4D flow MRI Improves Computational Fluid Dynamics Analysis of Aortic Dissection
Sylvana García-Rodríguez1, Jon Wrobel1, Alejandro Roldán-Alzate1,2, and Christopher J. François1

1Radiology, University of Wisconsin-Madison, Madison, WI, United States, 2Mechanical Engineering, University of Wisconsin-Madison, Madison, WI, United States


The effects of MRI-derived three-directional velocity profiles implemented at the inlet of aortic dissection (AD) computational fluid dynamics (CFD) simulations were investigated. Two AD models were generated from in vivo MRA data using 3D printing. In vitro 4D Flow MRI was performed on the AD phantoms at two flow rates. Normal and multidirectional blood flow vectors at the AD inlet was measured from 4D Flow MRI data and used in CFD simulations. Significant differences were found in pressure distribution in response to inlet boundary condition definitions. Peak velocity and wall shear stress were also affected by inlet condition definition.


To investigate the effects of MRI-derived three-directional profiles implemented at the inlet of aortic dissection (AD) on computational fluid dynamics (CFD) simulations.


In aortic dissection (AD) a tear in the wall of the aorta results in a false lumen which can lead to complications such as aneurysm growth, rupture, end-organ malperfusion and hypertension.1 CFD, in combination with medical imaging, has been increasingly used for patient-specific cardiovascular modeling of AD. The few existing AD CFD studies use inlet flow profiles based on literature, ultrasound, two-dimensional (2D) flow magnetic resonance imaging (MRI) and normal volunteer data.2-6 Four-dimensional (4D) flow MRI can improve patient-specificity, enabling multidirectional blood velocity measurement.7 This study implemented three-directional velocity inlets in AD CFD models based on an in vitro system that incorporates 3D printed models, allowing simulation of various physiological conditions. Two input boundary conditions were compared: CBC) constant velocity derived from 4D Flow MRI flow quantification and V3BC) three-directional velocity distribution extracted from 4D Flow MRI data.


Following an IRB-approved and HIPAA-compliant protocol, computed tomography angiography (CTA) data from two patients (54 year-old male and 55 year-old female) with acute descending thoracic AD, both of which ultimately required surgical repair, were used to generate models for 3D printing.

Patient Specific Models: In vitro anatomical models were segmented from CTA images using Mimics (Materialise; Leuven, Belgium). Surfaces were fixed and smoothed as needed (3-matic, Materialise), and the geometries were made hollow. SolidWorks was used to add tubing connections at inlets and outlets. Inlet (ascending aorta) and outlet planes (descending aorta and main arteries) were defined. The two geometries were exported in STereoLithography (STL) file format to be used for 3D printing and further CFD pre-processing. The two 3D geometries were printed to scale using selective laser sintering.

In vitro MRI: Each physical model (Figure 1) was connected to a perfusion pump (Stockert SIII Heart-Lung Machine) that circulated water infused with gadofosveset trisodium contrast agent (Lantheus, N. Billerica, MA) at 3 and 4 L/min. In vitro 4D Flow MRI8 was performed on a 3T scanner (MR750, GE Healthcare, Waukesha, WI) with the following parameters: 320 x 320 x 320 mm FOV (1.25 mm isotropic spatial resolution), TR/TE = 6.9/2.2 ms, FA = 15, VENC = 120 cm/s, and scan time 11-12 minutes. 4D Flow MRI data was processed in EnSight (CEI Inc, Apex, NC) to generate velocity streamlines and quantifying volumetric flow rate, including three-component velocity at a plane (inlet plane) in the ascending aorta.

CFD: CFD was performed in Fluent (ANSYS, Inc.; Canonsburg, PA) to simulate the in vitro system. Water fluid properties were used (density 1000 kg/m3; viscosity 0.001 Pa*s). No-slip boundary conditions were defined at rigid walls in accordance with the physical models. Two inlet boundary conditions were compared (Figure 2): Constant boundary condition (CBC) which assumed a constant velocity, normal to the inlet, using 4D Flow MRI-average flow divided by plane cross area; multidirectional velocity boundary condition (V3BC) which assumed three-directional velocities spatially distributed throughout the inlet plane based on 4D Flow MRI measurements. Outlet velocity at the aortic branch arteries was defined as quantified with a flowmeter, and the remaining outlets were assigned zero pressure outlet boundary conditions.

