LSA: Latent Style Augmentation Towards Stain-Agnostic Cervical Cancer Screening

2025-03-09

AI Summary

In cervical cancer whole-slide image (WSI) screening, inter-scanner staining variability induces severe domain shift; existing patch-level stain augmentation methods fail to scale to gigapixel WSIs—causing intra-WIS staining inconsistency and incurring prohibitive storage and computational overhead due to offline augmentation. This paper proposes the first online latent-space stain augmentation framework for WSIs, comprising WSI-level consistent stain augmentation (WSAug) and a latent-space Stain Transformer that synthesizes multi-style staining in real time at the feature level, eliminating the need to store augmented images. Trained exclusively on data from a single scanner, our method significantly improves classification accuracy across multiple unseen scanners. It is the first approach to achieve scalable, low-overhead, and clinically feasible stain robustness transfer—from patch-level to WSI-level—enabling practical deployment in heterogeneous digital pathology environments.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view Learning
• Search and Optimization: Distributed Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

The deployment of computer-aided diagnosis systems for cervical cancer screening using whole slide images (WSIs) faces critical challenges due to domain shifts caused by staining variations across different scanners and imaging environments. While existing stain augmentation methods improve patch-level robustness, they fail to scale to WSIs due to two key limitations: (1) inconsistent stain patterns when extending patch operations to gigapixel slides, and (2) prohibitive computational/storage costs from offline processing of augmented WSIs.To address this, we propose Latent Style Augmentation (LSA), a framework that performs efficient, online stain augmentation directly on WSI-level latent features. We first introduce WSAug, a WSI-level stain augmentation method ensuring consistent stain across patches within a WSI. Using offline-augmented WSIs by WSAug, we design and train Stain Transformer, which can simulate targeted style in the latent space, efficiently enhancing the robustness of the WSI-level classifier. We validate our method on a multi-scanner WSI dataset for cervical cancer diagnosis. Despite being trained on data from a single scanner, our approach achieves significant performance improvements on out-of-distribution data from other scanners. Code will be available at https://github.com/caijd2000/LSA.

Problem

• Addresses domain shifts in cervical cancer screening due to staining variations.
• Proposes Latent Style Augmentation for efficient, online WSI-level stain augmentation.
• Enhances robustness of WSI-level classifiers across different imaging environments.

Innovation

• Latent Style Augmentation for WSI-level stain consistency
• Stain Transformer simulates styles in latent space
• Efficient online augmentation reduces computational costs