Contributors of this dataset are members of the International Skin Imaging Collaboration (ISIC), an international academia and industry partnership designed to reduce skin cancer morbidity and mortality through the development and use of digital skin imaging applications. ISIC engages both the dermatology and computer vision communities and works to achieve its goals by developing standards and guidelines to improve the quality, privacy, and interoperability for digital skin imaging, by making available a large and expanding open-source archive of quality labeled skin images, and by holding machine learning Grand Challenges for the computer science community in association with leading computer vision conferences. Since 2016, ISIC has hosted 5 Grand Challenges that have all been focused on developing AI for diagnostic classification using dermoscopy images. Studies have demonstrated that top-performing algorithms outperform the average dermatologist on single image diagnosis in preselected lesions and limited diagnostic classes. One study demonstrated the clinical utility of a winning algorithm prospectively in the hands of real clinicians and in front of real patients. However, these applications are confined to settings that involve the presence of a dermatologist at the time of photo capture.
Contrary to prior ISIC Challenges, the ISIC 2024 Challenge will not feature dermoscopy photos. To better mimic lower-quality non-dermoscopic images, we will use standardized cropped images of lesions from total body photography. The SLICE-3D (“Skin Lesion Image Crops Extracted from 3D TBP”) dataset contains images of over 400,000 distinct skin lesions images for training from seven dermatologic centers from around the world. The test set contains 500,000 additional images from a different set of patients. De-identified images were systematically extracted from sensitive 3D Total Body Photographs and are comparable in optical resolution to smartphone images. Algorithms trained on lower quality images could improve clinical workflows and detect skin cancers earlier if deployed in primary care or non-clinical settings, where photos are captured by non-expert physicians or patients. Such a tool could prompt individuals to visit a specialized dermatologist. This dataset circumvents many inherent limitations of prior datasets and may be used to build upon previous applications of skin imaging for cancer detection. Images were collected from Memorial Sloan Kettering Cancer Center (USA), Hospital Clinic de Barcelona (Spain), the University of Queensland (Australia), Medical University of Vienna (Austria), University of Athens (Greece), Melanoma Institute Australia (Australia), the University Hospital of Basel (Switzerland), Alfred Hospital (Australia), and FNQH Cairns (Australia).