We much appreciate and mainly agree with many statements in the letters of Bystryn1 and Lambert et al.2 The 2 letters also give us the opportunity to clarify some important points. As indicated in our article, it was not our principal interest to develop or implement an apparatus to improve the diagnostic accuracy of pigmentary skin lesions. We would even rather no longer speak of automatic diagnosis, but rather of auxiliaries for clinical decisions. In fact, in the conclusions of our study we state that for the moment, digital dermoscopy analysis may be useful for training the eye to perform epiluminescence microscopy without a computer, perhaps to confirm a computer's epiluminescence microscopy diagnosis, and may also be a potentially formidable tool for research and teaching. Nevertheless, specific methods of pattern recognition more sophisticated than stepwise linear discriminant analysis (eg, nonlinear and/or nonparametric classification techniques, neural networks, etc) could improve diagnostic accuracy. From this point of view suitable methods (ROC [receiver operating characteristic] curves) that enable the cost of the clinical decision to be shifted, as Bystryn suggests, would be also recommended, thus increasing the sensitivity at the expense (as little as possible) of specificity. However, in our study we aimed simply to investigate the effective discriminant power of digital dermoscopy analysis, extracting only those variables able to give statistically significant data to help distinguish melanomas from nevi.
Rubegni P, Cevenini G, Burroni M, Perotti R, Dell'Eua G, Andreassi L. Digital Dermoscopy Analysis of Pigmented Skin Lesions: An Important Auxiliary for Clinical Decision and Not for Automatic Diagnosis. Arch Dermatol. 2001;137(3):378. doi:
Coronavirus Resource Center
Customize your JAMA Network experience by selecting one or more topics from the list below.
Create a personal account or sign in to: