Background: An acceptable level of reliability is a prerequisite for the introduction of epiluminescence microscopy (ELM) into the diagnosis of pigmented skin lesions.
Objective: To assess the effects of a training program on the reproducibility and accuracy of ELM as compared to clinical evaluation.
Methods: Before and after the program, 48 clinical images and their ELM counterparts were independently evaluated by seven participants. Reproducibility was measured by the kappa index, accuracy by the rate of exact diagnoses (RED) assuming histology as a reference.
Results: We observed (i) no effect of training on clinical reproducibility, (II) an improved reproducibility of ELM diagnoses of non-melanocytic lesions (NML) and melanomas but not of melanocytic nevi (MN), (iii) a greater increase in the clinical RED of NML and melanomas compared with MN and (iv) a similar though more pronounced increase in the RED on ELM.
Conclusions: The effects of training were greater for ELM as compared to clinical diagnosis, and for NML and melanomas as compared to MN.