Dermoscopy and digital dermoscopy analysis of palmoplantar 'equivocal' pigmented skin lesions in Caucasians

Abstract

Background/aim: The diagnosis of palmoplantar melanoma is often delayed and misdiagnosis is common, due to frequently unusual clinical presentation. We used a digital dermoscopy analyzer with a series of palmoplantar pigmented skin lesions (PP-PSL), and we compared sensitivity, specificity and diagnostic accuracy obtained with digital dermoscopy analysis (DDA) and classical dermoscopy.

Methods: Digital dermoscopy images of 107 PP-PSL were retrospectively obtained from the database of images of 3 Italian centers. The lesions (25 melanomas and 82 nevi) were all removed because of the presence of clinical and/or dermoscopic suspicious features. All digital images were analyzed using appropriate algorithms, and the diagnostic accuracy of the model was calculated. For comparison, dermoscopic images were clinically evaluated by two dermatologists and the Cohen ĸ concordance with DDA was calculated.

Results: The stepwise logistic regression analysis selected only 5 parameters out of 49. The logistic model achieved a sensitivity of 96% and a specificity of 87.8%. The Cohen ĸ concordance, evaluated by the Landis and Koch scale, supplied a substantial agreement between dermoscopy and DDA.

Conclusions: DDA might be a useful diagnostic instrument in the evaluation of preselected PP-PSL. However, these findings should be confirmed in a formal clinical trial.