Stress echocardiography is a clinically established, cost-effective technique for detecting and characterizing coronary artery disease by imaging the left ventricle (LV) of the heart at rest and then after either exercise or pharmacologically-induce stress to reveal ischemia. However, acquisitions are heavily operator dependent, two-dimensional (2D), and interpretation is generally based on qualitative assessment. While a variety of quantitative 2D approaches have been proposed in the research literature, none have been shown to be superior to the still highly variable qualitative visual comparison of rest/stress echocardiographic image sequences for detecting ischemic disease. Here, we propose that the way forward must focus on a new computational image analysis paradigm for quantitative 4D (three spatial dimensions plus time) stress echocardiography. Our strategy integrates information derived from both radiofrequency (RF) and B-mode echocardiographic images acquired using a matrix array probe. The integrated analysis system will yield accurate and robust measures of strain and strain rate - at rest, stress and differentially between rest and stress - that will identify myocardial tissue at-risk after dobutamine-induced stress. This work wil involve the development of novel (1) phase-sensitive, correlation-based RF ultrasound speckle tracking to estimate mid-wall displacements, (2) ma- chine learning techniques to localize the LV bounding surfaces and their displacements from B-mode data, (3) a meshless integration approach based on radial basis functions (RBFs) and Bayesian reasoning/sparse coding to estimate dense spatiotemporal parameters of strain and strain rate and (4) non-rigid registration of rest and stress image sequences to develop unique, 3D differential deformation parameters. The quantitative approach will be validated with implanted sonomicrometers and microsphere-derived flows using an acute canine model of stenosis. The ability of deformation and differential deformation derived from 4D stress echocardiography to detect new myocardial tissue at-risk in the presence of existing infarction will then be determined in a hybrid acute/chronic canine model of infarction with superimposed ischemia. The technique will be translated to humans and evaluated by measuring the reproducibility of our deformation and differential deformation parameters in a small cohort of subjects. Three main collaborators will team on this work. A group led by Matthew O'Donnell from the University of Washington will develop the RF-based speckle tracking methods. An image analysis group led by the PI James Duncan at Yale University will develop methods for segmentation, shape tracking, dense displacement integration and strain computation. A cardiology/physiology group under Dr. Albert Sinusas at Yale will perform the acute and chronic canine studies and the human stress echo studies. A consultant from Philips Medical Systems will work with the entire team to bridge the ultrasound image acquisition technology.
The detection and diagnosis of coronary artery disease are commonly performed using stress echocardiography, a test that is performed 3 million times in the United States annually. Our new methods will enable the accurate and robust quantification of changes in myocardial deformation in 4D (three spatial dimensions over time) due to stress using a cost efficient and noninvasive imaging technology. The resulting methodology may lead to more sensitive and accurate detection of stress-induced ischemia that will better guide clinical decision making.
|Stendahl, John C; Sinusas, Albert J (2015) Nanoparticles for Cardiovascular Imaging and Therapeutic Delivery, Part 2: Radiolabeled Probes. J Nucl Med 56:1637-41|
|Compas, Colin B; Wong, Emily Y; Huang, Xiaojie et al. (2015) Correction to ""Radial Basis Functions for Combining Shape and Speckle Tracking in 4D Echocardiography"". IEEE Trans Med Imaging 34:690|
|Parajuli, Nripesh; Compas, Colin B; Lin, Ben A et al. (2015) Sparsity and Biomechanics Inspired Integration of Shape and Speckle Tracking for Cardiac Deformation Analysis. Funct Imaging Model Heart 9126:57-64|
|Stendahl, John C; Sinusas, Albert J (2015) Nanoparticles for Cardiovascular Imaging and Therapeutic Delivery, Part 1: Compositions and Features. J Nucl Med 56:1469-75|
|Huang, Xiaojie; Dione, Donald P; Compas, Colin B et al. (2014) Contour tracking in echocardiographic sequences via sparse representation and dictionary learning. Med Image Anal 18:253-71|
|Arnal, Bastien; Nguyen, Thu-Mai; O'Donnell, Matthew (2014) Toric focusing for radiation force applications using a toric lens coupled to a spherically focused transducer. IEEE Trans Ultrason Ferroelectr Freq Control 61:2032-41|
|Compas, Colin B; Wong, Emily Y; Huang, Xiaojie et al. (2014) Radial basis functions for combining shape and speckle tracking in 4D echocardiography. IEEE Trans Med Imaging 33:1275-89|
|Zhu, Yun; Papademetris, Xenophon; Sinusas, Albert J et al. (2009) A Dynamical Shape Prior for LV Segmentation from RT3D Echocardiography. Med Image Comput Comput Assist Interv 5761:206-213|