonstrated that our method outperforms previous open-sourced methods in terms of image intrinsic property and semantic information.
Paper Structure (11 sections, 6 equations, 3 figures, 1 table)

This paper contains 11 sections, 6 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methods
  3. DeStainer
  4. Feature Fusion Module
  5. ReStainer
  6. Loss Functions
  7. Experiment
  8. Dataset and Implementation Details
  9. Results
  10. Ablation Study
  11. Conclusion

Figures (3)

  • Figure 1: (a): Examples of H&E and IHC pairs of different HER2 levels.(b): Stain Separation for H&E and IHC image. The first column is the original images. The second column is the shared H channel. The third column is the other channel, E channel for H&E, and DAB channel for IHC
  • Figure 2: The overall architecture of our framework. DeStainer performs stain separation and aligns the H channel of H&E with that of IHC images. The Feature Fusion Module integrates the output of DeStainer with the classification information of the HER2 level. Restainer is used to generate the corresponding IHC images. In the framework, y is the HER2-level label of the input IHC. $f$ represents a simple network, C and D are comparator and discriminator respectively.
  • Figure 3: Visualization of generated IHC with all HER2 levels of different methods