Magnetic Resonance Imaging and Computed Tomography Image Fusion–Based Aorta and Coronary Artery Model for In Silico Feasibility Evaluation of Perfusion With an Ascending Aortic Pump Using Computational Fluid Dynamics

Abstract

Left ventricular assist devices are mechanical pumps that provide full or partial support of the circulation in patients with varying degrees of heart failure. These devices are usually connected to the heart through the ventricular apex and placed in an abdominal pocket or pericardial sac, which requires open heart surgery. However, investigators have suggested advantages of catheter deployment of an intra-aortic pump, positioned in the ascending aorta using an anchoring device [1]. This simulation study explores the hemodynamic effects of a continuous flow, axial flow pump deployed in the ascending aorta (AAo), specifically focusing on: (a) perfusion of the coronary arterial circulation and (b) the effect of induced nonphysiologic, swirling flow discharged by the pump on perfusion to head-neck vessels of the aortic arch.A 3D numerical model of an adult aorta with coronary arterial branches was reconstructed by fusion of segmented vascular anatomies extracted from magnetic resonance (MR) angiography and contrast-enhanced computed tomography (CT) data resolving the coronary arteries (Fig. 1(a)). A cylindrical object, representative of the pump, was virtually positioned into the AAo with a diameter ∼2/3rd of the aortic dimension at the midplane of insertion. Incompressible, Newtonian blood flow (ρ = 1060 kg/m3; μ = 0.00371 Pa.s) and rigid, impermeable walls were assumed. The pressure field upstream of the pump was modeled using a validated in-house artificial compressibility computational fluid dynamics solver used extensively in previous studies for in silico cardiovascular flow assessment. A constant AAo inflow was modeled at the aortic root (Reynolds number, Re = 1000), and coronary outlet flow splits were prescribed to ensure normal coronary perfusion of 3% (∼163.28 ml/min) of cardiac output (CO = 5.44 l/min). The running average flow field was analyzed after 1 second of unsteady flow. The flow downstream of the pump was assessed in a second set of steady-state simulations conducted in ansys fluent 14.0 (ANSYS Inc., Cannonsburg, PA). Simulative of a pump outlet, an inlet mass flow rate was specified with local rotational components at the downstream model inlet. Swirling effects of increasing intensity were modeled by specifying the percentage tangential component (0%, 20%, and 40%) of inflow, where no swirling inlet component was an approximation of off-pump aortic flow. Aortic outlets were numerically modeled by defining constant resistance boundary conditions tuned in a manner such that a 40%–60% mass flow split between head-neck vessels and descending aorta (DAo) was maintained.In the upstream model, in order to maintain coronary flow at 3% of CO, during steady AAo outflow, a remarkably low pressure gradient (<0.25 mm Hg) was found necessary to be maintained across the aortic root (Fig. 1(b)). Increased swirling component (from 0% to 40%) modeled at the outlet of the pump resulted in decreased mass flow in the brachiocephalic artery (BCA) (from 1.78 to 1.62 l/min, p < 0.001) with a corresponding increase flow to the DAo (from 3.33 to 3.54 l/min, p < 0.001), owing to the Venturi effect created by virtue of the swirling flow at the base of the BCA. Mass flow rate in the left common carotid and left subclavian artery did not show significant variation with inlet swirling component changes. The mean wall shear stress in the transverse aortic arch was observed to increase and became more uniform as the swirling component was increased (Fig. 1(c)), which was congruent with the noted increases in the DAo flow split.An AAo blood pump is promising in regard to its potential transcatheter deployment. Swirling/helical flow patterns from the axial aortic pumps bear closer resemblance to physiological flows in a normal aortic arch. Device anchoring in the AAo that, by design, facilitates annular reverse flow between pump and AAo walls could potentially avert issues relating to improper coronary perfusion. New boundary condition implementations facilitating more realistic modeling of the suction effects of the AAo pump is a subject of ongoing efforts.