A numerical investigation on the effect of RF coil feed variability on global and local electromagnetic field exposure in human body models at 64 MHz

Abstract

Purpose: This study aims to investigate how the positions of the feeding sources of the transmit radiofrequency (RF) coil, field orientation direction with respect to the patient, and patient dimensions affect the global and local electromagnetic exposure in human body models.

Methods: Three RF coil models were implemented, namely a specific two-source (S2) feed and two multisource feed configurations: generic 32-source (G32) and hybrid 16-source (H16). Thirty-two feeding conditions were studied for the S2, whereas two were studied for the G32 and H16. The study was performed using five human body models. Additionally, for two of the body models, the case of a partially implanted lead was evaluated.

Results: The results showed an overall variation due to coil feeding conditions of the whole-body specific absorption rate (SAR) of less than 20%, but deviations up to 98% of the magnitude of the electric field tangential to a possible lead path. For the analysis with the partially implanted lead, a variation of local SAR at the tip of the lead of up to 60% was observed with respect to feed position and field orientation direction.

Conclusion: The results of this study suggest that specific information about feed position and field orientation direction must be considered for an accurate evaluation of patient exposure. Magn Reson Med 79:1135-1144, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Keywords: SAR; finite difference time domain (FDTD); interventional MRI; partially implanted leads; virtual population.

FIG. 1

Electrical characterization of the RF coil models. (a) 3-D view of the computational model. The computational RF body coil system was modeled to match the physical coil geometry: (b) S2, (c) G32, and (d) H16. RF, radiofrequency; G32, generic 32-source model; H16, hybrid 16-source model; S2, specific two-source model.

FIG. 2

Numerical setup and data analysis definition. (a) Human body model Duke loaded in the RF coil landmarked at the heart, the black line shows the trajectory used for the E_tan extraction. The ${\mathbf{B}}_{1\text{RMS}}^{+}$ at the central axial slice for the area occupied by the body is shown. All the numerical results were normalized based on the ${\mathbf{B}}_{1\text{RMS}}^{+}$ average equal to 3 μT in the selected area. Head and feet end-rings are defined. (b) S2 feed rotation within the coil end-ring where α defines the central angle of the two feeds at 90° used as 50 Ω voltage sources in the simulations. The field orientation direction (CW and CCW) also is shown. A total of 32 feeding conditions were studied with the specific two-source coil model. (c) Extraction line trajectory shown both inside (P₀–P₁) and outside (P₁–P₂) the body for the E_tan evaluation. (d) Extraction line trajectories for the five human body models used for the analysis. In Glenn and Duke, ${\mathrm{P}}_0^{\ast }$ indicates the point corresponding to the tip of the implant used in the models with partially implanted lead. CCW, counterclockwise; CW, clockwise; E_tan, electric field tangential to a predefined trajectory; RF, radiofrequency; T, tesla; RMS, root mean square.

FIG. 3

WbSAR results for the five human body models used for the analysis. The radar plots report the values with respect to Δπ and Δφ for the S2 model. The bold external circle indicates the coil with each human body model inside plotted in scale. The table reports the WbSAR mean and standard deviations with respect to all the variables: Δπ and Δφ for S2 and Δφ for G32 and H16. G32, generic 32-source model; H16, hybrid 16-source model; S2, specific two-source model;; Std = standard deviation; W, watts; WbSAR, whole body average specific absorption rate.

FIG. 4

║E_tan║ profiles for each human body model studied along the extraction trajectory (Figs. 2c and 2d). The gray area of each plots identifies the extraction section inside the body. Results are shown as stripes (identified by two colored lines), including Δπ and Δφ variability for the S2 model. Profiles obtained with G32 and H16 are plotted as black lines, with Δφ variability reported by solid (clockwise) and dotted (counterclockwise) lines. E_tan, electric field tangential to a predefined trajectory; G32, generic 32-source model; H16, hybrid 16-source model; S2, specific two-source model.

FIG. 5

Values of SAR_tip obtained for the different lead exposures (Δπ and Δφ) with each of the two human body models selected for the analysis: Glenn (a) and Duke (b). Whereas the peak SAR depends on the lead exposure, the spatial distribution around the electrically small tip of the lead depends only on tip geometry. Thus, the profiles reported apply to either the uSAR, 0.1gSAR, 1gSAR, or the 10gSAR normalized to its maximum value (see Table 2; values in bold and italic). CCW, counterclockwise; CW, clockwise; gSAR, gram average SAR, specific absorption rate; SAR_tip, SAR near the lead tip; uSAR, unaveraged SAR; W, watts.

FIG. 6

Map of uSAR around the tip of the partially implanted lead for one of the tested exposure conditions in Glenn and Duke. Values are reported as isosurfaces of SAR level with respect to the maximum value. The highest values were observed at the lead tip at the interface between bare tip and insulation and at the location where the lead exited the vein. SAR, specific absorption rate; uSAR, unaveraged SAR; W, watts.