Analýza a modelování zájmových struktur z echokardiografie
Loading...
Downloads
8
Date issued
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Vysoká škola báňská - Technická univerzita Ostrava
Location
Signature
Abstract
That diploma thesis is interested in implementation algorithm of Active Contours based on statistical parameters of pixels for segmentation of heart structures in Echocardiography. After implementation of created algorithm and separation of heart's structures is possible to meassure parametrs in ventricles and atriums of heart. The application output is grafical user interface contains all of these functionalities created in MATLAB as the all implemented codes. Chosen method was established on science publications from research in the first capture of thesis and evaluation of knowledge from these publications. Part of theory is the basis of Echokardiography, ultrasound images, ultrasound and last but not least was based on clinic problems and patology which you find out in this examination with determination of physiological parametrs of heart structures (heart ventricules, atriums). The practical part of thesis desribe steps of all segmentation process and that is preprocessing image, segmentation and meassuring parameters in separated model and that is lenght, width, center of gravity and volume of indicated by extracting from heart ventriculum or atrium. The final result is become the graphical user interface which include all of steps in interactive view for users. The next part of the thesis is dealt with testing in effectivity of implemented algorithm on images which was loaded by noise ( speckle, Gaussian and Salt and Pepper noise). All of data was segmentated by the same algorithm with the same set of parametrs like in native data and then was found out the similarity between segmentated models from noisy data and segmentated models from native data by structural similarity, mean squared error, correlation and by the DICE coefficient. The end of this thesis was evaluated regarding to results of these tested procedures and thereby of implemented algorith based on Active Contours.
Description
Subject(s)
Echocardiography, active contours, segmentation of images, image processing, filtration, ultrasound, cardiology, Salt and Pepper noise, Gaussian noise, speckle noise, correlation, sructural similarity, mean squared error, DICE coefficient, GUI