Reflectance confocal microscopy (RCM) is a non-invasive and in-vivo imaging modality with potential for great impact in detection and classification of skin lesions such as melanocytic nevi. However, RCM skin images are visually very different from standard histological sections and challenging to interpret visually. Thus automated and semi-automated processing can dramatically ease clinical adoption of this technology. A challenging problem in such analysis is detection of the Dermis/Epidermis junction. One confounding, but potentially informative, skin feature is the presence of wrinkles. We have developed an automated two-step wrinkle detector for RCM images which has good sensitivity and even higher specificity. However all attempts at automated methods for this problem are likely to require, in practice, significant intervention from clinicians or highly trained readers. This manual labeling is tedious and time consuming. Thus in parallel with development of automated methods, we are developing an interactive segmentation approach that combines human guidance, sophisticated segmentation software with integrated visualization (using the software program Seg3D), and machine learning methods designed to maximally accelerate and improve segmentation with minimal user input by learning the user’s preferences and guiding the user towards the most useful regions in the image for manual intervention.