cancer computational pathology computer vision deep learning histopathology life sciences
This page describes the COBRA (Classification Of Basal cell carcinoma, Risky skin cancers and Abnormalities) skin pathology dataset, which comprises over 7000 histopathology whole-slide-images related to the diagnosis of basal cell carcinoma skin cancer, the most commonly diagnosed cancer. The dataset includes biopsies and excisions and is divided into four groups. The first group contains about 2,500 BCC biopsies with subtype labels, while the second group includes 2,500 non-BCC biopsies with different types of skin dysplasia. The third group has 1,000 labelled risky cancer biopsies, including rare and dangerous types like melanoma, squamous cell carcinoma, and Merkel cell carcinoma. Finally, the fourth group contains 1,000 BCC excisions with or without free tumor margins, with 300 fully annotated. The dataset will be released over 2023 in stages, starting with the BCC and non-BCC biopsies, followed by BCC excisions, and finally the risky cancer biopsies.
As required
CC BY-SA-NC 4.0
https://daangeijs.github.io/cobra/
Radboud University Medical Center
See all datasets managed by Radboud University Medical Center.
https://www.computationalpathologygroup.eu/members/daan-geijs/
COBRA was accessed on DATE from https://registry.opendata.aws/cobra.
arn:aws:s3:::cobra-pathologyus-west-2aws s3 ls --no-sign-request s3://cobra-pathology/