A Multicentric Dataset for Training and Benchmarking Breast Cancer Segmentation in H&E Slides

2025-10-02

AI Summary

Existing public breast cancer histopathology datasets exhibit insufficient morphological diversity, limiting model generalizability across heterogeneous patient populations and hindering biomarker validation. To address this, we introduce BEETLE—the first multi-center, H&E-stained whole-slide image segmentation dataset specifically designed to capture rare morphologies, including diffuse lobular carcinoma infiltration and ductal carcinoma in situ. BEETLE comprises 587 annotated cases spanning diverse molecular subtypes and histological grades, enabling semantic segmentation of invasive/non-invasive epithelium, necrosis, and other clinically relevant structures. Images were acquired using seven distinct digital slide scanners; annotations underwent multiple rounds of expert review and are accompanied by standardized clinical metadata. An independent external test set is provided for robust evaluation. Experiments demonstrate that BEETLE substantially improves cross-institutional model generalization. As an open-source resource, BEETLE enables reproducible, benchmarkable, and quantitatively rigorous automated analysis of breast cancer histopathology.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view Learning
• Knowledge Representation and Reasoning: Ontologies

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Security and Privacy: Data transparency and provenance

Abstract

Automated semantic segmentation of whole-slide images (WSIs) stained with hematoxylin and eosin (H&E) is essential for large-scale artificial intelligence-based biomarker analysis in breast cancer. However, existing public datasets for breast cancer segmentation lack the morphological diversity needed to support model generalizability and robust biomarker validation across heterogeneous patient cohorts. We introduce BrEast cancEr hisTopathoLogy sEgmentation (BEETLE), a dataset for multiclass semantic segmentation of H&E-stained breast cancer WSIs. It consists of 587 biopsies and resections from three collaborating clinical centers and two public datasets, digitized using seven scanners, and covers all molecular subtypes and histological grades. Using diverse annotation strategies, we collected annotations across four classes - invasive epithelium, non-invasive epithelium, necrosis, and other - with particular focus on morphologies underrepresented in existing datasets, such as ductal carcinoma in situ and dispersed lobular tumor cells. The dataset's diversity and relevance to the rapidly growing field of automated biomarker quantification in breast cancer ensure its high potential for reuse. Finally, we provide a well-curated, multicentric external evaluation set to enable standardized benchmarking of breast cancer segmentation models.

Problem

• Addressing limited morphological diversity in breast cancer segmentation datasets
• Providing multiclass annotations for underrepresented breast cancer morphologies
• Establishing standardized benchmarks for breast cancer segmentation model evaluation

Innovation

• Multicentric dataset with diverse breast cancer morphologies
• Combines biopsies from multiple centers and public datasets
• Provides standardized evaluation set for segmentation benchmarking