ANALYSIS OF MULTIDIMENSIONAL PATIENT DATA WITH CARDIOVASCULAR DISEASES THROUGH MEDI- CAL IMAGE PROCESSING, MODELING, AND MACHINE LEARNING TECHNIQUES
Date Issued
January 15, 2025
Type
Διδακτορική Διατριβή
Abstract
Cardiovascular diseases remain the leading cause of mortality globally, driving the need
for improved diagnostic and predictive tools. This thesis integrates computational fluid
dynamics (CFD), radiomic analysis, and automated segmentation techniques with ad-
vanced imaging modalities to enhance the assessment of coronary artery disease (CAD)
and systemic vascular conditions.
Endothelial shear stress (ESS), calculated through CFD models based on coronary com-
puted tomography angiography (CCTA), is evaluated in relation to stenosis severity,
plaque volume and myocardial perfusion. The research explores how variations in ESS
correlate with functional changes in myocardial perfusion, as measured by positron
emission tomography (PET). These findings provide insights into the role of hemody-
namic forces in CAD and underscore the potential of ESS as a diagnostic marker.
Radiomic analysis is applied to CCTA-derived plaque features to extract biomarkers
predictive of myocardial perfusion abnormalities. The integration of these biomarkers
with PET and single-photon emission computed tomography (SPECT) data signifi-
cantly enhances diagnostic accuracy compared to traditional imaging metrics. This ap-
proach highlights the utility of radiomics in non-invasive CAD risk stratification.
Additionally, the thesis develops automated segmentation methods for analyzing the
aorta in PET/CT imaging, enabling precise and reproducible assessment of metabolic
activity in large-vessel vasculitis. These techniques improve workflow efficiency and
provide new tools for evaluating systemic inflammatory conditions.
By combining ESS analysis, radiomic biomarkers, and automated segmentation, this
thesis establishes a comprehensive framework for non-invasive cardiovascular diag-
nostics. The findings contribute to advancing patient-specific risk assessment and man-
agement strategies, with significant implications for clinical practice.
Subjects