Detailed Evaluation of Five 3D Speckle Tracking Algorithms Using Synthetic Echocardiographic Recordings

Abstract

A plethora of techniques for cardiac deformation imaging with 3D ultrasound, typically referred to as 3D speckle tracking techniques, are available from academia and industry. Although the benefits of single methods over alternative ones have been reported in separate publications, the intrinsic differences in the data and definitions used makes it hard to compare the relative performance of different solutions. To address this issue, we have recently proposed a framework to simulate realistic 3D echocardiographic recordings and used it to generate a common set of ground-truth data for 3D speckle tracking algorithms, which was made available online. The aim of this study was therefore to use the newly developed database to contrast non-commercial speckle tracking solutions from research groups with leading expertise in the field. The five techniques involved cover the most representative families of existing approaches, namely block-matching, radio-frequency tracking, optical flow and elastic image registration. The techniques were contrasted in terms of tracking and strain accuracy. The feasibility of the obtained strain measurements to diagnose pathology was also tested for ischemia and dyssynchrony.

A Framework for the Generation of Realistic Synthetic Cardiac Ultrasound and Magnetic Resonance Imaging Sequences From the Same Virtual Patients

“…Consequently, regional strain analysis is currently applied as a supplementary diagnosis method, but is still not ready for full clinical implementation in spite of their potential impact in many applications [6] [12]. One of the reasons is the fact that a well-established validation of regional strain imaging is still missing at the moment, despite the existence of a large number of dedicated quantification algorithms both in US [13] and tagged MR [14]. A comprehensive and thorough quantitative validation 2 of these methods based on dense strain measurements would thus represent a major progress.…”

Cardiac motion estimation with dictionary learning and robust sparse coding in ultrasound imaging

“…where α p ∈ R q is a sparse vector and q > n. The learning of a dedicated overcomplete dictionary can be performed either offline, using a set of ground-truth motions, or online, using the current estimation. In this paper, two dictionaries have been trained offline using highly realistic simulated motion fields [11]. The resulting dictionaries allow us to capture typical patterns of vertical and horizontal cardiac motions.…”

Optical flow estimation in ultrasound images using a sparse representation

“…Recent advances have made possible to simulate realistic cardiac US image sequences with ground-truth [17], enabling the use of learning-based methods for cardiac flow estimation. In this context, we propose to use a learned flow dictionary that captures typical patterns of cardiac motion.…”

“…The synthetic evaluation dataset provided in [17] contains simulated US sequences, which are to our knowledge among the most realistic in the recent US literature. The ground-truth flow fields provide the possibility of training the flow dictionaries as well as evaluating the OF estimation accuracy.…”