Features extracting from fluorescence images and features sequential scanning for classification purposes

The purpose from this paper is to introduce a sample of parameters that can be lead to reliably discrimination of the malignant from benign diseases. Diverse parameters extracted from fluorescence images by applying adaptively learnt or predefined filters. The K-nearest neighbor´s algorithm is used to classify the skin lesions; a technique of sequential scanning of the parameters that could be applied to find an optimal set of parameters that would improve classification accuracy. This classification approach is modular and enables easy inclusion and exclusion of parameters. This facilitates the evaluation of their significance related to the skin cancer diagnosis. We have implemented a parameter scanning scheme which allows automatic optimization of the K-nearest neighbor classifier and indicates which features are more relevant for the diagnosis problem.