A Simulation Study of a Novel Patient-Adjustable MRI Coil for Safe Pediatric Imaging in Children with Cardiac Implantable Electronic Devices (CIEDs)

A substantial and growing population of individuals with conductive implants face persistent challenges in accessing magnetic resonance imaging (MRI) due to risks associated with radiofrequency (RF) heating of implants. Recently, a novel approach based on altering MRI electric fields at the location of individual’s implants has shown promise in substantially reducing RF heating in adult patients with deep brain stimulation devices. In this study, we present the results of electromagnetic and thermal simulation studies demonstrating the applicability of this novel technique in pediatric patients with cardiac implantable electronic devices (CIEDs) at 1.5T MR. The outcomes contribute valuable insights into the development of new MRI transmit coils that can be mechanically adjusted on a patient-specific basis to accommodate patients with conductive implants, particularly in the pediatric population.

SECTION I. Introduction

Magnetic Resonance Imaging (MRI) has revolutionized our approach to understanding human biology and pathology. Unfortunately, despite continuous advances, MRI remains inaccessible to a large and growing group of patients with conductive implants [1], [2]. The problem is exacerbated in children, for whom MR-conditional devices are not readily available, and manufacturers are less willing to invest as the pediatric market is smaller than the adult market.

The major risk of MRI in patients with conductive implants is radiofrequency (RF) heating of the tissue due to the antenna effect. This occurs when the electric field of MRI scanner couples with the implanted device, amplifying the specific absorption rate (SAR) of the radiofrequency energy in the tissue surrounding the implant. Serious injuries have underscored the severity of this problem [3]. Over the past decade, various methods have been suggested by us and others to reduce RF heating in active implantable medical devices (AIMDs) during MRI. These methods include the use of novel implant materials [4], [5] , exploration of MRI platforms with alternative field polarization [6], [7] and surgical modification of implant trajectory to minimize its interaction with MRI fields [8 -​13]. An alternative approach involves altering the MRI technology itself to accommodate patients with implants. This concept, recently explored successfully in adults with deep brain stimulation (DBS) devices [14], [15], takes the form of field-shaping methods where the MRI transmit coil is manipulated to generate a low electric field at the location of an individual patient’s implant, significantly reducing RF heating [16], [17]. Here, we present the results of the first comprehensive simulation study demonstrating the applicability of this concept to children with cardiac implantable electronic devices (CIEDs) at 1.5T MR.

We simulated a linearly polarized MRI birdcage transmit coil, known to generate a slab-like region of zero electric field within its load. We demonstrated that by rotating the coil around bodies of children with CIEDs, it is possible to align the implanted device with the null-E field region of the coil. This alignment, in turn, reduced RF heating by 93% on average compared to using the conventional body coil that is integrated with the MRI scanner. Additionally, we simulated the transfer function (TF) of lead models and demonstrated that once calibrated for a single coil position, the TF can reliably predict RF heating of leads when the coil rotates to different angles around the body. The latter is particularly important, as it enables a quantitative assessment of in-vivo RF heating generated by the rotating coil across large patient populations.

SECTION II. Methods

A. Pediatric Body Model, CIED Configurations, and Rotating MRI Coil

A simplified pediatric body model consisting of three tissue classes was created from segmented MRI images of a 29-month-old child. The images were post-processed to form a tetrahedral mesh suitable for finite element simulations (Figure 1). A model of an unshielded, low-pass, linearly polarized MRI birdcage coil was developed based on a physical prototype built in our lab (16-leg, length = 290 mm, diameter = 350 mm). The coil was tuned to 64 MHz, corresponding to MRI at 1.5 T, using 10 pF capacitors distributed along the legs. The pediatric body model was positioned within the coil with the heart at the isocenter.

Eight distinct CIED lead models were created to represent both clinically realistic endocardial and epicardial lead trajectories, as well as scenarios known to generate excessive RF heating (See Figure 2 for more details on trajectory selection). The lead models consisted of a 25 cm copper wire with a conductivity (σ₁) of 5.8×10^7 S/m, a relative permittivity (ε_r1) of 0.99, and a cross-section of 0.8 mm. This wire was embedded in urethane insulation with a conductivity (σ₂) of 0 S/m, a relative permittivity (ε_r2) of 3.5, and a thickness of 1.6 mm, featuring a 1 cm exposed tip.

