High-frame-rate contrast-enhanced echography for myocardial perfusion assessment

Research output: Types of ThesisDoctoral ThesisInternal

15 Downloads (Pure)

Abstract

Cardiovascular disease remains a significant global burden, standing as the leading cause of death. Annually, millions of people are diagnosed with angina and myocardial infarction (MI) worlwide, with over 3 million cases reported in the European Union alone. Acute MI, particularly ST-segment elevation MI (STEMI) is associated with substantial morbidity and mortality. Prompt primary percutaneous intervention (PPCI) is the preferred treatment, reopening occluded vessels with a high success rate. Despite this, the no-reflow (NR) phenomenon, characterized by microvascular obstruction, can lead to inadequate myocardial perfusion even after successful epicardial flow restoration, thereby increasing the risk of adverse outcomes post-MI.
Current diagnostic methods used to assess myocardial perfusion and diagnose NR have certain limitations, including varying levels of accuracy and potential harm if performed repeatedly due to the presence of ionizing radiation. Echocardiography emerges as a promising option due to its accessibility, cost-effectiveness, and absence of ionizing radiation. The adoption of gas-filled coated bubbles with the size of few micrometers (microbubbles) as ultrasound contrast agents (UCAs) has improved echocardiography sensitivity to detect blood flow. However, challenges remain in producing reliable perfusion images due to noise, image artifacts, variability, and operator dependence despite the use of UCAs.
This thesis addresses these challenges by developing contrast imaging techniques that improve contrast-enhanced ultrasound (CEUS) image quality and contrast detection compared to the standard imaging methods. It could lead to CEUS becoming a reliable method for assessing myocardial perfusion and monitoring NR. This would have significant clinical implications, improving decision-making in patient treatment and understanding NR mechanism.
First, in Chapter 2 we explored the capabilities and limitations of the spatiotemporal implementation of SVD as a clutter filter to separate microbubble and tissue signals. We conducted an in vitro experiment and found that SVD is ineffective in detecting slow flow during tissue motion, which is the condition of myocardial perfusion. Thus, we introduced Independent Component Analysis (ICA) as a post-processing technique to improve contrast detection by exploiting the distinct statistical distributions of microbuble and tissue signal. Our in vitro results show that ICA improves SVD detection by 7-10 dB during motion.
Next, we incorporated the nonlinear response of microbubbles to enhance contrast detection. The standard technique to exploit this property is the Multi-Pulse Contrast Scheme (MPCS), including techniques like pulse inversion (PI), amplitude modulation (AM), and their combination (AMPI). On top of the motion artifacts, the nonlinear propagation through the microbubble cloud also reduces contrast detection when using the MPCS technique. In Chapter 3, we developed a contrast detection technique using higher-order singular value decomposition (HOSVD), a generalization of SVD that works on high-order tensors. HOSVD is applied to a beamformed IQ image series, with spatial, temporal, and pulsing sequence as the input dimensions. We conducted both in vitro and in vivo experiments to evaluate the efficacy of this technique in comparison to the current standard methods. in vitro results, particularly in scenarios involving motion, demonstrated that HOSVD outperformed the existing standard contrast detection schemes by over 10 dB. We also validated the viability of HOSVD implementation in a more realistic in vivo experiment utilizing a cardiac porcine model. HOSVD showed the superior capability to detect microbubble signal within the myocardium, surpassing the AM contrast-to-background ratio by up to 19 dB. On Chapter 4, we furthered our investigation with the porcine cardiac model to visualize coronary vascular dynamics and identify perfusion deficits by inducing occlusion in the left anterior descending (LAD) artery. We successfully differentiated between fast and slow coronary flows, assumed to represent flow in larger vessels and perfusion. This distinction enhanced our confidence in evaluating myocardial perfusion. Moreover, we accurately visualized the affected area in the myocardium caused by the LAD occlusion.

To improve reproducibility and errors due to out-of-plane motion, we explored the implementation of 3D ultrasound imaging. The 3D part of the thesis begins with Chapter 5. We developed an image reconstruction technique for data acquired with the in-house developed spiral array probe. We adjusted the "lags" on the spatial coherence (SC) beamforming and show it capabilities to improve image quality compared to traditional Delay and Sum (DAS) technique. Our in vitro and in vivo experiments, employing a chicken embryo model results showed that SC outperformed conventional DAS beamforming by up to 25 dB. In chapter 6, the focus shifts to in vivo studies using big animal model. We utilize porcine kidney model to visualize renal cortex microcirculation and measure Doppler velocity in vessels. The results are validated against commercial Doppler and contrast-enhanced imaging. Super-resolution imaging is introduced to enhance the visualization of microvascular structures.
Chapter 7 builds upon the previous techniques by implementing HOSVD on high-frame-rate volumetric images, beamformed with SC technique. In vitro and in vivo experiments demonstrate the improved contrast detection capabilities of this approach. The chapter concludes with successful visualization of the coronary artery using HOSVD and SC on a porcine model.

Lastly, in Chapter 8, we presented the key findings, challenges, and limitations of this thesis. We also shared some technical insights based on unpublished results and provided technical suggestions and clinical implications for future research.
Original languageEnglish
Awarding Institution
  • Erasmus University Rotterdam
Supervisors/Advisors
  • van der Steen, Ton, Supervisor
  • Vos, Rik, Co-supervisor
Award date17 Jan 2024
Place of PublicationRotterdam
Print ISBNs978-94-6469-692-9
Publication statusPublished - 17 Jan 2024

Fingerprint

Dive into the research topics of 'High-frame-rate contrast-enhanced echography for myocardial perfusion assessment'. Together they form a unique fingerprint.

Cite this