Analysis of skin lesions with pigmented networks

This paper deals with image processing for automatic analysis of pigmented skin lesions. Special emphasis is put on pigmented networks in epiluminescence microscopy (ELM) images. The purpose of processing images of pigmented skin lesions is the visualisation for education and training as well as for diagnosis support. We use a colour based segmentation applied to the Karhunen-Loeve transform of the RGB colour vectors to separate pigmented lesions from the skin and proceed with histogram equalisation and greyscale morphology to enhance and filter the pigmented networks. We demonstrate that we can enhance pigmented networks in skin lesions visually, and make them accessible to further analysis and classification.