Non-invasive ultrasound monitoring of regional carotid wall structure and deformation in atherosclerosis

Thesis (Ph. D.)--Harvard--Massachusetts Institute of Technology Division of Health Sciences and Technology, 2001.Includes bibliographical references (p. 223-242).Atherosclerosis is characterized by local remodeling of arterial structure and distensibility. Developing lesions either progress gradually to compromise tissue perfusion or rupture suddenly to cause catastrophic myocardial infarction or stroke. Reliable measurement of changes in arterial structure and composition is required for assessment of disease progression. Non-invasive carotid ultrasound can image the heterogeneity of wall structure and distensibility caused by atherosclerosis. However, this capability has not been utilized for clinical monitoring because of speckle noise and other artifacts. Clinical measures focus instead on average wall thickness and diameter distension in the distal common carotid to reduce sensitivity to noise. The goal of our research was to develop an effective system for reliable regional structure and deformation measurements since these are more sensitive indicators of disease progression. We constructed a system for freehand ultrasound scanning based on custom software which simultaneously acquires real-time image sequences and 3D frame localization data from an electromagnetic spatial localizer. With finite element modeling, we evaluated candidate measures of regional wall deformation.(cont.) Finally, we developed a multi-step scheme for robust estimation of local wall structure and deformation. This new strategy is based on a directionally-sensitive segmentation functional and a motion-region-of-interest constrained optical flow algorithm. We validated this estimator with simulated images and clinical ultrasound data. The results show structure estimates that are accurate and precise, with inter- and intra-observer reproducibility surpassing existing methods. Estimates of wall velocity and deformation likewise show good overall accuracy and precision. We present results from a proof-of-principle evaluation conducted in a pilot study of normal subjects and clinical patients. For one example, we demonstrate the combination of 2D image processing with 3D frame localization for visualization of the carotid volume. With slice localization, estimates of carotid wall structure and deformation can be derived for all axial positions along the carotid artery. The elements developed here provide the tools necessary for reliable quantification of regional wall structure and composition changes which result from atherosclerosis.by Raymond C. Chan.Ph.D