Self-attentive adversarial stain normalization

Authors

Aman Shrivastava, University of Virginia
William Adorno, University of Virginia
Lubaina Ehsan Lubaina Ehsan, University of Virginia
Syed Asad Ali, Aga Khan UniversityFollow
Sean R. Moore, University of Virginia
Beatrice C. Amadi, University of Zambia
Paul Kelly, University of Zambia
Sana Syed, University of Virginia
Donald E. Brown, University of Virginia

Document Type

Article

Department

Paediatrics and Child Health; Women and Child Health

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.

Comments

Volume, issue, and pagination are not provided by the author/publisher

Publication (Name of Journal)

arXiv preprint arXiv

Recommended Citation

Shrivastava, A., Adorno, W., Lubaina Ehsan, L., Ali, S., Moore, S. R., Amadi, B. C., Kelly, P., Syed, S., Brown, D. E. (2019). Self-attentive adversarial stain normalization. arXiv preprint arXiv.
Available at: https://ecommons.aku.edu/pakistan_fhs_mc_women_childhealth_paediatr/864