BT18 Is a mole-mapping device for patients with multiple and/or atypical naevi superior to standard photography in a National Health Service pigmented lesion clinic?

• Published in: British journal of dermatology (1951) 2024-06, Vol.191 (Supplement_1), p.i197-i197
• Main Authors: Hira, Sundip Kaur, Smith, Abbey, Hinchcliffe, Mark, Heir, Vanessa, Mitra, Angana, Smith, Hayley, Muinonen-Martin, Andrew
• Format: Article
• Language: English
• ISSN: 0007-0963; 1365-2133
• DOI: 10.1093/bjd/ljae090.415

Mole mapping is an effective way to facilitate the early detection of melanoma in high-risk patients with multiple and/or atypical naevi. Current UK guidance recommends baseline photography (preferably dermoscopic) and clinical review for monitoring [National Institute for Health and Care Excellence. Assessing melanoma. Available at: https://www.nice.org.uk/guidance/ng14/chapter/Recommendations#assessingmelanoma (last accessed 20 March 2024)]. Our aim was to evaluate the use of a mole-mapping device with artificial intelligence (AI) in our National Health Service specialist pigmented lesion clinic and to compare this to standard medical photography. We carried out a pilot in January 2023 comparing our current photography software, Medical Image Manager (MIM), to the Canfield IntelliStudio Gen3 with DermaGraphix® (Canfield Imaging Systems, Fairfield, NJ, USA). The IntelliStudio combines a semiautomated photoacquisition system with mole-tracking software, allowing linked comparison of two-dimensional macro and dermoscopic images. The software flags new or changed lesions, and AI known as DExI version 1.0 (Dermatology Explained Intelligence) risk stratifies lesions as low or high risk for malignancy. The technology was assessed on its efficiency of use, its ability to identify new or changed naevi, and the accuracy of DExI. Twelve patients with multiple and/or atypical naevi were recruited: seven had multiple visits and eight had images captured with both the IntelliStudio and standard photography by the clinical photography team. On average DermaGraphix® was 5.4 times faster (P = 0.03) and needed 54% fewer clicks per lesion (P = 0.02) compared with MIM to assess all marked naevi, macro and dermoscopically. Consultants identified lesions for dermoscopy ranging from 2 to 32 per patient (mean 11.9). The number of lesions per patient identified by DermaGraphix® ranged from 152 to 2721 (mean 700) and naevi counts ranged from 45 to 749 (mean 255). The software flagged a few new and changed lesions in most patients with follow-up images. The toggle function allowed these to be rapidly identified as either benign inflammatory lesions or false positives, mainly attributable to variable body positioning and artefacts such as clothing. DExI identified 13 lesions as high risk for malignancy, 5 of which scored as low risk on another visit. These lesions were reviewed by dermatologists; all were clinically stable and benign. In conclusion, the IntelliStudio was considered