Statistical modeling and classification of reflectance confocal microscopy images

Abstract

This paper deals with the characterization and classification of reflectance confocal microscopy images of human skin. A special attention will be given to the identification and characterization of the lentigo, a phenomenon that originates at the dermo-epidermic junction of the skin. Confocal images are acquired at different skin depths with a high resolution. For each depth, the histograms of pixel intensities are determined, and well statistically modelled with a generalized gamma distribution (GGD). The scale, shape and translation parameters associated with the GGD are estimated using a new natural gradient descent algorithm showing fast convergence properties when compared to state-of-the-art estimation methods. Results show that the estimated parameters can be used to classify clinical images of lentigo and healthy patients. They also show that the scale and shape parameters are good features to identify and characterize the presence of lentigo in skin tissues.