Using Observed Contrast Motion During Cardiac Catheterization to Tune Patient Specific Coronary Blood Flow Simulations

Stephen Moore, Olivia Smith, Kerry Halupka, Sergiy Zhuk
IBM Research


In the assessment of coronary artery disease, virtual Fractional Flow Reserve (vFFR) is an emerging technology that uses patient specific of Computational Fluid Dynamics simulations to infer the pressure loss through a stenosis, replacing an effective, but expensive technique based on the use of a pressure wire. For an accurate representation of the hemodynamics, appropriate patient specific boundary conditions are crucial. To date, most vFFR efforts make use of reduced order lumped parameter models for the inlet and outlet/s of the coronary arterial tree, but suffer from the inability to specify the associated parameters in a patient specific manner. When applying vFFR in a catheter laboratory setting using X-Ray angiograms as the basis for creating the geometrical model, there is some indirect functional information available through observing the motion of the radio-opaque contrast agent. In this work, we present a novel method for tuning the peripheral resistances. The approach is based on the inclusion of the catheter in the geometrical model and simulating the process of fluoroscopy. Given initial estimates for the outlet peripheral resistances, a simulation of the contrast release can be performed and a misfit function in terms of the observed and simulated arrival times of the contrast agent along the tree can be defined. Based on this approach an iterative method is employed to optimise the resistances in order to match the contrast motion, thereby reconstructing the underlying velocity and pressure fields. We present proof of principle results on synthetically generated coronary geometries with multiple segments, demonstrating that given knowledge of the afferent blood pressure and the contrast release flowrate alone (both of which are available in a catheter laboratory setting), the optimisation algorithm can successfully reconstruct the flowfield, providing a potential new means to improve simulation-based technologies in this area.