Intraplaque Hemorrhage Quantification using Carotid Magnetic Resonance Imaging

Abstract

As a critical feature of vulnerable atherosclerotic plaque, intraplaque hemorrhage (IPH) is associated with fast plaque progression and subsequent cerebral ischemic events. Although presence/absence of intraplaque hemorrhage is a recognized biomarker for plaque vulnerability assessment, quantitative measurement of the intraplaque hemorrhage signals on MRI may contribute greater utilities to access the evolution of intraplaque hemorrhage and its vulnerability. Recent studies have demonstrated that intraplaque hemorrhage volume and signal intensity could either progress or regress along time and that the change differs between symptomatic and asymptomatic plaques. Therefore, intraplaque hemorrhage quantification may contribute to a precise assessment of atherosclerosis’s clinical risk. However, intraplaque hemorrhage detection and segmentation have mainly relied on manual review, which is not only time-consuming but also prone to measurement errors due to flexible window display level settings. Furthermore, carotid MRI often suffers from complex motion problems, causing degradation of image quality and inaccurate vessel wall delineation. In this dissertation, a semi-automatic method was developed based on histological validation for intraplaque hemorrhage detection and quantification from magnetic resonance imaging (MRI). First, intraplaque hemorrhage detection criteria was established on the widely used Magnetization-Prepared Rapid Acquisition Gradient-Echo (MP-RAGE) sequence and recently developed Simultaneous Noncontrast Angiography and intraPlaque hemorrhage (SNAP) sequence. Both adjacent soft tissue and local median values on MP-RAGE were found to be good intensity normalization references for intraplaque hemorrhage detection, while the sternocleidomastoid muscle on the SNAP reference image was chosen for SNAP image intensity normalization. Second, a volumetric image processing method was developed on 3D SNAP and reproducibility for intraplaque hemorrhage quantitative measures were demonstrated. Lastly, a non-marker-attached motion detection and correction technique based on structured light was proposed for both abrupt and bulk motion correction in carotid MRI for better vessel wall delineation, which could also potentially improve delineation for intraplaque hemorrhage and other plaque components.