Diagnostic Imaging Study for the Melanoma Advanced Imaging Dermatoscope (mAID)

The purpose of this study is to calculate the sensitivity and specificity of a novel imaging device and associated software algorithm in detecting early stage melanoma versus nevi of the skin. The instrument, which was invented by the PI, for the purposes of this study, will be loaned to three external (to Rockefeller) institutions and used on patients who are scheduled for biopsy of pigmented lesions. The purpose of correlating the output screening result of the novel device and the output diagnosis of the gold standard histology analysis procedure is so that these two diagnoses can be compared to generate the number of true positives, true negatives, false positives and false negatives for the novel device. The purpose of disseminating the device to the external institutions is to achieve the appropriate power such that the specificity can be evaluated at 99% sensitivity. The rationale for the power needed in the study is that in order to be clinically useful, the device needs to be extremely sensitive (i.e. 99%) because false negative diagnosis is a dangerous situation, leading to potential progression of melanoma, the most deadly form of skin cancer. The scientific hypothesis is that the diagnostic mechanism for superficial melanoma is the light tissue interaction that occurs between the blue shifted wavelengths (i.e. blue light, ultra violet light) and the superficial epidermis while the mechanism for diagnosis of deeper melanoma (i.e. Breslow depth \>0.5mm) is the light/tissue interaction that occurs between the red shifted light (i.e. red light, infrared light) and the portion of the pigmented lesion that lies within the dermis. These hypotheses were fueled by initial observations that the diagnostic sensitivity and specificity were wavelength dependent in a study that looked at only the red, green and blue wavelengths as available in traditional digital dermoscopy imaging. The initial finding was that of the multiple features analyzed, more features were statistically significant diagnostics in the blue channel but there were (a relative minority) other features that fared better in the red channel. It is hypothesized that the diagnostic features that did better in the red channel were features of deep melanin while the superficial regions, which should theoretically be atypical in ALL melanomas, were evident in the quantitative endpoint metrics generated from the blue channel.