Analytical and dimensional morphometry in early diagnosis cutaneous melanoma with dermoscopic images.

Abstract

Objective: To give to the clinician an objective numerical assessment tool to evaluate melanomas so that a diagnosis can be reached with the assistance of computerized procedures.

Study design: The approach adopted for differential diagnosis of melanomas and nevi considers diverse morphologic characteristics intrinsic to the lesions, that is, shape, size, and symmetry in total independence of pigmentation, and proposes that this information can be evaluated quantitatively and separately by morphometric procedures with statistically valid independent numeric variables that guarantee objectivity and, from a method point of view, consistency.

Results: The results show that the differential diagnosis on malignant and benign lesions is made on five variables, which all describe the fine irregularities of the contour and have a high significance in comparing melanomas to nevi.

Conclusion: The multivariate discriminant analysis demonstrates the ability of the analytic variables to discriminate 88% of the lesions, rising to 90% if two-dimensional variables are included.