Myocardial motion tracking for fast cardiac ultrasound imaging

Cardiovascular diseases are a major public health problem, and their diagnosis relies on a clinical examination that can be challenging. Studies show that the assessment of left-ventricular (LV) myocardial function, i.e. the ability of the heart muscle to propel blood through the circulation, leads to more effective diagnosis and treatment. Myocardial deformation imaging based on echocardiography has recently been introduced to assess regional strain and strain rate (SR). Among the techniques proposed so far, speckle tracking echocardiography (STE), which allows a semi-automatic and angle-independent quantification of myocardial deformation over the cardiac cycle, has been shown useful in a multitude of cardiac conditions. However, in current clinical practice, STE is based on conventional B-mode imaging at relatively low frame rate (FR) that is typically lower than 80 Hz, thereby limiting the time resolution of cardiac mechanics' assessment. With the available FR, some of the cardiac mechanical phases, which are known to be associated with very fast motion and deformation of the myocardium, cannot be studied accurately, although they might contain potentially important information on cardiac (patho-) physiology. Lately, advanced imaging techniques have been developed to enable high frame rate (HFR) ultrasound imaging of the heart and, thus, move towards a better assessment of the cardiac function. Indeed, it was shown that HFR imaging can provide additional information on timing of the cardiac mechanical phases and can improve quality of the time-dependent parameters. The HFR techniques can be divided into two groups: one is based on the simultaneous transmission into different directions of multiple focused beams, while the second one is based on defocused waves (diverging/plane waves), along with massive parallel receive beamforming and coherent compounding. The aim of this thesis was therefore to introduce a novel STE algorithm specifically developed for HFR imaging. Indeed, since the abovementioned STE algorithms operate at low FRs, they cannot be directly applied to HFR imaging as interframe motion becomes very small. Hence, first, the HFR STE method was developed, validated, and tested in vivo. Then, to further improve the performance of the proposed method different HFR scan techniques, beamforming algorithms, and clutter filters were investigated. Afterwards, the novel motion estimator was challenged in clinical settings where it was used to estimate novel biomarkers to assess cardiac diastolic function and to quantitatively analyze stress testing data. Finally, at the end of this PhD project, during an internship at the University of Florence (Italy), HFR STE was brought closer to the clinical environment by estimating motion on a high-end scanner in HFR mode and by implementing a computationally efficient version of the proposed algorithm on an experimental machine.

Publication year:2021

Accessibility:Closed