3d Cine Dense Mri: Ventricular Segmentation And Myocardial Strain Analysis

Abstract

3D cine DENSE MRI: Ventricular segmentation and myocardial

strain analysis

Daniel Alejandro Auger 11 February 2013

Displacement encoding with stimulated echoes (DENSE) is a quantitative magnetic

resonance imaging (MRI) technique which measures myocardial displacement at a

pixel resolution. Regional myocardial strain can then be calculated, providing a useful

clinical measure of myocardial function. Right ventricular (RV) function is difficult to

quantify because of its asymmetrical geometry, thin myocardial wall and eccentric

motion. Novel motion and strain analysis techniques using 3D cine DENSE are

presented for the RV. Three dimensional tissue tracking methods offer the ability to

quantify RV strain at a previously unattainable spatial resolution. The RV is divided

into four anatomical segments in order to provide regional measures of strain across

the cardiac cycle. Results compare favorably to previous studies, showing variation in

strain according to structure. Segmenting the myocardium from 3D cine DENSE data

is necessary for regional myocardial strain quantification. This prohibitively time

consuming step is a limiting factor for the use of 3D cine DENSE in a clinical

environment. A semi-automated left ventricular segmentation algorithm is introduced

to minimize the data analysis time. The segmentation is initialized by interactively

placing guide points along the left ventricles borders at a single cardiac phase. A

mathematical model is fitted to the points and used to generate epicardial and

endocardial surfaces. Each surface is then propagated across the cardiac cycle using

the displacement information inherent in the phase data. Segmentation results

compare well to corresponding manually defined contours, and the time required to

segment an entire three dimensional data set is reduced by 10 fold, with only 5.6% of

contours requiring readjustment. Finally, methods are presented to quantify biventricular

cardiac dyssynchrony using strain measurements obtained from multiple

planes of 2D cine DENSE data. Results illustrate the variation of regional strain

across the cardiac cycle between healthy and diseased hearts. Dyssynchrony is

quantified using time to onset and time to peak strain metrics. Results illustrate biventricular strain variations reflecting right and left bundle branch blocks.