CAncer MEtastases in LYmph nOdes challeNge (CAMELYON) Dataset
Radboud University Medical Center
Whole-slide pathology images of lymph nodes from the first computational pathology challenge.
All data from the CAMELYON challenges, the first to use whole-slide images in computational pathology. The goal was to help pathologists identify breast cancer metastases in sentinel lymph nodes, with slide-level and region-level annotations.
Access
- Cost
- FreeNo charge to download; hosted in an AWS S3 bucket.
- License
- Public domainCC0
- Sign-up
- No sign-up
- How you get it
- API and Bulk downloadTIFF, XML annotations
What's in it
- One row is
- One whole-slide image with annotations
- Covers
- Medical centers in the Netherlands
- Years
- Not stated
- Updated
- One-time release
- Size
- Not stated
Quality
A landmark benchmark with high-quality expert annotations. Slides are huge and need pathology image tooling.
- Gigapixel images
- Scanner and center differences
Connects to
- International Skin Imaging Collaboration (ISIC) Archive: Another medical imaging benchmark
- The Cancer Genome Atlas: Cancer data that includes pathology slides
Questions it could answer
- How well do models generalize across hospitals and scanners?
- What is the smallest metastasis detectors find reliably?