Computational modeling of the thermal effects of flow on radio frequency-induced heating of peripheral vascular stents during MRI

Abstract

Purpose. The goal of this study was to develop and validate a computational model that can accurately predict the influence of flow on the temperature rise near a peripheral vascular stent during magnetic resonance imaging (MRI).Methods. Computational modeling and simulation of radio frequency (RF) induced heating of a vascular stent during MRI at 3.0 T was developed and validated with flow phantom experiments. The maximum temperature rise of the stent was measured as a function of physiologically relevant flow rates.Results. A significant difference was not identified between the experiment and simulation (P > 0.05). The temperature rise of the stent during MRI was over 10 °C without flow, and was reduced by 5 °C with a flow rate of only 58 ml min^-1, corresponding to a reduction of CEM₄₃from 45 min to less than 1 min.Conclusion. The computer model developed in this study was validated with experimental measurements, and accurately predicted the influence of flow on the RF-induced temperature rise of a vascular stent during MRI. Furthermore, the results of this study demonstrate that relatively low flow rates significantly reduce the temperature rise of a stent and the surrounding medium during RF-induced heating under typical scanning power and physiologically relevant conditions.

Keywords: MRI; MRI safety; RF-induced heating; medical devices; stents.

Figure 1.

A schematic of the geometrical domains (air, gel phantom and flow channel) used in the simulation. The diameter of the RF coil is 60 cm and the longest dimension of the phantom is 65 cm. The red square denotes the temperature point probe location used in the simulation.

Figure 2.

A schematic showing the location of the equipment in the MR scanner room and the adjacent equipment room. The fiber optic temperature probe and silicone tubing from the peristaltic pump and reservoir were routed through the RF waveguide between the two rooms.

Figure 3.

A photograph of the experimental setup including the flow phantom without gel, the vascular stent positioned on the exterior of the flow channel inside the phantom, and temperature probes. Two additional probes are shown in the photograph that were not used during the experimental measurements. The probe on the right side of the phantom (background) and the probe at the distal end of the stent were not used during the experiment when the phantom was filled with gel and positioned at isocenter in bore of the MRI system. The flow phantom is shown on the patient table of the MRI system prior to advancing to isocenter in the bore of the MRI system.

Figure 4.

The experimental measurements of RF induced temperature rise versus time without flow (dashed grey line) and with a flow rate of 192 mL/min (dashed black line), 2240 mL/min (dashed grey line) and computer simulation without flow (solid lines) at each flow rate, with temperature contours of the stent and gel from the simulation, all at a console-reported SAR of 5.5 W/kg.

Figure 5.

The MR-powered RF temperature rise of the stent at varied volumetric flow rates for the experimental measurements (grey circles) and computer simulation (red diamonds) in the context of thermal dose, CEM₄₃. The physiologic flow rate in the external iliac artery of normal subjects at rest is approximately 0.35 L/min (48). The error bars denote ± 1σ for the experimental measurements.

Figure 6.

A Bland-Altman analysis shows the agreement of the simulation with the experimental measurements at all volumetric flow rates. The mean difference (solid black line) between the simulation and experiment was −0.13 °C and the limits of agreement (black dashed lines) are −0.13 °C ± 1.96(SD), where SD is the standard deviation (σ=0.6827). The upper and lower limits of agreement are 1.21 °C and −1.47 °C, respectively.

Figure 7.

A Bland-Altman analysis showing the average CEM₄₃ values from the experiment and simulation at each volumetric flow rate. The mean difference (solid black line) between the simulation and experiment was −0.94 minutes and the limits of agreement (black dashed lines) are −0.94 min ± 1.96(SD), where SD is the standard deviation (σ=8.4660). The upper and lower limits of agreement are 15.7 min and −17.5 min, respectively.