Diagnosis of pigmented skin lesions by dermoscopy: web-based training improves diagnostic performance of non-experts

Background: Dermoscopy has been shown to enhance the diagnosis of melanoma. However, use of dermoscopy requires training and expertise to be effective. Objectives: To determine whether an Internet-based course is a suitable tool in teaching dermoscopy, and to evaluate the diagnostic value of pattern analysis and diagnostic algorithms in colleagues not yet familiar with this technique. Methods: Sixteen colleagues who were not experts in dermoscopy were asked to evaluate the dermoscopic images of 20 pigmented skin lesions using different diagnostic methods (i.e. pattern analysis, ABCD rule, seven-point checklist and Menzies' method), before and after an Internet-based training course on dermoscopy. Mean ± SEM sensitivity, specificity and diagnostic accuracy, and kappa (κ) intraobserver agreement were evaluated for each diagnostic method before and after training for the 16 participants. Differences between mean values were assessed by means of two-tailed Wilcoxon rank-sum tests. Results: There was a considerable improvement in the dermoscopic melanoma diagnosis after the Web-based training vs. before. Improvements in sensitivity and diagnostic accuracy were significant for the ABCD rule and Menzies' method. Improvements in sensitivity were also significant for pattern analysis, whereas the sensitivity values were high for the seven-point checklist in evaluations both before and after training. No significant difference was found for specificity before and after training for any method. There was a significant improvement in the κ intraobserver agreement after training for pattern analysis and the ABCD rule. For the seven-point checklist and Menzies' method there was already good agreement before training, with no significant improvement after training. Conclusions: We demonstrated that Web-based training is an effective tool for teaching dermoscopy