Browsing by Author "Franco Leiva, Pamela Alejandra"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Hemodynamics and biomechanics assessment of the heart and great vessels by cardiovascular magnetic resonance imaging
    (2022) Franco Leiva, Pamela Alejandra; Uribe Arancibia, Sergio A.; Pontificia Universidad Católica de Chile. Escuela de Ingeniería
    Cardiac magnetic resonance imaging (MRI) is the gold standard technique for assessing cardiac function. Moreover, cardiac MRI also provides a unique technique called 4D Flow MRI that includes velocity images of the three-orthogonal planes within a 3D volume for the entire cardiovascular system throughout the cardiac cycle. It allows obtaining several hemodynamic parameters providing the evaluation of several cardiovascular diseases. Nevertheless, 4D Flow MRI and processing methods suffer from several issues, e.g., prolonged scanning times, incorrect flow measurements, and missing the clinical relevance through calculating several hemodynamic parameters. In this Thesis, three research articles intended to tackle some of these previous issues. The first article compares the uni-directional Dual Velocity-Encoding (VENC) PC-MRI methods for different noise levels and proposes a correction algorithm for the Optimal Dual-VENC (Carrillo et al., 2018), which is based on theoretical considerations. The second article describes a methodology for quantitative evaluation of intraventricular hemodynamics using a single segmentation from a 4D Flow dataset and a finite-element method. Our approach was able to identify abnormal flow patterns in a small cohort of dilated cardiomyopathy patients and can be applied to any other cardiovascular disease. The third article provides a comprehensive overview of the relative performance of different machine learning algorithms applied over 4D flow data for bicuspid aortic valve aortopathy classification. For that purpose, we analyzed and extracted multiple correlation patterns of hemodynamic parameters, finding which parameters showed high collinearity between them, which allows us to diminish their size to a few variables. This investigation of this thesis for assessing different Dual-VENC reconstruction techniques, image processing, data quantification, pattern recognition, and machine learning in three independent articles. Thought the fact that the topics aborded in the articles were not tested together, future research may combine all these topics to investigate and improve the examination in the cardiovascular system.