Figure 1:

(A) Segmented MRI data of a 29-month-old child were utilized to generate 3D models representing the child's silhouette, brain, and heart. Distinct conductivity and permittivity values were assigned to each organ to represent the heterogeneity of the body. (B) The tetrahedral meshes underwent post-processing for finite element simulations. (C) A 3D rendering view (left) and simulated body model (right).

The coil was rotated around the pediatric body model in a total of 10 positions. To obtain an accurate temperature profile, a high-resolution mesh region surrounding the entire implant was created for the electromagnetic simulations. The initial mesh was configured such that the maximum element size was less than 1 mm on the entire wire core, less than 2 mm for the wire insulation and the tissue region surrounding the entire lead, as well as for the lead insulation.

The simulations were also conducted utilizing a conventional Siemens body coil operating in circular polarization (CP) mode and tuned to 64 MHz. The specific coil parameters were reported in our previous study [8]. The input power of the coil was adjusted to generate an average ${\text{B}}_1^ +$ of 2μT on an axial circular plane (with a 2 cm diameter) for each lead trajectory at each angle. The ANSYS High Frequency Structure Simulator (HFSS) convergence criterion was set to limit the maximum change in the magnitude of scattering parameters (ΔS), with a threshold of ΔS ≤ 0.02.

Figure 2:

Lead trajectories implanted into the phantom for calibration and validation of the simulated transfer function. Care was taken to include orthogonal trajectories to maximize the variability of incident electric field across different cases as suggested in [18].

The electromagnetic (EM) simulation was then integrated with the transient thermal solver of ANSYS Workbench 2021 R1 to evaluate the temperature rise (ΔT) near the tip of the implant similar to our previous works [19], [20]. The simulation employed Pennes’ bioheat equation, excluding perfusion, with specific heat capacity and isotropic thermal conductivity values assigned for each model component (Figure 3). The system was subjected to 280 seconds of RF radiation at 1.5 T, and the resulting ΔT near the lead tip was recorded for each trajectory as the coil was rotated through different angles.

B. Transfer Function Simulation and Validation

The concept of the lead transfer function, originally introduced by Park et al. to simplify electromagnetic simulations [21] , has been adopted by the ISO TS 10974 [22] guideline as the gold standard approach for the prediction of in-vivo RF heating of elongated implants in an MRI environment. The technique involves developing and validating a mathematical model of the implanted lead that quantifies the lead’s response to a well-controlled incident electric field. This model is then used to predict RF heating of the lead when exposed to in-vivo MRI fields, which can be estimated from electromagnetic simulations in realistic body models. Specifically, the predicted RF heating can be expressed as: $$\begin{equation*}\Delta {T_{predicted{\text{ }}}} = C\mid \int_0^l {\left( {{{\left. {TF(x)*{E_{tan}}(x)dx} \right|}^2}} \right.} \tag{1}\end{equation*}$$

Here TF(x) is the lead’s transfer function, giving the level of RF heating observed at the lead’s tip in response to exposure to an infinitesimal tangential electric field impinging on the lead at position x along its length. E_tan is the tangential component of an arbitrary incident electric field along the length of the lead, which generates $\Delta {T_{predicted}}$. C is a calibration factor that can be determined by measuring RF heating along an arbitrary lead trajectory and extracting E_tan along that trajectory from numerical simulations as follows: $$\begin{equation*}C = \frac{{\Delta {T_{simulated{\text{ }}}}}}{{\mid \int_0^l {\left( {{{\left. {TF(x)*{E_{tan}}(x)dx} \right|}^2}} \right.} }}\tag{2}\end{equation*}$$

In a conventional MRI scanner, the TF only needs to be calibrated once, and then it is validated by comparing the predicted RF heating with experimentally measured RF heating for a set of lead trajectories that generate a dynamic range of heating. However, as a rotating MRI coil could potentially behave like multiple distinct coils, it is unclear how the concept of the transfer function would apply to it. Here, we examined whether the transfer function of a simulated CIED lead, calibrated at a single coil position, could reliably predict RF heating of leads along different trajectories and at other coil positions.

SECTION III. Results

We found that for all trajectories, there existed an optimal coil rotation angle that effectively reduced temperature increase (ΔT) at the tips to well below 1°C. On average, we found a substantial reduction of approximately 93% in RF heating compare to a conventional body coil. Figure 4 illustrates the alignment of the low E-field slab on an axial plane when the coil was positioned in its optimal angle for each lead trajectory.

