Nonlinear Deterministic Methods for Computer Aided Diagnosis in Case of Kidney Diseases

摘要

This paper proposes a set of nonlinear deterministic methods derived from chaos theory that can serve as computed aided diagnosis tools for kidney diseases based on computer topographies (CT). These procedures target the classification of the analyzed tissue samples in normal, malign and benign affected and also enhanced visualization of the CT images. The classification methods consist in estimating the fractal dimension of the kidney tissue and, respectively, the correlation dimension of the attractor obtained from the spatial series associated to the kidney image. The enhanced visualization method associates a fractal map to the analysed image. The methods are tested on 120 CTs presenting normal and modified tissue. The degree of trustworthiness of the methods while dealing with classifications is discussed based on statistical results and samples of fractal maps associated to the images are also presented.