Qifeng Wei

Comparing Shape Tracking, Speckle Tracking, and a Combined Method for Deformation Analysis in Echocardiography

Left ventricular (LV) deformation analysis can provide valuable information about cardiac function. Echocardiography is a non-invasive method for imaging the motion of the heart. Two methods that have been used for quantitative deformation analysis in echocardiography are shape tracking and speckle tracking. Shape tracking provides good displacement values on the boundaries of the myocardium, while speckle tracking provides more accurate tracking across the myocardium. Combining these two complementary sources of information can provide more accurate displacement values over the entire myocardium. These methods combine the two information sources using adaptive radial basis functions over multiple frames. Ultrasound data was acquired on three normal canines. Radial strain values were compared between the shape tracking, speckle tracking, and combined methods to show improvement when using both sources of information. Strain values were calculated from MR tagged data for comparison.

3D Elasticity imaging using principal stretches on an open-chest dog heart

Ultrasound strain imaging has demonstrated its ability to quantitatively assess myocardial viability and contractility altered by myocardial ischemia. However, current ultrasound strain imaging methods still use lower dimensional methods to monitor 3D heart motion. Some 3-D tracking algorithms have also been developed recently in different groups. Quantitative analysis using current methods depends on ultrasound probe orientation and selection of a centroid. To address this, 3D elasticity imaging derived using principal stretches independent of the centroid point is used to assess the contractility of myocardial fibers with 3D data from a commercial 3D scanner on an open-chest dog heart. In this study, an open-chest dog was performed according to a Yale institutional animal protocol and 3D radio frequency (RF) volume data were acquired using an commercial 2D phased array (iE33, Philips, Andover, MA) placed in front of the anterior wall of the left-ventricle with a small water stand-off. 3D speckle tracking was applied to estimate 3D displacement with tracking resolution of 1.2 mm in the axial direction and 4.5 mm in azimuthal and zenithal directions. Volume-to-volume tracking results were accumulated referenced to the heart geometry at the end of diastole until the end of systole. Three principal stretches were estimated using eigenvalue decomposition of the derived right Cauchy deformation tensor at each point. Initial results of strains based on principal stretch and the principal direction has successfully demonstrated that heart wall is thickening along the radial direction and shortening along the longitudinal direction during systole. Unlike 1D or 2D methods, 3D speckle tracking can estimate myocardiums three-dimensional motion. Three strains based on the principal stretches were derived from 3D displacements to assess myocardial contractility independent of probe position and the selection of the centroid point.

Novel method of measuring valvular regurgitation using three-dimensional nonlinear curve fitting of doppler signals within the flow convergence zone

Mitral valve regurgitation (MR) is among the most prevalent and significant valve problems in the Western world. Echocardiography plays a significant role in the diagnosis of degenerative valve disease. However, a simple and accurate means of quantifying MR has eluded both the technical and clinical ultrasound communities. Perhaps the best clinically accepted method used today is the 2-D proximal isovelocity surface area (PISA) method. In this study, a new quantification method using 3-D color Doppler ultrasound, called the field optimization method (FOM), is described. For each 3-D color flow volume, this method iterates on a simple fluid dynamics model that, when processed by a model of ultrasound physics, attempts to agree with the observed velocities in a least-squares sense. The output of this model is an estimate of the regurgitant flow and the location of its associated orifice. To validate the new method, in vitro experiments were performed using a pulsatile flow loop and different geometric orifices. Measurements from the FOM and from 2-D PISA were compared with measurements made with a calibrated ultrasonic flow probe. Results show that the new method has a higher correlation to the truth data and has lower inter- and intra-observer variability than the 2-D PISA method.

Multi-frame radial basis functions to combine shape and speckle tracking for cardiac deformation analysis in echocardiography

Quantitative analysis of left ventricular motion can provide valuable information about cardiac function. Echocardiography is a noninvasive, readily available method that can generate real time images of heart motion. Two methods that have been used to track motion in echocardiography are shape tracking and speckle tracking. Shape tracking provides reliable tracking information on the boundaries of the myocardium, while speckle tracking is reliable across the myocardium. The complementary nature of these methods means that combining them can lead to a better overall understanding of ventricular deformation. The methods presented here use radial basis functions to combine displacements generated from the two methods using information from multiple sequential frames. Ultrasound data was acquired for six canines at baseline and also, for three of these, after myocardial infarction induced by surgical coronary occlusion. Mean segmental radial strain values showed significant decreases in the infarct regions. Comparison to tagged MRI strain values for two of the animals showed good correlation.

3D elasticity imaging on an open-chest dog heart

Myocardial ischemia and infarction alter myocardial viability and contractility. We have hypothesized that contractility changes can be detected by ultrasound strain imaging. Current ultrasound strain imaging methods are mainly 1D and 2D. However, heart motion is complex and 3D. Previous studies showed that 3D speckle tracking on a left ventricular (LV) phantom and a 3D LV simulation reduced low dimensional tracking error. Before applying this method clinically, 3D speckle tracking was tested using 3D RF data acquired with a commercial 3D scanner (iE33, Philips, Andover, MA) on an openchest dog heart. Image data were recorded before and after occlusion of the left anterior descending (LAD) coronary artery to detect acute ischemia induced by this occlusion. Following a local animal protocol, 3D RF volume data were acquired on an open-chest dog heart using a commercial 2D phased array (X7-2, Philips, Andover, MA) placed in front of the anterior wall of the left-ventricle with a small water stand-off. The LAD artery was occluded to produce acute ischemia. Data acquisition before and after occlusion includes 52 volumes/cycle and 46 volumes/cycle (frame rate = 77 Hz), respectively. Each data volume covered 77.5 degrees in the azimuthal direction, 70 degrees in the zenithal direction, and 6.5 cm in depth (anterior wall) at a transmit frequency of 3.8 MHz. 3D speckle tracking was applied to estimate strain with tracking resolution of 1 mm in the axial direction and 7 mm in azimuthal and zenithal directions. Frame-to-frame tracking results were accumulated referenced to the heart geometry at the end of diastole. 3D speckle tracking successfully estimated the displacement in three directions. Radial strains at the end of systole derived from accumulated 3D displacements referenced to the end of diastole in the zoomed region detected abnormal wall thinning due to LAD occlusion. Unlike 1D or 2D methods, 3D speckle tracking can estimate 3D motion and reduce low dimensional tracking error for 3D complex cardiac motion, as suggested in our previous studies. By accounting for out-of-plane motion, strain imaging using 3D tracking may allow more accurate detection of abnormal myocardial deformation associated with myocardial ischemia or infarction.