Abstract 1870: A Novel Method of Spatially-Dense 3D Speckle Tracking on 4D Echocardiography for Strain Analysis: with Validation by Sonomicrometry

Abstract

Background: We have developed a method for spatially-dense 3D speckle tracking that tracks speckle features in myocardium to allow calculation of full 3D displacement and strain from 4D full-volume echo images and have tested its accuracy compared to sonomicrometry in open-chest pigs. Methods: The 4D images were acquired using a Philips iE33 on 6 open-chest piglets. An array of 6 sonomicrometers (sono crystals) was implanted as 2 triangular sets at apex and mid-levels in the LV myocardium. The pair-wise crystal distances were computed offline using Sonometrics® software to yield 3D displacements. For 4D computation of speckle based mechanics, Cartesian coordinate images were entered into an envelope detection program, then a volume-based non-rigid registration was used to obtain optimal displacement between end diastolic (ED) and end systolic (ES) volumes. The displacement field was modeled using B-splines, parameterized by the location of a set of control points. Optimal locations of the control points were obtained by minimizing the sum of squared intensity differences between the ED and ES volumes. Sono crystals were located based on anatomic markers in the ED volume before the adjacent speckle patches were tracked, to ensure that image-derived strains were computed for segments matching the distances between sono crystal pairs. The method was implemented in MATLAB. Results: The comparison of the systolic strains from dense 4D speckle tracking showed a strong correlation (r 2 = 0.82) with sonomicrometry for 44 segmental strains. Conclusions: These studies show that the method we have developed can accurately compute 3D displacement and segmented LV strain from 4D data.