StainGAN: Stain Style Transfer for Digital Histological Images

2018/04/04 by M Tarek Shaban, Shaban, M Tarek, Christoph Baur +5 · 10 citations
Computer Science · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #cs.CV

paper · pdf · doi:10.48550/arxiv.1804.01601

Submitted to MICCAI 2018

arxiv created 2018/04/04 · arxiv updated 2018/04/06

Abstract

Digitized Histological diagnosis is in increasing demand. However, color variations due to various factors are imposing obstacles to the diagnosis process. The problem of stain color variations is a well-defined problem with many proposed solutions. Most of these solutions are highly dependent on a reference template slide. We propose a deep-learning solution inspired by CycleGANs that is trained end-to-end, eliminating the need for an expert to pick a representative reference slide. Our approach showed superior results quantitatively and qualitatively against the state of the art methods (10% improvement visually using SSIM). We further validated our method on a clinical use-case, namely Breast Cancer tumor classification, showing 12% increase in AUC. The code will be made publicly available.

Cited by

• When normalization hallucinates: unseen risks in AI-powered whole slide image processing
• K-Stain: Keypoint-Driven Correspondence for H&E-to-IHC Virtual Staining
• Adaptive Stain Normalization for Cross-Domain Medical Histology
• Transparency of medical artificial intelligence systems
• TRUST: Token-dRiven Ultrasound Style Transfer for Cross-Device Adaptation
• MorphGen: Morphology-Guided Representation Learning for Robust Single-Domain Generalization in Histopathological Cancer Classification
• Pathology Foundation Models are Scanner Sensitive: Benchmark and Mitigation with Contrastive ScanGen Loss
• Staining normalization in histopathology: Method benchmarking using multicenter dataset
• StainPIDR: A Pathological Image Decouplingand Reconstruction Method for Stain Normalization Based on Color Vector Quantization and Structure Restaining
• SiliCoN: Simultaneous Nuclei Segmentation and Color Normalization of Histological Images

Related

• Structure-Preserving Multi-Domain Stain Color Augmentation using Style-Transfer with Disentangled Representations
• Stain Style Transfer of Histopathology Images Via Structure-Preserved Generative Learning
• Histopathological Stain Transfer using Style Transfer Network with Adversarial Loss
• Region-guided CycleGANs for Stain Transfer in Whole Slide Images
• Neural Stain-Style Transfer Learning using GAN for Histopathological Images