The SLICE-3D dataset - Skin Lesion Image Crops Extracted from 3DTBP for skin cancer detection - dataset description

Sep 21, 2026

Uploaded by Oliver

Algorithms for skin cancer detection can benefit triage, especially for populations with limited access to dermatologic specialists. The SLICE-3D (“Skin Lesion Image Crops Extracted from 3DTBP”) dataset provides over 400,000 distinct, de-identified skin lesion image crops collected from seven dermatologic centers worldwide. Images were systematically extracted from sensitive 3D Total Body Photographs and are comparable in optical resolution to smartphone photos, enabling AI models to train on broader-quality inputs. The dataset aims to reduce selection bias and support earlier detection in primary care and non-clinical settings.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Background & Summary

Motivations for improved skin cancer triage

Limitations of existing open datasets

What SLICE-3D provides (coverage and data origin)

Clinical workflow relevance and deployment scenarios

[www. nature.com/scientificdata](www. nature.com/scientificdata) oPEN DATA DESCRIPToR The SLICE-3D dataset: 400,000 skin lesion image crops extracted from 3DTBP for skin cancer detection Nicholas R. Kurtansky1 ✉, Brian M. D’alessandro2, Maura C. Gillis1, Brigid Betz-Stablein3, Sara E. Cerminara4, Rafael Garcia5, Marcela Alves Girundi6, Elisabeth Victoria Goessinger4, Philippe Gottfrois4, Pascale Guitera6,7, Allan C. Halpern1, Valerie Jakrot6, Harald Kittler8, Kivanc Kose1, Konstantinos Liopyris9, Josep Malvehy10,11, Victoria J. Mar12,13, Linda K. Martin6,7,14, Thomas Mathew6, Lara Valeska Maul15, Adam Mothershaw3, Alina M. Mueller4, Christoph Mueller8, Alexander A. Navarini4, Tarlia Rajeswaran16, Vin Rajeswaran16, Anup Saha5, Maithili Sashindranath12, Laura Serra-García17, H. Peter Soyer3, Georgios Theocharis9, Ayesha Vos13, Jochen Weber1 & Veronica Rotemberg1 AI image classification algorithms have shown promising results when applied to skin cancer detection. Most public skin cancer image datasets are comprised of dermoscopic photos and are limited by selection bias, lack of standardization, and lend themselves to development of algorithms that can only be used by skilled clinicians. The SLICE-3D (“Skin Lesion Image Crops Extracted from 3DTBP”) dataset described here addresses those concerns and contains images of over 400,000 distinct skin lesions from seven dermatologic centers from around the world. 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 nonexpert 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. Background & Summary Algorithms that can distinguish benign from malignant skin lesions with sufficient accuracy could improve triage for skin cancer detection and greatly benefit populations without access to specialized dermatologic care1–8. Current algorithms typically use images captured with a handheld medical device called a dermatoscope9, which uses a magnifying lens and a lighting source to illuminate morphologic features not otherwise visible to the naked eye 10, 11. Dermatoscopes are used commonly in dermatology clinics as an aid to evaluate skin lesions and diagnose skin cancers such as melanoma (MM)10, 12, 13, basal cell carcinoma (BCC)14, and squamous cell 1Dermatology Service, Department of Medicine, MemorialSloan Kettering Cancer Center, NewYork, NewYork, USA. 2Canfield Scientific, Inc., Parsippany, New Jersey, USA. 3Frazer Institute, The University of Queensland, Dermatology Research Centre, Brisbane, Queensland, Australia. 4Department of Dermatology, University Hospital of Basel, Basel, Switzerland. 5Computer Vision and Robotics Institute, University of Girona, Girona, Spain. 6 Melanoma institute Australia, Sydney, Australia. 7 Faculty of Medicine and Health, University of Sydney, Sydney, Australia. 8ViDiR Group, Department of Dermatology, Medical University of Vienna, Vienna, Austria. 9University of Athens Medical School, Athens, Greece. 10Dermatology Department, Hospital Clínic Barcelona, Universitat de Barcelona, IDIBAPS, Barcelona, Spain. 11centro de investigación Biomédica en Red de enfermedades Raras (ciBeR eR), instituto de Salud carlos iii, Barcelona, Spain. 12School of Public Health and Preventive Medicine, Monash University, Melbourne, VIC, Australia. 13Victorian Melanoma Service, Alfred Hospital, 55 Commercial Road, Melbourne, VIC, 3004, Australia. 14School of Clinical Medicine, Faculty of Medicine & Health, University of New South Wales, Sydney, Australia. 15Department of De

FAQs

What problem does the SLICE-3D dataset target in skin cancer detection AI?

It targets limitations of existing datasets, such as selection bias and lack of standardization, by providing a large, de-identified collection intended for more widely usable algorithm development and triage use-cases.

How many images and lesions does SLICE-3D include, and where do they come from?

SLICE-3D contains images of over 400,000 distinct skin lesions extracted from seven dermatologic centers worldwide.

Why are SLICE-3D images useful for real-world deployment beyond specialized clinics?

The cropped images are comparable in optical resolution to smartphone photos, supporting training and use in settings where images are captured by nonexpert physicians or patients for telehealth or primary care workflows.

Sep 21, 2026

Uploaded by Oliver