P2A-9 Analysis of 4D Ultrasound for Dynamic Measures of Cardiac Function

Abstract

Real-time three-dimensional echocardiography (RT3DE) offers an efficient way to obtain complete 3D images of the heart over an entire cardiac cycle in just a few seconds of imaging. The complex 3D wall motion and temporal information contained in these 4D data sequences has the potential to enhance and supplement clinical diagnoses of the heart. However, today cardiac examinations mostly rely on 2D echocardiography, and 4D cardiac imaging dynamics is often not included. Our method is based on optical flow which is used to track myocardial motion. With a manual initialization of endocardium and epicardium at end-diastolic frames, the proposed method can automatically compute a myocardial motion field throughout an entire cardiac cycle from RT3DE, from which directional displacements, strains, and cardiac torsion can be automatically derived. In this paper, the method is validated using canine experimental data as well as clinical patient data. RT3DE data sets were taken before and after a proximal LAD occlusion under an IRB approved canine protocol. Strain analysis was performed on both data and averaged according to the ASE 16-segment left ventricular model. Radial displacement correctly indicated the outward motion of the anterior section with a reduction in global motion after the occlusion. The thickening analyses results yielded negative radial strain values for the mid-anterior section throughout the whole cardiac cycle. 3D visualization of ED-ES radial displacements and strains also indicated the dyskinetic regions in the mid-anterior segment as well as all apical segments. These findings coincide with a proximal LAD occlusion expected from the 16-segment model. Both MR tagging and RT3DE data were taken from patient data. Strain measures for frames during the systolic phase were used for comparison, which yielded a 2.18% root-mean-squared error in estimating radial strain when compared to the same phase of MR tagging. A strong correlation between strains measured by RT3DE and MR tagging with r=0.91 was observed, which outperformed recent validation studies using 2D methods [1]. Thus our method may provide cardiologists with a new and effective tool for measuring 3D strain and torsion, finally allowing clinicians to routinely measure abnormal wall motion in a quantitative and inexpensive fashion.