vix.ing · top · new · best · stats · spec

Pix2Pix-based Stain-to-Stain Translation: A Solution for Robust Stain\n Normalization in Histopathology Images Analysis

2020/02/03 by Pegah Salehi, Salehi, Pegah, Abdolah Chalechale +1 · 3 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · Engineering · #AI in cancer detection #Cell Image Analysis Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Image Processing Techniques and Applications #Image and Video Processing (eess.IV) #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2002.00647

openalex publication_date 2020/02/03 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

The diagnosis of cancer is mainly performed by visual analysis of the\npathologists, through examining the morphology of the tissue slices and the\nspatial arrangement of the cells. If the microscopic image of a specimen is not\nstained, it will look colorless and textured. Therefore, chemical staining is\nrequired to create contrast and help identify specific tissue components.\nDuring tissue preparation due to differences in chemicals, scanners, cutting\nthicknesses, and laboratory protocols, similar tissues are usually varied\nsignificantly in appearance. This diversity in staining, in addition to\nInterpretive disparity among pathologists more is one of the main challenges in\ndesigning robust and flexible systems for automated analysis. To address the\nstaining color variations, several methods for normalizing stain have been\nproposed. In our proposed method, a Stain-to-Stain Translation (STST) approach\nis used to stain normalization for Hematoxylin and Eosin (H&E) stained\nhistopathology images, which learns not only the specific color distribution\nbut also the preserves corresponding histopathological pattern. We perform the\nprocess of translation based on the pix2pix framework, which uses the\nconditional generator adversarial networks (cGANs). Our approach showed\nexcellent results, both mathematically and experimentally against the state of\nthe art methods. We have made the source code publicly available.\n

Cited by

Related