Cardiac image analysis: morphology, function, and dynamics

This dissertation presents algorithms for: (1) the computation of left ventricular (LV) Ejection Fraction (EF) from cardiac cine-MR images with the bias and the limits of agreement comparable to the inter-observer variability inherent in manual methods, and (2) an anatomically relevant parametric shape-motion analysis of the left anterior descending (LAD) coronary artery. Cardiovascular disease (CVD) is the leading cause of death world-wide. Substantial research effort focused on the improvement of the cardiovascular imaging modalities, has achieved increase in spatio-temporal resolution to enable precise physiological analysis of the heart. However, the enormous amount of data generated by these methods renders the manual analysis as a formidable task. The development of computer-assisted image analysis methods aid in clinical diagnosis, prognosis and monitoring of CVD. This dissertation presents a computational framework, which combines the adaptive fuzzy connectedness-based region segmentation approach with energy-minimizing dynamic programming, and the elastically adaptive physics-based deformable model frameworks. Specifically, this framework has been applied for the automated delineation of the LV myocardial contours in each of the acquired cine-MR images, and preliminary shape-motion analysis of the LAD coronary artery.