COBRA: Over 7,000 Skin Cancer Histopathology Whole-Slide Images

Radboud University Medical Center created the COBRA dataset, comprising over 7,000 histopathology whole-slide-images for diagnosing basal cell carcinoma and other skin cancers. The collection includes about 2,500 BCC biopsies with subtype labels, 2,500 non-BCC biopsies, 1,000 labelled risky cancer biopsies, and 1,000 BCC excisions. The dataset was scheduled for release in stages throughout 2023.

Use Cases

• Training classifiers for basal cell carcinoma (BCC) subtype identification based on labelled biopsy images.
• Developing models to differentiate BCC from other skin dysplasias based on the non-BCC biopsy group.
• Building detection systems for rare and dangerous skin cancers like melanoma based on the risky cancer biopsy set.
• Analyzing tumor margins in excision samples based on the group with free margin annotations.

Strengths

• Over 7,000 whole-slide-images provide a substantial base for model training.
• Includes 300 fully annotated BCC excision samples for detailed margin analysis.
• Covers a diverse range of skin pathologies, including common BCC, dysplasias, and rare cancers like melanoma.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

Radboud University Medical Center

Collection Method

Collection of biopsies and excisions for histopathology analysis.

Time Range

Release scheduled for 2023.

Freshness

Release was scheduled for 2023; current availability status is unknown.

Geography

null