Melanocytic Lesions Screening through Particle Swarm Optimization

Abstract

Early detection of malignant melanoma, which is the most dangerous skin cancer, significantly improves the chances of curing it. For this reason, dermatologists are looking for new methods for the examination of suspicious lesions that changes their shape over time. The author investigates in this chapter some algorithms which may be used for automated diagnosis of skin lesions. First algorithm performs the image segmentation by edge detection, which plays an important role in identifying borders of the lesion. Next algorithm uses the Particle Swarm Optimization (PSO) paradigm for recognizing the images of the same melanocytic nevus taken at different moments of time. The idea is that a novel view of an object can be recognized by simply matching it to combinations of known views of the same object. The main difficulty in implementing this idea is determining the parameters of the combination of views. The space of parameters is very large and we propose a PSO approach to search this space efficiently. The effectiveness of this approach is shown on a set of real images captured with a camera under different angles of view.