Generative models for color normalization in digital pathology and dermatology: Advancing the learning paradigm

• Published in: Expert systems with applications 2024-07, Vol.245, p.123105, Article 123105
• Main Authors: Salvi, Massimo, Branciforti, Francesco, Molinari, Filippo, Meiburger, Kristen M.
• Format: Article
• Language: English
• Subjects: Color constancy; Deep learning; Digital dermatology; Digital pathology; Generative adversarial networks; Stain normalization
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2023.123105

[Display omitted] •A color normalization algorithm is proposed for digital pathology and dermatology.•A novel learning paradigm is presented to generalize heuristic algorithms.•The color normalization task is formulated as an image-to-image translation problem.•The generalization ability of the approach is tested on multiple external datasets. Color medical images introduce an additional confounding factor compared to conventional grayscale medical images: color variability. This variability can lead to inconsistent evaluation by clinicians and the misinterpretation or suboptimal learning process of automatic quantitative algorithms. To mitigate the potential negative consequences of color variability, several color normalization strategies have been developed, proving to be effective in standardizing image appearance. In this paper, we present a novel paradigm for color normalization using generative adversarial networks (GANs). Our method focuses on standardizing images in the field of digital pathology (stain normalization) and dermatology (color constancy), where high color variability is consistently observed. Specifically, we formulate the color normalization task as an image-to-image translation problem, ensuring a pixel-to-pixel correspondence between the original and normalized images. Our approach outperforms existing state-of-the-art methods in both the digital pathology and dermatology fields. Extensive validation using public datasets demonstrate the effectiveness of our color normalization results on entirely external test sets. Our framework exhibits strong generalization capability on unseen data, making it suitable for inclusion in the pipeline of automatic quantitative algorithms to reduce color variability and improve segmentation and/or classification performance. Lastly, we provide the source code of our models to encourage open science.