Metody shlukové analýzy pro biomedicínské signály

Abstract

The bachelor thesis deals with regional image segmentation based on cluster analysis. The aim of this work is realisation and design of selected non-hierarchical methods k-means and fuzzy c-means. Testing was performed on real medical data which is affected by artificial additive noise. Part of this work is a chapter dealing with tissue modeling on selected medical images for the purpose of their subsequent extraction and chapter for implementation of cluster analysis on 1D ECG signals with the aim of decomposing characteristic segments of these signals. Main reason for testing was performed on 3 medical image datasets and the result was evaluated based on the correlation coefficient and the mean square error. The last step was to evaluate the achieved results transferred mainly to the graphic form. This was followed by the creation of a graphical-user interface to simplify testing. The result of this work is a dynamic evaluation of non-hierarchical clustering techniques for creating mathematical models of tissues depending on the intensity of noise.