Dissertations & Theses (Open Access)

NONPARAMETRIC STATISTICAL AND MACHINE LEARNING METHODS IN ASSESSING CARDIAC RISK USING POSITRON EMISSION TOMOGRAPHY SCANS

WEILU HAN, UTHealth School of Public Health

Abstract

Coronary Artery Disease (CAD) mainly results from a diffuse process of cholesterol deposition, scarring, and calcification (hardening of the arteries) throughout the major coronary arteries supplying blood flow to the heart muscle. CAD is the most common form of heart disease in the United States and the primary cause of heart attacks. More than 12 million Americans have been diagnosed with this disease 1. In the U.S., CAD accounts for one-third of all deaths each year. Twenty to forty percent of middle-aged people have early or advanced coronary artery disease due to atherosclerosis, most without symptoms or knowledge of their condition. In atherosclerosis, cholesterol builds up in pockets (plaques) embedded in the walls of coronary arteries beneath their inner lining. Risk factors for CAD includes age, family history, smoking, obesity, and high blood pressure. The University of Texas Health Science Center at Houston and The Weatherhead PET Center is where PET Imaging of the heart was first developed and continues to be the most advanced center for helping physicians solve complicated patient CAD management problems. PET diagnoses heart disease non-invasively with 96-98% accuracy in individuals with or without symptoms of the disease, permitting treatment even before symptoms appear. In the last 10 years, there have been a paradigm shift in CAD care utilizing PET as a key to diagnosis and clinical decision-making.