COMPUTER AIDED DIAGNOSIS METHODS BASED ON FRACTAL AND SPATIAL SERIES ANALYSIS FOR KIDNEY CT IMAGES

摘要

The goal of this paper is to present a set of algorithms developed based on fractal and spatial series analysis that can be applied to computer aided diagnosis for discrimination between normal and modified kidney tissue. These algorithms were tested on 120 computer tomography (CT) images of normal, benign and malign affected renal tissue. Two of the algorithms provide useful numerical results that can be gathered to form statistics and provide a classification of the kidney tissue in normal and malign affected, while the third method can be used for enhanced visualization that proves its usefulness in the case of images that can not be classified. The conclusions of the study on the selected set of CT images are that distinction between normal and malign tissues can be done with high accuracy and significantly better results are obtained from CT images taken with contrast substances while using the correlation dimension. The enhancing procedure can give insights on the problem at hand when the statistics fails.