Characteristics of publicly available skin cancer image datasets: a systematic review

AI Generated Summary*

Searches of MEDLINE, Google, and Google Dataset Search yielded 29 skin cancer image datasets and 20 atlases. Of these, 21 datasets (106,950 images) and 17 atlases were open access, while eight datasets and three atlases had regulated access. Among the 14 datasets stating origin, 11 came exclusively from Europe, North America, or Oceania, and 19 of 21 open access datasets held only dermoscopic or macroscopic images. Metadata coverage was uneven: age 76.4%, sex 77.5%, body site 74.4% of images, but ethnicity 1.3% and Fitzpatrick skin type 2.1%. In the three datasets with skin type information, only ten images were type V and one type VI. Reporting was limited and variable, and the review calls for reporting standards.

*This summary was generated by AI and is published unedited. Oku does not alter these summaries. It may contain errors or omissions and is provided for general informational purposes only. Oku does not guarantee its accuracy, completeness, or reliability. For authoritative information, please refer to the original, peer-reviewed article.