Self-Attentive Adversarial Stain Normalization

TL;DR

This paper introduces SAASN, a self-attentive adversarial method for stain normalization in histopathology images, reducing variability across labs and improving consistency for diagnosis and deep learning models.

Contribution

SAASN is a novel unsupervised generative adversarial approach that incorporates self-attention to enhance detail and structural preservation during stain normalization.

Findings

SAASN outperforms existing stain normalization methods.

It produces more detailed and structurally consistent images.

Demonstrates robustness across different biopsy datasets.

Abstract

Hematoxylin and Eosin (H&E) stained Whole Slide Images (WSIs) are utilized for biopsy visualization-based diagnostic and prognostic assessment of diseases. Variation in the H&E staining process across different lab sites can lead to significant variations in biopsy image appearance. These variations introduce an undesirable bias when the slides are examined by pathologists or used for training deep learning models. To reduce this bias, slides need to be translated to a common domain of stain appearance before analysis. We propose a Self-Attentive Adversarial Stain Normalization (SAASN) approach for the normalization of multiple stain appearances to a common domain. This unsupervised generative adversarial approach includes self-attention mechanism for synthesizing images with finer detail while preserving the structural consistency of the biopsy features during translation. SAASN demonstrates consistent and superior performance compared to other popular stain normalization techniques on H&E stained duodenal biopsy image data.