Artificial intelligence and skin cancer
AI is already used in medical diagnostics. Could the diagnosis and therapy of pigmented skin lesions benefit from it in a realistic clinical scenario?
Two differental AI apps, in two different scenarios
The research team tested the AI application under realistic clinical conditions in two skin cancer centres, the MedUni University Department of Dermatology in Vienna and the Sydney Melanoma Diagnostic Centre in Australia. The prospective study consisted of two scenarios: AI was used in scenario A for suspected skin cancer changes and in scenario B for patients with many moles. The AI-assisted application was compared in both cases with both experienced and less experienced physicians.
In scenario A, 172 suspected pigmented lesions (of which 84 were malignant) were examined in 124 patients; in scenario B, the research team examined 5,696 pigmented lesions (of which 18 were malignant) in 66 patients. Two different artificial intelligence-based smartphone applications were used: a new 7-class AI algorithm and an ISIC algorithm already used in preliminary retrospective studies.
In scenario A, the 7-class AI algorithm showed equivalent diagnostic accuracy compared to experts, while it was significantly superior to less experienced physicians. The ISIC algorithm, on the other hand, performed significantly worse with respect to experts, but better with respect to inexperienced users.
Inferiority when compared to experts
In terms of therapeutic decisions, the 7-class algorithm was significantly inferior to experts but superior to inexperienced users. The results suggest that an AI-assisted smartphone application for skin cancer diagnosis makes equally good diagnostic decisions as experts in a real-life clinical scenario.
However, when it came to making treatment decisions, the experts were superior to the AI. Dr Kittler summarized the assessment of findings by indicating that:
'The AI application tends to remove more benign lesions in treatment recommendations than the experts. If this is taken into account, the AI application can certainly be used. One must also keep in mind that if it were used uncritically, too many false-positive results would have to be clarified'.
- Menzies SW, Sinz C, Menzies M, Lo SN, Yolland W, Lingohr J, Razmara M, Tschandl P, Guitera P, Scolyer RA, Boltz F, Borik-Heil L, Herbert Chan H, Chromy D, Coker DJ, Collgros H, Eghtedari M, Corral Forteza M, Forward E, Gallo B, Geisler S, Gibson M, Hampel A, Ho G, Junez L, Kienzl P, Martin A, Moloney FJ, Regio Pereira A, Ressler JM, Richter S, Silic K, Silly T, Skoll M, Tittes J, Weber P, Weninger W, Weiss D, Woo-Sampson P, Zilberg C, Kittler H. Comparison of humans versus mobile phone-powered artificial intelligence for the diagnosis and management of pigmented skin cancer in secondary care: a multicentre, prospective, diagnostic, clinical trial. Lancet Digit Health. 2023 Oct;5(10):e679-e691. doi: 10.1016/S2589-7500(23)00130-9. PMID: 37775188.