TY - JOUR
T1 - Geometrically induced wall shear stress variability in CFD-MRI coupled simulations of blood flow in the thoracic aortas
AU - Perinajová, Romana
AU - Juffermans, Joe F.
AU - Westenberg, Jos J.M.
AU - van der Palen, Roel L.F.
AU - van den Boogaard, Pieter J.
AU - Lamb, Hildo J.
AU - Kenjereš, Saša
PY - 2021
Y1 - 2021
N2 - Aortic aneurysm is associated with aberrant blood flow and wall shear stress (WSS). This can be studied by coupling magnetic resonance imaging (MRI) with computational fluid dynamics (CFD). For patient-specific simulations, extra attention should be given to the variation in segmentation of the MRI data-set and its effect on WSS. We performed CFD simulations of blood flow in the aorta for ten different volunteers and provided corresponding WSS distributions. The aorta of each volunteer was segmented four times. The same inlet and outlet boundary conditions were applied for all segmentation variations of each volunteer. Steady-state CFD simulations were performed with inlet flow based on phase-contrast MRI during peak systole. We show that the commonly used comparison of mean and maximal values of WSS, based on CFD in the different segments of the thoracic aorta, yields good to excellent correlation (0.78–0.95) for rescan and moderate to excellent correlation (0.64–1.00) for intra- and interobserver reproducibility. However, the effect of geometrical variations is higher for the voxel-to-voxel comparison of WSS. With this analysis method, the correlation for different segments of the whole aorta is poor to moderate (0.43–0.66) for rescan and poor to good (0.48–0.73) for intra- and interobserver reproducibility. Therefore, we advise being critical about the CFD results based on the MRI segmentations to avoid possible misinterpretation. While the global values of WSS are similar for different modalities, the variation of results is high when considering the local distributions.
AB - Aortic aneurysm is associated with aberrant blood flow and wall shear stress (WSS). This can be studied by coupling magnetic resonance imaging (MRI) with computational fluid dynamics (CFD). For patient-specific simulations, extra attention should be given to the variation in segmentation of the MRI data-set and its effect on WSS. We performed CFD simulations of blood flow in the aorta for ten different volunteers and provided corresponding WSS distributions. The aorta of each volunteer was segmented four times. The same inlet and outlet boundary conditions were applied for all segmentation variations of each volunteer. Steady-state CFD simulations were performed with inlet flow based on phase-contrast MRI during peak systole. We show that the commonly used comparison of mean and maximal values of WSS, based on CFD in the different segments of the thoracic aorta, yields good to excellent correlation (0.78–0.95) for rescan and moderate to excellent correlation (0.64–1.00) for intra- and interobserver reproducibility. However, the effect of geometrical variations is higher for the voxel-to-voxel comparison of WSS. With this analysis method, the correlation for different segments of the whole aorta is poor to moderate (0.43–0.66) for rescan and poor to good (0.48–0.73) for intra- and interobserver reproducibility. Therefore, we advise being critical about the CFD results based on the MRI segmentations to avoid possible misinterpretation. While the global values of WSS are similar for different modalities, the variation of results is high when considering the local distributions.
KW - Aorta
KW - CFD
KW - MRI
KW - Segmentation
KW - Wall shear stress
UR - http://www.scopus.com/inward/record.url?scp=85105252517&partnerID=8YFLogxK
U2 - 10.1016/j.compbiomed.2021.104385
DO - 10.1016/j.compbiomed.2021.104385
M3 - Article
C2 - 33894502
AN - SCOPUS:85105252517
SN - 0010-4825
VL - 133
JO - Computers in Biology and Medicine
JF - Computers in Biology and Medicine
M1 - 104385
ER -