Detekce a modelování zájmových objektů z MR mozkových dat

Abstract

The aim of this bachelor thesis is the design of an automated algorithm for segmentation and modeling of pathological areas of MR brain imaging data. Two methods are used for segmentation purposes, namely Otsu multiregional segmentation and active contouring method implemented in MATLAB. Subsequently, these methods are compared to the purpose of evaluating the efficiency of segmentation of MR images. Furthermore, the effectiveness of using image pre-processing to improve segmentation results and subsequently extracting pathological areas is evaluated. This work describes general principles of segmentation of image data, SW and methods used to detect pathological regions of the brain. The proposed algorithms were tested on a dataset of 21 MR frames.