Método dialético de busca e otimização para análise de imagens de ressonância magnética.

Abstract

Multispectral image analysis is a relatively promising field of research with applications in several areas, such as remote sensing and medical imaging. Biology, Psychology and Social Sciences are intrinsically connected to the very roots of the development of algorithms and methods in Computational Intelligence, as it is easily seen in approaches like genetic algorithms, evolutionary programming and particle swarm optimization. However, Philosophy appears to be still considered a sort of enigmatic knowledge, despite the power of generalization and the systematic nature of investigative methods like dialectics. Here we claim that Philosophy can be also considered as a source of inspiration. In this work we propose a new non-supervised classifier based on dialectics as defined by Hegel’s works and the philosophical school of the Philosophy of Praxis, to classify synthetic multispectral magnetic resonance images. This work proved that such a classifier can reach results as good as those obtained by Kohonen’s self-organized networks. Herein this work we also generated new optimization methods based on dialectics: a canonical version and a version based on the Principle of Maximum Entropy. These methods were evaluated using benchmark functions and applied to optimize k-means clusters according to cluster validity indexes, getting good results when compared with canonical k-means. Dialectical classification was also performed in two case studies, involving the study of the progress of Alzheimer’s disease and detection of activated areas in functional imaging, demonstrating that the dialectical classifier can get good classification results when the initial number of classes is unknown.