Moderní metody segmentace obrazu s využitím prvků umělé inteligence

Abstract

This thesis deals with the issue of medical images processing along with the application of artificial intelligence elements on image data. Today, artificial intelligence is a very effective and necessary tool not only in terms of optimizing image segmentation. Genetic algorithms (GA) have also proven themselves successful in solving various optimization problems in many areas of human activity. Image segmentation is used to extract objects of interest, to differentiate individual tissues in an image and their subsequent classification. The main problem this process is facing is incorrect distribution of individual parts of the image, which could lead to misdiagnosis and potential complications in the planning or realization of surgeries. The author mainly focuses on the usage and comparison of optimization evolutionary and genetic algorithms in terms of regional segmentation of biomedical images that can solve the shortcomings of conventional methods. In particular these include ABC, PSO, DPSO and GA algorithms. The thesis among others also includes the comparison of individual methods and testing simulation environment for evaluation of individual approaches of image segmentation, where their functionality is compared and verified on real medical images.