The influence of training on the recognition of gross features of dermoscopy images

Background: In a dermoscopic examination, besides structural components, inexperienced clinicians should also be able to recognize the gross features of the images. Aim: The aim of this study is, whether or not an inexperienced clinician has problems in the recognition of gross features of the images on dermoscopic examination. Methods: Two dermatologists, of whom one was experienced in the field of dermoscopy and the other was not, examined 161 dermoscopic images of melanocytic lesions in the gross features of their borders. Inner and outer borders were defined for each lesion. Both dermatologists separately evaluated the borders of the lesions for irregularity, asymmetry, and wideness of fading. For subjective image analysis they scored each lesion by using the four-point ordinal scale. For computerized image analysis they manually marked borders with dots, by using a computer program. We used quadratic-weighted kappa for interobserver reliability assessments for subjective scores and intraclass correlation coefficients (ICC) for automatically calculated scores. Results: In a subjective evaluation the inexperienced observer used a higher score than the experienced observer and the kappa values were between 0.241 - 0.286. ICC for the automatically calculated scores were between 0.357 and 0.522. According to both the outer and the inner borders, the concordance between experienced and inexperienced observers was almost perfect in measurements of diameter, perimeter, and area (ICC scores were between 0.948 and 0.990). Conclusions: An inexperienced person, in comparison with an experienced person, sees lesions in the same sizes, but in different shapes on dermoscopy. Therefore, it is advisable that making learners familiar with the borders of lesions should be included in the training on dermoscopy.