Metody regionální segmentace pro identifikaci objektů zájmu z medicínských obrazů – laboratorní úloha

Abstract

Visual imaging information plays a fundamental role in almost every aspect of our lives. Nowadays, most of this information presents itself the digital form of an image. This kind of imagery is ubiquitous, from a television picture, digital photographs to a CT image. The thesis deals with pretreatment and the following segmentation of medical images. Medical images include footage from CT and MRI. The aim of the thesis is to evaluate the pretreatment image application using the median filter technique, successive by the usage of Otsu thresholding and K-means clustering methods. Attained results will be used for an efficient evaluation of each method used. The efficiency of the methods will be evaluated via evaluation parameters. MSE, PSNR, and a correlation coefficient will be used as the parameters for the evaluation. The entire thesis makes use of the MATLAB software. The final analysis processes the important information relating to the features of the method used. The thesis ends with the proposition of an educational laboratory task.