Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Nov 21;8(1):708.
doi: 10.1038/s41746-025-02070-7.

Automated triage of cancer-suspicious skin lesions with 3D total-body photography

Nicholas R Kurtansky  1 , Maura C Gillis  2 , Noel C F Codella  3 , Brian M D'Alessandro  4 , Zongyuan Ge  5   6 , Pascale Guitera  7   8   9 , Allan C Halpern  2 , Harald Kittler  10 , Josep Malvehy  11   12 , Konstantinos Liopyris  13 , Victoria J Mar  14   15 , Linda K Martin  7   8   16 , Lara Valeska Maul  17   18 , Alexander Navarini  19 , Tarlia Rajeswaran  20 , Vin Rajeswaran  20 , Nadia Reichman  2 , H Peter Soyer  21 , Jochen Weber  2 , Siyuan Yan  6   22 , Veronica Rotemberg #  2 , Kivanc Kose #  2
Affiliations

Automated triage of cancer-suspicious skin lesions with 3D total-body photography

Nicholas R Kurtansky et al. NPJ Digit Med. .

Abstract

Careful selection of skin lesions that require expert evaluation is important for early skin cancer detection. Yet challenges include lack of cost-effective asymptomatic screening, geographical inequality in access to specialty dermatology, and long wait times due to exam inefficiencies and staff shortages. Machine learning models trained on high-quality dermoscopy photos have been shown to aid clinicians in diagnosing individual, hand-selected skin lesions. In contrast, models designed for triage have been less explored due to limited datasets that represent a broader net of skin lesions. 3D total body photography is an emerging technology used in dermatology to document all apparent skin lesions on a patient for skin cancer monitoring. A multi-institutional and global project collected over 900,000 lesion crops off 3D total body photos for an online grand challenge in machine learning. Here we summarize the results of the competition, 'ISIC 2024 - Skin Cancer Detection with 3D-TBP', demonstrate superiority of a model that utilized intra-patient context against a prior published approach, and explore clinical plausibility of automated atypical skin lesion triage through an ablation study.

PubMed Disclaimer

Conflict of interest statement

Competing interests: B. D’Alessandro is an employee of Canfield Scientific, Inc. N. Codella is a Microsoft employee and holds diverse investments in the technology and healthcare sectors. P. Guitera has participated in Advisory Boards for MSD and L’Oreal and received honoraria from Metaoptima PTY and travel support from L’Oréal. Neither of these is relevant for this paper. A. Halpern receives consultation fees from Canfield Scientific, Inc. A. Navarini and L.V. Maul received a grant from Canfield Scientific, Inc. for physician's salary in a separate study that had no influence on this manuscript. H.P. Soyer is a shareholder of MoleMap NZ Limited and e-derm consult GmbH and undertakes regular teledermatological reporting for both companies. H.P. Soyer is a medical consultant for Canfield Scientific Inc. and a medical advisor for First Derm. V. Rotemberg is a consultant for Inhabit Brands, Inc. and Atria Institute, and receives in-kind support from Kaggle and AWS. The other authors do not declare any competing interests.

Figures

Fig. 1
Fig. 1. Public and private leaderboard frequency and bivariate distributions.
a Shows a scatterplot of submissions according to their scores on the public (x-axis) and private (y-axis) leaderboards. While participant scores on the public and private leaderboard were highly correlated (Pearson′s correlation coefficient = 0.988), the deviation from the line of perfect concordance suggests some overfitting occurred. b Presents a histogram of scores on the private leaderboard. c Presents a histogram of scores on the public leaderboard.
Fig. 2
Fig. 2. Lesion risk scores stratified by patient.
Lesion risk score predictions by the winning model for each of the 91 patients whose tiles contained a melanoma. The risk score (y-axis) is displayed for melanomas (mean = 0.958, median = 0.988) as red dots, and all lesions (mean = 0.465, median = 0.437) as grey boxplots, stratified by each patient (x-axis). The patients are sorted by highest scoring melanoma. Lesion scores closer to 1 on the y-axis indicate higher model-estimated risk and lesions closer to 0 on the y-axis indicate lower model-estimated risk.
Fig. 3
Fig. 3. Association of lesion characteristics and ML-modelled risk.
Waterfall graph presents correlations between each continuous metadata feature and the mean lesion risk score rank (ascending order) over the top 500 placing submissions of ISIC’24.
Fig. 4
Fig. 4. ML model diagram.
Diagram of the winning model from the ISIC’24 competition, which was the subject to the ablation study.
See this image and copyright information in PMC

References

    1. Marchetti, M. A. et al. Prospective validation of dermoscopy-based open-source artificial intelligence for melanoma diagnosis (PROVE-AI study). NPJ Digit. Med.6, 127 (2023). - PMC - PubMed
    1. Kurtansky, N. R. et al. Effect of patient-contextual skin images in human- and artificial intelligence-based diagnosis of melanoma: Results from the 2020 SIIM-ISIC melanoma classification challenge. J. Eur. Acad. Dermatol. Venereol. 8, 1489–1499 (2024). - PMC - PubMed
    1. Marchetti, M. A. et al. Results of the 2016 International Skin Imaging Collaboration International Symposium on Biomedical Imaging challenge: comparison of the accuracy of computer algorithms to dermatologists for the diagnosis of melanoma from dermoscopic images. J. Am. Acad. Dermatol78, 270–277.e1 (2018). - PMC - PubMed
    1. Codella, N. et al. Skin lesion analysis toward melanoma detection 2018: a challenge hosted by the International Skin Imaging Collaboration (ISIC). arXiv. 10.48550/arXiv.1902.03368 (2019).
    1. Codella, N. C. F. et al. Skin lesion analysis toward melanoma detection: a challenge at the 2017 International Symposium on Biomedical Imaging (ISBI), hosted by the International Skin Imaging Collaboration (ISIC). In: Proc. IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018) 168–172 (IEEE, 2018). https://ieeexplore.ieee.org/document/8363547/

LinkOut - more resources