Skin Melanoma DELM Automatic Classification by Wigner Distribution Related Synthesis Techniques

The development of an equipment for Computer Assisted Digital Dermatology (CADD) as a basis for the creation of Active Support Systems (ASS) for early diagnosis of melanotic disease is presented. It is based on a Digital EpiLuminescence Microscopy (DELM) tecnique, taking advantage of polarized light guided by optical fibers. To discriminate between malignant and benign melanocytic lesions, several dermatoscopical features have been proposed by different research groups, over the years. Nevertheless many are the attempts to reach reliable and objective classification procedures. We use, as a reference, the work by Stanganelli and Kenet. A Biomedical Engineering analyisis allows us to develop structured reference parameter tables, which enable to extract maximal information content from experimental dermatological data. The automatic classification parameters taken into account can be divided into two mainclasses: Characteristic Size Descriptor (point size, local, and global) and Intrinsic Descriptor (morphological, geometrical, chromatical, others). They offer the best structured operative synthetic description, when compared to usual specialisict ones. Those Structured Parameter Descriptors are the most suitable tool to ther effective development of reliable Active Support Systems for diagnostic applications in Dermatology. Reliable quantitative information can be derived from those parameters by usual Wigner Distribution related synthesis techniques. In the present paper, we present an improved approach to the traditional 2D Geometrical Syntetical Descriptors, based on a modification of a traditional Algebraic Moment Invariant Set. The improved approach allows us to reach overthree orders of magnitude higher precision than traditional techniques.