Maximum Likelihood Motion Estimation in 3D Echocardiography through Non-rigid Registration in Spherical Coordinates

Abstract

Abstract. Automated motion tracking of the myocardium from 3D echocardiography provides insight into heart's architecture and function. We present a method for 3D cardiac motion tracking using nonrigid image registration. Our contribution is two-fold. We introduce a new similarity measure derived from a maximum likelihood perspective taking into account physical properties of ultrasound image acquisition and formation. Second, we use envelope-detected 3D echo images in the raw spherical coordinates format, which preserves speckle statistics and represents a compromise between signal detail and data complexity. We derive mechanical measures such as strain and twist, and validate using sonomicrometry in open-chest piglets. The results demonstrate the accuracy and feasibility of our method for studying cardiac motion.

Three-dimensional myocardial strain estimation from volumetric ultrasound: Experimental validation in an animal model

“…These observations are consistent with a previous clinical study in which we compared the 3D segmental strain estimates against those obtained with 2D techniques (r=0.63 for LL , r=0.41 for CC and no significant correlation for RR ) [10]. Myronenko et al developed a comparable registration method, but they only reported sonomicrometry correlations for global twist and results for strain measurements were lumped for all directions [5]. These findings are also in line with the performance of the current state-of-the-art commercial 3D speckle tracking methods, which typically rely on block-matching based algorithms.…”

“…In the ultrasound society these are traditionally subdivided into either phase-based techniques applied on the radio-frequency data using correlation, time-or phase-shift techniques to estimate motion [2]; and block-matching techniques on Bmode data [3] [4]. A third approach is to use image registration techniques where the cardiac deformation field is parametrized using smooth basis functions [5][6] [7]. Our lab has previously presented such a 3D strain estimation method (3DSE-method) based on elastic registration of subsequent volumes [6].…”

Temporal diffeomorphic free-form deformation: Application to motion and strain estimation from 3D echocardiography

“…This agreement was quantified using 31 This is a pre-print version The final version can be downloaded from http://www.sciencedirect.com/ covariance indexes of average and segmental strain curves. Future extension of the TDFFD algorithm will focus on incorporating similarity metrics adapted to statistical characteristics of US speckle noise (Myronenko et al (2009)) and its extension to incorporate compounding strategies to improve the limited field-of-view in 3D US sequences of heart failure patients with dilated LV similar to Piella et al (2011). Velocity field at spatiotemporal location x, t parameterized by p x k 0 (p) := ϕ0(x, t k , p) Trajectory of a point x from time t = 0 to time t k q i,j,k,l := (q i , q j , q k , q l )…”

Comparison of image registration methods for composing spectral retinal images

“…The nal transformation is obtained by b-spline interpolation using the grid nodes as control points. Several similarity measures were quantitatively evaluated in this framework: Correlation coe cient (CC), Similarity measure by Myronenko et al (CD2) [8], Similarity measure by Cohen and Dinstein (MS) [2], Mutual information (MI) [15], Minimization of residual complexity (RC) [7], Sum of absolute di erences (SAD) and Sum of squared di erences (SSD).…”