In Silico Investigation of the Impact of Catheter Type on Quantitative Digital Subtraction Angiography

Quantitative Digital Subtraction Angiography (qDSA) is a method for estimating arterial blood velocity from DSA images by tracking the movement of iodine contrast oscillations.{1} Previous work in a porcine model of hepatic transarterial embolization demonstrated qDSA can measure in vivo blood flow changes.{2} However, the influence of injection catheter design on qDSA is unknown. In this study, a well-controlled and repeatable Computational Fluid Dynamics (CFD) simulation platform is used to analyze differences in qDSA results obtained with different catheters.

Materials and Methods:

The geometry of the injection catheter was hypothesized to influence the time-varying iodine concentration at the site of injection and subsequent velocity measurements. This was tested in silico using i) a CFD simulation of fluid transport through the catheter and vessel (OpenFOAM, two-fluid mixing, detached eddy turbulence), ii) an x-ray projection simulator that converts 3D iodine concentrations to 2D images, and iii) a published qDSA analysis algorithm.{1} A 6-mm diameter cylindrical vessel was modeled with pulsatile blood flow and mean velocities of 24, 29, 35, 41, and 45 cm/s. 5 Fr end-hole and straight flush catheters were simulated. For each, a 2.0 mL/s injection of iohexol 300 mgI/mL was modeled. CFD simulations were carried out for 7 s and x-ray images were generated at 30 frame/s. The raw image input to the qDSA algorithm was the contrast measured along a vessel segment centerline vs. time. Differences in the raw image inputs for the two catheters were quantified with the normalized root mean square difference (nRMSD). Differences in qDSA results were evaluated by comparing the linear fits of qDSA velocity versus known CFD velocity for each catheter.

Results:

The nRMSD between the raw image datasets for the two catheter geometries was 17% – 82%, depending on blood velocity. For the end-hole catheter, the linear fit between qDSA velocity and CFD velocity had an R^2 value ranging from 0.79 to 0.99 (mean 0.97) and a slope ranging from 0.25 to 1.14 (mean 0.77), depending on the vessel segment analyzed (5-7.5 cm from catheter tip, 5-10 cm length). For the flush catheter, R^2 ranged from 0.90 to 0.99 (mean 0.97) and the slope was 1.92 to 3.42 (mean 2.74).

Conclusion:

This in silico study indicates that catheter type impacts the image input to the qDSA algorithm and the qDSA velocity results. The calibration between qDSA velocity and true velocity should account for catheter type as well as the position of the analyzed vessel segment relative to the catheter tip.