In Partial Fulfillment of the Requirements for the Degree of
Doctor of Philosophy
Will defend his dissertation
Cardiovascular Disease (CVD) is one of the leading causes of death, both in United States and all around the globe. One of the primary causes of CVD is coronary artery atherosclerosis, also known as coronary artery disease (CAD). Recent studies have shown that the presence of calcified coronary plaques, as detected from non-contrast computed tomography (CT) data has a significant predictive value for CAD. Consequently, several CAD risk scores have been developed based on the data collected by CT. However, in spite of a vast amount of CAD-related information available from CT, only a small fraction of it is being used in the existing risk scores. This is due to the lack of robust image analysis methods for the automated extraction of CAD-related information from non-contrast CT data. The long term goal of our research, to which this dissertation contributes, is to develop a set of computational methods for the automated extraction of CVD biomarkers from non-contrast CT data.
The specific objectives of this dissertation are: (i) To develop a general knowledge-based method for the segmentation of organs in medical images with particular emphasis on the incorporation of knowledge-driven constraints into the segmentation problem, and to apply this method to the problem of heart segmentation in non-contrast CT data; (ii) To develop a method for the delineation of the inner thoracic region in non-contrast CT data; and (iii) To develop a method for the estimation of coronary artery zones in non-contrast CT data.
The main contributions of this dissertation are: (i) development of a general knowledge-driven Markov Random Field model for image segmentation that uses prior information about appearance, location and shape to collectively constrain the solution space of the segmentation problem; (ii) development of a graph-based method for the delineation of the inner thoracic region in non-contrast CT data; and (iii) development of a learning-based method for the estimation of coronary artery zones in non-contrast CT data. The accuracy and robustness of the proposed methods is demonstrated by an extensive set of experimental results.