Mapping of the full-tensor myocardial deformation using high-frame-rate ultrasound strain imaging

Abstract

Echocardiography, namely cardiac ultrasound, has been the clinically predominant imaging method to evaluate the structural and functional changes of the heart because of its noninvasiveness, high temporal resolution, portability, non-ionizing radiation, and low cost. Ultrasound strain imaging (USI) methods processed on echocardiograms (or cardiac USI) have been shown capable of detecting abnormal myocardial kinematics (i.e., displacement and strain) in various cardiac dysfunctions and may thus aid clinical diagnosis of cardiovascular disease, the leading killer worldwide. However, cardiac USI remains a less than truly reliable diagnostic tool because of several unsolved challenges. First, the estimation accuracy of displacement and strain in the direction orthogonal (i.e., lateral) to the ultrasound beam is poorer in orders of magnitude than that in the direction along the ultrasound beam (i.e., axial), leading to the unsatisfactory reproducibility and the incomplete full-tensor strain mapping of cardiac USI. Besides, the anisotropic properties of the myocardium stemming from the underlying myofiber arrangement are well-recognized, but their potential impacts on the performance of cardiac USI have been overlooked. In addition, single-focus image reconstruction is clinically utilized to acquire echocardiograms, but acceptable image quality and therefore reliable strain estimates can only be achieved at the focal zone. Meanwhile, the frame rate for full-view echocardiogram acquisitions is relatively low (< 100 Hz), which is insufficient for accurate strain estimation using radio-frequency (RF) data and for capturing short-lived cardiac events. This dissertation therefore aims to tackle the aforementioned challenges by developing a high-frame-rate cardiac USI framework that integrates coherent compounding of unfocused transmit waves and robust cardiac strain estimation. A two-dimensional (2D) cross-correlation-based (CC-based) USI method was first systematically assessed in silico, in vitro, and in vivo. The USI method utilized cross-correlation on RF signals to estimate in-plane tissue displacements and full-tensor strains, and its optimal performance in various kinematic scenarios was explored. Performance of the USI method subjecting to the anisotropic mechanical and acoustic properties was further examined in simulated anisotropic media and in vitro porcine skeletal muscles to investigate the impact of tissue anisotropy on strain estimation. Moreover, the underlying tissue strain and the sonographic signal-to-noise ratio (SNRs) were corroborated to account for the impact of the anisotropic mechanical and acoustic properties, respectively. Finally, a high-frame-rate cardiac USI framework was developed that incorporated the 2D CC-based USI method and coherent diverging wave compounding, with its feasibility of full-tensor myocardial strain mapping assessed using open-chest in vivo animal and transthoracic human experiments. In the open-chest animal study, all the strain components in cardiac and principal coordinates showed good reproducibility and excellent agreement with previously reported measurements from magnetic resonance imaging (MRI), the current gold standard. In the transthoracic human study, myocardial strains could be accurately estimated in the septal wall at relatively smaller depths but were degraded at larger depths under a combined effect of reduced image spatial characteristics. In conclusion, the aforementioned findings collectively demonstrated the feasibility of full-tensor myocardial kinematic assessment using high-frame-rate cardiac USI, which is deemed highly translational towards reliable clinical diagnosis.