Skip to Main Content

LV Strain Quantification from 4D Echocardiography

Accurate, comprehensive quantification of regional left ventricular (LV) deformation is crucial for detection, risk stratification, and management of patients with ischemic heart disease. In this BRP, supported by NIH, four partners from two academic institutions and industry will work together to develop and validate an integrated imaging/image analysis system that will accurately, robustly, and reproducibly quantify regional LV strain and strain rate from four-dimensional (3 spatial dimensions and time) echocardiographic (4DE) image sequences. Intramural displacement will be estimated using a phase-sensitive-correlation-based speckle tracking approach being developed by a team led by Matthew O'Donnell at the University of Michigan. This information will be derived from rediofrequency (RF) signal data acquired from an ultrasound array, giving access to beam-formed acoustic data, using an approach being developed by a team lead by Jeff Powers, PhD from Philips Medical Systems. Displacement information at the myocardial surface will be derived from B-mode images using a shape-tracking strategy being developed by a team lead by James Duncan, PhD at Yale University, who will also serve as the PI of the BRP. Intramural and surface displacement information will be combined using an integrated segmentation/deformation estimator based on a biomechanical model (also being developed at Yale), to provide comprehensive, 4D estimates of myocardial strains, strain rates and material parameters.

The approach will be validated/evaluated using phantoms and in vivo testing in collaboration with a team lead by Yale cardiologist Albert Sinusas, MD. In vivo evaluation will include experiments based on acute and chronic canine models of ischemic injury for the quantification of transmurality of injury, subsequent LV remodeling and response to ACE inhibito therapy. Clinical feasibility will be established in human studies. The DE-derived indices of LV deformation will be shown to be comparable to those derived from Magnetic Resonance (MR) tagging.