The International Skin Imaging Collaboration (ISIC) is a global initiative aimed at reducing deaths from melanoma and minimizing unnecessary biopsies by improving the early detection of melanoma through digital imaging and artificial intelligence (AI). To achieve this, ISIC works on developing digital imaging standards and fosters collaboration between dermatology and computer science communities. Although melanoma is its initial focus, ISIC’s work has broad implications for enhancing AI-driven diagnostic tools across the full spectrum of skin conditions, including non-melanoma skin cancers and inflammatory skin diseases.
How we collaborated
The Data Nutrition Project partnered with ISIC through initiative member Memorial Sloan Kettering Cancer Center to co-develop custom dataset documentation—referred to as “nutrition labels”—for ISIC datasets. These labels provide critical transparency about how the datasets should be used and highlight potential limitations, particularly around the representation of diverse skin types. Given the clinical consequences of underrepresentation, especially for individuals with darker skin tones who are at higher risk of misdiagnosis, this collaboration aimed to ensure datasets used to train AI are responsibly documented and more equitably representative.

Through this joint effort, we aim to support ISIC’s broader mission of developing and promoting standards for quality, privacy, and interoperability in dermatologic imaging. Clear, standardized documentation not only helps ensure the clinical utility of images by reinforcing quality thresholds (such as resolution and focus) but also addresses privacy concerns unique to dermatology, such as protecting privacy in body images. Ultimately, we hope this work contributes to the establishment of robust imaging standards that improve care across diagnostic, educational, and telehealth applications while advancing equity and trust in AI-powered dermatology.
