Pigmented lesion classification poses a challenging task for dermatologists and pathologists alike given the variable clinical and histologic presentations. Melanoma has increased potential for morbidity and mortality compared to non-melanoma skin cancers, and it is now the third most commonly diagnosed cancer in the United States (Marghoob et al., 2009;Welch et al., 2021;Whiteman et al., 2016). While dermatologists require the fewest number of biopsies per diagnosis of cutaneous malignancy, both providers and patients must nevertheless accept that any skin biopsy exchanges diagnostic information for scar tissue (Nelson et al., 2019). Finding a diagnostic balance can be difficult, especially on the cosmetically sensitive regions of the face. Dermatologists have a 65-80% accuracy rate in melanoma diagnosis without the use of epiluminescence microscopy (dermoscopy). Artificial intelligence (AI) has potential to be a useful adjunct tool to assist dermatologists with this challenging diagnosis (Argenziano & Soyer, 2001). Thus, it is imperative that dermatologists understand the basics of AI.

No abstract