High Framerate, GPU-Accelerated ARFI Log(VoA)-Based Carotid Plaque Characterization With Concurrent Vector Doppler-Derived Wall Shear Stress Estimation for Stroke Risk Assessment

Stroke, a primary cause of death and disability worldwide, is most often ischemic and results from carotid atherosclerotic plaque rupture. To prevent ischemic stroke, carotid plaque excision by carotid endarterectomy (CEA) surgery is clinically indicated by the presence of neurological symptoms and high degree of stenosis. However, it is estimated that less than 20% of CEA procedures actually prevent stroke in asymptomatic patients or in symptomatic patients with moderate stenosis. Rather than stenosis grade, plaque rupture risk is more accurately predicted by vulnerable composition and high wall shear stress (WSS), a quantification of the frictional force of blood flow on vessel walls. Therefore, reliable and accessible methods for noninvasively interrogating these hallmarks of plaque rupture could improve stroke risk assessment for better indication of need for CEA. Recently, ultrasound imaging techniques have been developed to separately image vulnerable plaque structure and composition and measure WSS. One such technique for imaging vulnerable plaque morphology, termed log variance of acceleration (log(VoA)), was demonstrated in clinical studies to delineate collagen, calcium, lipid rich necrotic core, and intraplaque hemorrhage. However, limitations of this technique included potential inability to differentiate features in patients with and without stroke-related neurological symptoms and insufficient framerate to concurrently measure WSS. Alternatively, high framerate plane wave (PW)-log(VoA) methods have the potential for simultaneous plaque characterization and WSS estimation. No previous work has combined these approaches for holistic atherosclerotic plaque imaging. This dissertation aims to improve on existing methods for stroke risk assessment in four steps. First, log(VoA)-derived plaque composition alone is shown to be insufficient for differentiating human subjects with and without neurological symptoms. Second, a high framerate PW-log(VoA) approach is demonstrated for plaque characterization in ex vivo human cadaveric carotid arteries. Third, the technology is adapted for concurrent log(VoA)-derived plaque imaging and 2D vector Doppler-derived WSS measurement using GPU-accelerated post processing, with validation by histology and fluid structure interaction (FSI) modeling. Fourth, an in silico framework combining FSI and ultrasound simulations of flow in human carotid geometries is presented to enable further development and validation of ultrasound-derived signal processing and WSS estimation techniques.