44 2033180199
Official Journal of
  • image

Sign up for email alert when new content gets added: Sign up

Computational Histological Staining and Destaining of RGB Images from Prostate Core Biopsies Using Generative Adversarial Neural Networks

Author(s): Nick Wells

Histopathology tissue samples are commonly accessible in two forms: paraffin-embedded unstained whole slide RGB pictures and non-paraffin-embedded stained Whole Slide RGB Images (WSRI). The Hematoxylin And Eosin Stain (H&E) is one of the most important stains in histology, however, it has various flaws relating to tissue preparation, staining techniques, slowness, and human error. Two unique techniques for training machine learning models for computational H&E staining and destaining of prostate core biopsy RGB pictures are presented. A conditional generative adversarial network is used in the staining model to learn hierarchical non-linear mappings between Whole Slide RGB Images (WSRI) pairs of prostate core biopsies before and after H&E staining. Using previously unknown non-stained biopsy photos as input, the trained staining model may then create computationally H&E-stained prostate core WSRIs. Natural dyes and pigments are harmless, eco-friendly alternatives to synthetic equivalents, and beetroot is one such natural color. The red hue of beetroot produced from this plant benefits significantly from betalain pigments. The physiochemical and spectrophotometric properties of beetroot, as well as the histology staining, were investigated in this study. The potential of various tissues was tested to ascertain tissue specificity. Histopathology is the visual study of the shape and morphology of a stained biopsy tissue slice pathologist using a microscope to diagnose various disease abnormalities. Following H&E, a histological examination was performed.