Appearance normalization of histology slides

Abstract

This paper presents a method for automatic color and intensity normalization of digitized histology slides stained with two different agents. In comparison to previous approaches, prior information on the stain vectors is used in the plane estimation process, resulting in improved stability of the estimates. Due to the prevalence of hematoxylin and eosin staining for histology slides, the proposed method has significant practical utility. In particular, it can be used as a first step to standardize appearance across slides and is effective at countering effects due to differing stain amounts and protocols and counteracting slide fading. The approach is validated against non-prior plane-fitting using synthetic experiments and 13 real datasets. Results of application of the method to adjustment of faded slides are given, and the effectiveness of the method in aiding statistical classification is shown.

Artificial Intelligence-Based Multiclass Classification of Benign or Malignant Mucosal Lesions of the Stomach

“…Staining inconsistencies may be attributed to multiple factors: dye, staining protocols of laboratories, fading, and digital scanners ( Vahadane et al, 2016 ; Roy et al, 2018 ). To ease the adverse impact upon the analytic accuracy of AI, multiple scholars have tried various methods to standardize color distribution in images ( Khan et al, 2014 ; Vicory et al, 2015 ; Bejnordi et al, 2016 ; Samsi et al, 2018 ). However, use of a single transformation function for each channel is rarely sufficient.…”

Spatial Statistics for Segmenting Histological Structures in H&E Stained Tissue Images

“…Previous normalization methods have utilized stain vector estimation methods [17] such as non negative matrix factorization [15]. We found these methods ineffective because the color distributions for some images in our dataset are skewed toward predominantly one stain, either hematoxylin or eosin.…”

A New Color Augmentation Method for Deep Learning Segmentation of Histological Images

“…Another advantage of Lab color space is that the selection of stained components can be achieved by a simple thresholding operation in the lightness channel. According to the Beer-Lambert law mentioned in [12], the transmission of light through a material can be modeled as…”

“…This is obtained by using an image augmentation technique that exploits the color transformation between different images, with a specific attention on the stained components within each sample. This allows the network to learn invariance to such variation, without the need to see these transformations in the labeled data [10], which is particularly important in the segmentation of histological images of human skin since the color variation is one of the most common variations [11,12]. It is shown below that such transformations can be efficiently implemented.…”