Automated cardiac motion recovery via active trajectory field models

Modern cardiovascular research continues to push the boundaries of medical imaging, developing innovative technologies to aid in the understanding, diagnosis, and treatment of cardiovascular dysfunction. For example, quantitative tissue tracking techniques such as displacement encoding with stimulated echoes (DENSE) in cardiac magnetic resonance (cMR) imaging and speckle tracking in echocardiography offer unprecedented insight into tissue motion and cardiac strain. However, the burden of complex manual data analyses still rests squarely on the shoulders of the cardiovascular scientist.;This dissertation develops a novel image analysis technique for automated cardiac motion recovery termed active trajectory field models (ATFMs). ATFMs take advantage of the intrinsic motion information of quantitative tissue tracking data to automatically recover accurate cardiac motion from noisy and incomplete multidimensional data acquisitions. The ATFM technique successfully fuses DENSE cMR and echocardiography, developing a model of cardiac motion from costly yet accurate DENSE cMR data and subsequently fitting this model to inexpensive yet incomplete echocardiographic acquisitions.;The first major contribution of this work is in the field of image analysis, developing a novel deformable modeling technique for automated motion recovery from a set of noisy and incomplete deformation vectors. ATFMs incorporate a characterization of tissue motion variability into the deformable modeling framework, constraining the segmentation solution to plausible motion fields without the need for finite-element approximations of complex biomechanical models. The second major contribution of this work is in the field of cardiovascular analysis, developing a novel automated algorithm for the recovery of meaningful cardiac motion from multidimensional DENSE cMR and echocardiographic imagery. ATFMs automatically recover accurate trajectory field solutions from noisy and incomplete quantitative tissue tracking data without the need for tedious and time consuming manual data analyses.;The ATFM technique is evaluated using a number of multidimensional DENSE cMR and echocardiographic datasets. On average, ATFM motion recovery produces radial and circumferential strain correlation values of 0.77 and 0.83 with respect to standard DENSE cMR and speckle tracking analysis techniques. These results indicate that cardiac trajectory data can be well characterized and motion can be successfully recovered from noisy and incomplete data acquisitions via the automated ATFM technique.