{"access":"open","annotations":"Diagnosis (melanoma, nevus, BCC, ...), often histopathology-confirmed; some subsets have lesion segmentation and dermoscopic attribute masks; patient metadata (age, sex, site).","article":null,"cancer_slugs":["malignant-melanoma","non-melanoma-skin-cancer-nmsc"],"details":"","doi":null,"formats":["JPEG","CSV"],"hf":null,"huggingface":null,"kind":"registry","license":"CC-0 / CC BY-NC per contributor","modalities":["dermoscopy"],"models":[],"name":"ISIC Archive \u2014 International Skin Imaging Collaboration","page":"/ai-oncology/datasets/isic-archive","provider":"ISIC (Memorial Sloan Kettering and partners)","size":{"items":70000,"notes_en":"tens of thousands of images with diagnosis; annual challenge subsets (e.g. 2020: 33,126 images)","notes_pl":"dziesi\u0105tki tysi\u0119cy obraz\u00f3w z rozpoznaniem; coroczne podzbiory konkursowe (np. 2020: 33 126 obraz\u00f3w)","unit":"dermoscopic images"},"slug":"isic-archive","sources":[{"label":"ISIC Archive","url":"https://www.isic-archive.com/"}],"summary":"The public backbone of skin-cancer AI; strongly skewed towards light skin tones, which every model card built on it should say.","tasks":["classification","detection","segmentation"],"url":"https://www.isic-archive.com/","verified_at":"2026-09-05T22:25:58.972520"}