Figure 3:

(A) Electromagnetic (EM) simulation setup showing the MRI coil rotated to positions φ=22.5° and φ=90° from the reference position where the feed was in front of patient’s nose (φ=0°). (B) Thermal simulations. The location of the inserted temperature probe is indicated. The overlaid temperature profile was overlaid on the heart and the wire for ID5. (C) The simulated temperature rise for an example lead trajectory (ID5) when the coil was positioned at φ=90°, featuring a 280-second RF exposure and a subsequent 50-second cooling period. The initial temperature of the simulation was set to room temperature (22°C). The specific heat and thermal conductivity values assigned to each model were presented in the table.

The transfer function (TF) of the lead was simulated using the reciprocal method [23]. In brief, the tip of the lead was excited with a monopole antenna, and the induced electric current along the lead was recorded, giving the uncalibrated TF. The magnitude and phase of the transfer function are illustrated in Figure 5B. The TF was calibrated such that the simulated and predicted temperature rises matched for the lead following trajectory ID3 (See Figure 2) in the coil positioned at φ=90°. Applying the same calibration factor (C) to the remaining trajectories and at different coil angles led to a robust agreement between simulated and predicted ΔT, with a correlation coefficient of 0.89. This demonstrates the accuracy of TF predictions aligning with simulated temperatures (Figure 6).

Figure 4:

(A) The body model within the linearly polarized rotating coil, illustrating the orientation of the low E-field slab on an axial plane. This depiction is provided for both the rotation to the coil's optimal position (on the right) (φ=315°) and the worst-case position (on the left) (φ=247.5°) for a specific trajectory (ID1). (B) The low E-field slab on the axial plane is depicted for the remaining seven trajectories when the coil is in the optimal position.

Figure 5:

(A) Transfer function simulation setup. A polyline, located 2 mm away from the center of the implanted wire, was created to extract the H-field which was used as a proxy for induced currents. (B) The magnitude and phase of the transfer function.

Figure 6:

Bar plot illustrating the simulated (in orange) and predicted (in green) ΔT varying the angle rotation of the MRI coil. The x-axis of each subplot progresses from left to right, representing the range from φ₁ to φ₁₀.

SECTION IV. Discussion and Conclusions

Cardiac pacing holds a crucial role in cardiology and healthcare, significantly influencing the management of diverse cardiac rhythm disorders and improving the quality of life for affected individuals. However, the potential risks associated with exposure to MRI, particularly RF heating at the tissue/tip interface of the implanted pacing device, restrict the use of advanced diagnostic opportunities. Especially in the case of children, where epicardial pacing systems are primarily employed by sewing the cardiac lead directly to the myocardium, the absence of MR-conditional labeling poses a challenge [24]. There are several emerging methods to address these concerns, including modifications to lead design and materials [25], [26], changes to MR technology utilizing multi-channel transceiver coils and parallel transmit (pTx) technology [17], or the implementation of surgical modification techniques [27]. While these approaches have demonstrated promising results, they face constraints such as the high cost-to-benefit ratio associated with medical device manufacturers. Additionally, notable drawbacks in clinical applications, such as B1+ inhomogeneity impacting imaging quality, and considerations in the operating room, need to be carefully addressed for their effective implementation.

An alternative approach was recently suggested featuring a mechanically rotatable linearly polarized MRI transmit coil, which created a low electric field region, strategically aligning with implant locations to alleviate RF heating during MRI in patients with DBS devices. Encouraged by the success observed in DBS devices, our investigation took a stride by assessing generalizability of this concept for body imaging in children with CIEDs. Our results, derived from comprehensive thermal simulations, revealed existence of an optimal rotation angle for a representative cohort of lead configurations implanted in a child body model, yielding an average reduction of 93% in RF heating. Additionally, we presented the outcomes of simulated, calibrated, and validated transfer function (TF) analyses for the implanted lead, demonstrating a strong alignment with simulated temperature increases. This not only facilitates a quantitative assessment of in-vivo RF heating generated by the rotating coil but also lays the foundation for subsequent studies to extend this evaluation across a diverse patient population.

It is important to note however, that results demonstrated in this study could not be generalized to other field strengths as RF heating is a resonance phenomenon highly dependent of the field frequency.

ACKNOWLEDGMENT

This work was supported by the NIH grant R01EB034377.