Model of images of structureless areas for the analysis of pigment patterns using artificial intelligence in oncodermatology
摘要
This study addresses the creation of a new model for structureless areas of pigmented skin neoplasms intended for use in hardware and software systems for pattern recognition in oncodermatology. The problem which development of the model seeks to solve is the difficult formalization of the metaphorical medical definition of this element of pigment patterns. The proposed model of structureless areas is based on analysis of the texture characteristics of dermatoscopic images. The study used a set of 400 images containing a variety of dermatoscopic structures and structureless areas. An experimental study demonstrated the adequacy of the proposed model, which supports discrimination between structureless areas and other structures in dermatoscopic images with accuracy of up to 92.2%.