A Study of a Ventricular Motion in Cardiac MRI using Deformable Model
MetadataShow full item record
Publisher:The Ohio State University
Series/Report no.:The Ohio State University. Department of Electrical and Computer Engineering Honors Theses; 2007
We experimented with a novel deformable model that track the right ventricle’s (RV) wall motion through complete cardiac cycle by using a snake-like approach. The model uses a complex Fourier shape descriptor parameterization for efficient calculation of forces that constrains contour deformation. Even though the complexity exists in RV boundary shape, the model tracks the contour correctly and shows the robustness in weak contrast and noisy edge map. We also present a quantitative evaluation of delineation accuracy by comparing manual segmented contours and semi-automatically segmented contour, to check the reliability of our deformable model. The extracted shapes shows that the error between two contours to be an average of two pixels from 256 pixels by 256 pixels of cardiac magnetic resonance images. We used the spatio-temporal characterization of ventricular wall motion, obtained by our model, to help classifying the Intra-ventricular dyssynchrony (IVD) in the LV - i.e. asynchronous activation of LV wall - by adding RV information of ventricular movement to existing data. The classifying method was to use a popular statistical pattern recognition method of the Principal Component Analysis and the Fisher’s Linear Discriminant Analysis. From a database contains 33 patients, our classifier produced correct classification performance of 87.9 % with the RV data, which shows the promising improved IVD classification as contrast to current criteria for selecting therapy, which provided the correct classification of just 84.8 % on the same database with only the LV data.
Items in Knowledge Bank are protected by copyright, with all rights reserved, unless otherwise indicated.