ABSTRACT
BACKGROUND: Dermoscopy revealed to be extremely useful in the diagnosis of early melanoma, the most important limitation being its subjectivity in giving a final diagnosis. To overcome this problem, several algorithms and checklists have been proposed. However, they generally demonstrated modest level of diagnostic accuracy, unsatisfactory concordance between dermoscopists and/or poor specificity. OBJECTIVE: To test a new methodological approach for the differentiation between early melanoma and atypical nevi, based on an integrated clinical-anamnestic dermoscopic risk scoring system (iDScore). METHODS: We selected a total of 435 standardized dermoscopic images of clinically atypical melanocytic skin lesion (MSL) excised in the suspect of malignancy (i.e. 134 early melanomas - MM - and 301 atypical nevi). Data concerning patient age and sex and lesion dimension and site were collected. A scoring classifier was designed based on this data set integrated with the dermoscopic evaluations performed by three experts blinded to histological diagnosis. RESULTS: A total of seven dermoscopic structures, three age groups (30-40 years, 41-60 years and >60 years), two maximum diameter categories (5-10 mm and >10 mm) and three body areas (i.e. frequently, chronically and seldom photoexposed sites) were selected by the scoring classifier as interdependently significant variables. The total risk score (S) of a lesion resulted from the simple sum of partial scores assigned to each selected variable. The iDScore-aided diagnosis showed an high accuracy (receiver operating characteristic-area under the curve = 0.903; IC: 95% = 0.887-0.918). A risk-based criticality scale corresponding to different S ranges was proposed. CONCLUSION: The iDScore checklist is proposed as a feasible and efficient tool to support dermatologists in non-invasive differentiation between atypical nevi and early MM on the basis of few selected clinical-anamnestic data and standardized dermoscopic features.
