Meeting Banner
Abstract #4500

Computer Assisted Branch Cut Placement for Computing 3D+t Biventricular Strain from Tagged MRI

Ming Li1, 2, Bharath Ambale Venkatesh3, Himanshu Gupta4, Steve G. Lloyd4, Louis J. Dell'Italia4, Thomas Stewart Denney Jr1, 2

1AU MRI Research Center, Auburn University, Auburn, AL, United States; 2Electrical and Computer Engineering Department, Auburn University, Auburn, AL, United States; 3Johns Hopkins University, Baltimore, MD, United States; 4Department of Medicine, Division of Cardiovascular Disease, University of Alabama at Birmingham, Birmingham, AL, United States

Myocardial strain from tagged magnetic resonance imaging is an important parameter of cardiac mechanical function. In previous work, three-dimensional biventricular strain throughout cardiac cycle was reconstructed using displacement measurements obtained by unwrapping harmonic phase. Ideally, multiples of 2 can be added to regions to unwrap a phase image, but noise and artifacts can create inconsistencies in the phase and cause unwrapping methods to fail. Previous work used manually-placed branch cuts to resolve these inconsistencies. In this abstract, we present an algorithm to automatically place branch cuts in both long and short axis tagged cardiac images and compute 3D+time biventricular strain.

Keywords

accurately agreement allowed annealing apart applied around assisted auburn audience automated automatic automatically axis blue border branch cardiac cardiologists cardiovascular cause cavities closed coefficients composed compute computed computer computing connections consecutive consistent consuming core correction correctly create cuts cycle deformation dell dense detected detection deviation diabetic dimensional discrete disease displacement dual electrical energy engineers error excellent exhaustive fail fast frames fully function gated grid harmonic harp heart hypertension hypertensive ideally identified identifying implemented inconsistencies incorrect inspecting johns long loop loops manual manually maps maximal mechanical medicine misaligned model myocardial myocardium negative noise optimize paper patient patients penalized penalizing physiologists pixels placed placement population positive prospectively pulmonary rapid reconstructed remove represent representative required residue residues resistant resolution search segmentation short simulated slices spaced spatial squared strain strains studies subject successfully systolic tagged tags target term terms throughout typical underwent unwrap unwrapped unwrapping user utilized validate variation ventricle ventricular view volunteer volunteers wraps