| Atherosclerosis results in a majority of ischemic events, making it a leading cause of death. The clinical manifestations of human coronary atherosclerosis extend from an asymptomatic state and stable angina to acute coronary events such as myocardial infarction, unstable angina, and sudden cardiac death. Additionally, atherosclerosis is known to exist in two forms, stable or hard plaques and unstable or soft plaques. ‘Stable’ plaques are characterized by a fibrous cap that is thick relative to the volume of lipid and necrosis. The fibrous matrix is rich in collagen and smooth muscle cells, hence hardening or stabilizing the plaque. In the case of soft plaques, particular geometrical features such as a thin fibrous cap and a large necrotic core with high lipid content cause regions of increased circumferential stress in the fibrous cap, making it ‘unstable’ and prone to rupture. Therefore, atherosclerotic plaque stability is related to histologic composition. However, current diagnostic tools do not allow adequate in vivo identification and characterization of plaques. The only methods of interpretation that exist currently are either qualitative and hence not reproducible or those that are based on analysis of texture from images, therefore subject to limitations in image capture. Intravascular ultrasound (IVUS) provides two dimensional cross-sectional views of the vessel, hence accurate information on vessel morphology. IVUS backscattered signals have the potential to provide real-time in vivo plaque composition via spectral analysis. In this dissertation, various spectral algorithms were compared to assess those best suited for IVUS biological signals. A combination of previously identified spectral parameters were employed collectively to characterize the spectra. Classification trees were then used to determine patterns in the parameters pertaining to various plaque components. Pseudo-histology overlays or tissue maps of entire artery cross-sections were visualized. Finally, the validation for these tissue maps was inbuilt since the classification schemes were computed from ultrasound data reflective of only homogeneous plaque components that were selected from corresponding histology. These ex-vivo histological validations display immense potential for assessment of plaque vulnerability. The developed classification trees make real-time analysis of IVUS data conceivable, enabling plaque characterization and further increasing the utility of IVUS. |
