Vývoj nástrojů pro segmentaci mozkových komor a tumoru mozku na snímcích za účelem následného vyhodnocení zájmových objektů

Abstract

This diploma thesis defines various terms that relate to pathophysiological description of the brain, including tumors, followed by a classification based on their severity, and their detection by medical imaging systems. It contains the description of development tools for image processing. In addition, the work in this scientific field is focused on the most impressive global segmentation researches and algorithms of the present time.

In the practical part of the thesis we describe the basic features and functions of the used development environment 〖FOTOM〗^NG. The next chapters are followed by the design solution and the actual implementation of image processing tools. The goal is to create a tool (or a module) to process the image from a magnetic resonance image for the subsequent accurate localization and identification of the pathological object, in my case the brain tumor. In my thesis, two segmentation modules that were programmed using the Java programming language (version 1.8) on the NetBeans IDE (version 8.2) and then implemented into the 〖FOTOM〗^NG system were created. Modules allow initial editing using binary thresholding, which helps automated segmentation itself. Here, the method is based on the detection of external border points and the method of geometric active contours Fast Marching Level-Set. The last part of my diplomathesis deals primarily with visualization and comparison of measured results