Results and Discussion

Velocity streamlines show a good agreement between in vitro and CFD models (Figure 3), showing areas of vorticity at the same locations in Model 1: ascending aorta and proximal to the tear. Results for Model 1 are presented in Table 1, demonstrating an increase in peak velocity (22% for 3 L/min, 18% for 4 L/min); pressure gradient (50% for 3 L/min, 44% for 4 L/min) and peak wall shear stress (27% for 3 L/min, 26% for 4 L/min) with the implementation of V3BC at the inlet (Figure 4). Renal flow distributions remained very similar with inlet variation. Some of the vorticity of the flow in Model 2 was reproduced in CFD streamlines. However, the complexity of the inlet plane velocity distribution makes the CFD model highly dependent on outlet boundary conditions.


Including the multidirectional velocity information at the inlets of CFD simulations in AD significantly alters peak velocities, pressure gradients, and peak wall shear stress. The results of this study highlight the importance of 4D Flow MRI in providing detailed velocity data in comparison to conventional 2D Flow MRI flow, which was represented by the CBC input. Further studies to explore outlet boundary conditions define the next future steps.


We acknowledge support from the UW Radiology R&D 2015. We also thank GE Healthcare for their support.


1. Khan IA, Nair, CK. Clinical, diagnostic, and management perspectives of aortic dissection. Chest 2002;122:311-328.

2. Chen D, Muller-Eschner, M, von Tengg-Kobligk, H, et al. A patient-specific study of type-B aortic dissection: evaluation of true-false lumen blood exchange. Biomed Eng Online 2013;12:1-16.

3. Cheng Z, Riga, CV, Bicknell, CD, et al. Analysis of flow patterns in a patient-specific aortic dissection model. J Biomech Eng-T ASME 2010;132:051007.

4. Alimohammadi M, Agu, O, Balabani, S, et al. Development of a patienti-specific simulation tool to analyse aortic dissections: asessment of mixed patient-specific flow and pressure boundary conditions. Med Eng Phys 2014;36:275-284.

5. Karmonik C, Bismuth, JX, Davies, MG, et al. Computational hemodynamics in the human aorta: a computational fluid dynamics study of three cases with patient-specific geometries and inflow rates. Technol Health Care 2008;16:343-354.

6. Tse KM, Chiu, P, Lee, HP, et al. Investigation of hemodynamics in the development of dissecting aneurysm within patient-specific dissecting aneurismal aortas using computational fluid dynamics (CFD) simulations. J Biomech 2011;44:827-36.

7. François CJ, Markl, M, Schiebler, ML, et al. Four-dimensional, flow-sensitive magnetic resonance imaging of blood flow patterns in thoracic aortic dissections. J Thorac Cardiov Sur 2013;145:1359-66.

8. Gu T, Korosec, FR, Block, WF, et al. PC VIPR: a high-speed 3D phase-contrast method for flow quantification and high-resolution angiography. AJNR Am J Neuroradiol 2005;26:743-749.


Figure 1. 3D printed physical representation of Model 1 (left) and Model 2 (right), connected to the in vitro system tubing.

Figure 2. Inlet plane velocity vectors measured by 4D Flow MRI (in vitro) and applied to CFD models as input.

Figure 3. Comparison of velocity streamlines from 4D Flow MRI scanning (in vitro) and resulting from CFD models.

Table 1. CFD results for Model 1, comparing effects of inlet boundary conditions.

Figure 4. Pressure and wall shear stress distributions resulting from CFD of Model 1. Arrows point to areas of higher pressure and wall shear stress, especially at 4 L/min.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)