Research ArticlesVirtual Double Staining: A Digital Approach to Immunohistochemical Quantification of Estrogen Receptor Protein in Breast Carcinoma SpecimensLykkegaard Andersen, Nina MD; Brügmann, Anja MD, PhD; Lelkaitis, Giedrius MD; Nielsen, Søren HT; Friis Lippert, Michael MSc; Vyberg, Mogens MDAuthor Information Institute of Pathology, Aalborg University Hospital, Aalborg, Denmark M.F.L. is an employee at Visiopharm. The remaining authors declare no conflict of interest. Reprints: Anja Brügmann, MD, PhD, Institute of Pathology, Aalborg University Hospital, Ladegaardsgade 3, Aalborg 9000, Denmark (e-mail: [email protected]). Applied Immunohistochemistry & Molecular Morphology: October 2018 - Volume 26 - Issue 9 - p 620-626 doi: 10.1097/PAI.0000000000000502 Buy Metrics Abstract Visual assessment of immunohistochemically detected estrogen receptor protein is prone to interobserver and intraobserver variation due to its subjective evaluation. The aim of this study was to validate a new image analysis system based on virtual double staining (VDS) by comparing visual and automated scorings of ER in tissue microarrays of breast carcinomas. Tissue microarrays were constructed of 112 consecutive resection specimens of breast carcinomas. Immunohistochemistry assays for ER and pancytokeratin was applied on separate serial sections. ER scoring was visually performed by 5 observers using the histoscore (H-score) method. The Visiopharm ER image analysis protocol (APP) software application using VDS technique was applied separating stromal cells from carcinoma and other epithelial cells based on the pancytokeratin reaction. Using color deconvolution, polynomial filters, and nuclear segmentation the APP determined the percentage of positive cells and their intensity, and calculated the resulting H-score. On the basis of 1% cutoff VDS was perfectly correlated with visual assessment (κ=1). Using H-score, a very high agreement between VDS and visual ER assessment was seen (R2=0.950). Image analysis has the attributes to eliminate the shortcomings of visual ER evaluation by generating automated, reproducible, and objective results of ER assessment. Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.