2021/04/05 by Milad Habibi, Habibi, Milad, Roshan M. D’Souza +5 · 1 citation
Medicine · Engineering · #Cardiovascular Health and Disease Prevention #Acute Ischemic Stroke Management #Medical Imaging and Analysis
paper · pdf · doi:10.48550/arxiv.2104.01971
High-fidelity patient-specific modeling of cardiovascular flows and\nhemodynamics is challenging. Direct blood flow measurement inside the body with\nin-vivo measurement modalities such as 4D flow magnetic resonance imaging (4D\nflow MRI) suffer from low resolution and acquisition noise. In-vitro\nexperimental modeling and patient-specific computational fluid dynamics (CFD)\nmodels are subject to uncertainty in patient-specific boundary conditions and\nmodel parameters. Furthermore, collecting blood flow data in the near-wall\nregion (e.g., wall shear stress) with experimental measurement modalities poses\nadditional challenges. In this study, a computationally efficient data\nassimilation method called reduced-order modeling Kalman filter (ROM-KF) was\nproposed, which combined a sequential Kalman filter with reduced-order modeling\nusing a linear model provided by dynamic mode decomposition (DMD). The goal of\nROM-KF was to overcome low resolution and noise in experimental and uncertainty\nin CFD modeling of cardiovascular flows. The accuracy of the method was\nassessed with 1D Womersley flow, 2D idealized aneurysm, and 3D patient-specific\ncerebral aneurysm models. Synthetic experimental data were used to enable\ndirect quantification of errors using benchmark datasets. The accuracy of\nROM-KF in reconstructing near-wall hemodynamics was assessed by applying the\nmethod to problems where near-wall blood flow data were missing in the\nexperimental dataset. The ROM-KF method provided blood flow data that were more\naccurate than the computational and synthetic experimental datasets and\nimproved near-wall hemodynamics quantification.\n