Carotid atherosclerosis, a disease in which plaque builds up inside the vessel wall, is a major cause of ischemic stroke. Traditionally, atherosclerosis risk stratification is heavily based on... Show moreCarotid atherosclerosis, a disease in which plaque builds up inside the vessel wall, is a major cause of ischemic stroke. Traditionally, atherosclerosis risk stratification is heavily based on the percentage of stenosis. However, a growing body of evidence suggests that luminal stenosis may not be the only cause of symptoms but the plaque composition may be more likely to impact the disease outcome. High-resolution vessel wall magnetic resonance imaging (VWMRI) is one of the most promising modalities for visualizing and evaluating carotid atherosclerotic plaque. The quantitative assessment of carotid atherosclerotic disease requires vessel wall segmentation and plaque classification, which is generally performed by manual delineations. However, manual contour tracing is labor-intensive, time-consuming and subject to inter-observer and inter-scan variability, which makes manual image analysis impractical for studies where large volume of data needs to be processed. Therefore, the main goal of this thesis is to: 1) develop approaches to automatically, robustly and reproducibly segment the carotid vessel wall and classify the atherosclerotic plaque from multi-spectral VWMRI; 2) validate the developed methods with reference standard; 3) extract the imaging biomarkers that can assist carotid artery disease evaluation. Show less
