Synopsis

Keywords: Non-Array RF Coils, Antennas & Waveguides, Non-Array RF Coils, Antennas & Waveguides, Dielectric resonator, loo-dipole

The goal of this work was to demonstrate that performance of multi-channel dielectric resonator antenna arrays for brain MRI at 7T can be substantially improved using a novel multi-feed, loop-dipole coupling mechanism. Simulations were conducted for different rectangular DRA geometries and dielectric constants. Three RF feed types were investigated: loop-only, dipole-only and loop-dipole. 16-channel, loop-dipole rectangular DRA arrays provided significant gains in B₁⁺, SAR efficiency and SNR vs. 8-channel bow-tie antenna array. The feasibility of multi-feed, loop-dipole approach was for a 24-channel DRA array was demonstrated.

Introduction

Dielectric resonator antenna (DRA) is a promising alternative for MRI at 7T which requires no additional decoupling circuits and only a minimal number of lumped components^1-3. Different RF feed mechanisms can be used to induce transverse electric (TE) modes in a rectangular DRA, e.g. loop coils or dipole antennas. While it was shown, that dipole-fed rectangular DRA can provide a good transmit efficiency in deeper located regions, e.g. prostate^4,5, it is unknown whether the same is true for human brain MRI at 7T, mainly due to differences in sample size and the sample-mode interaction⁶. Multi-channel DRA for brain MRI at 7T was developed by Winter et al.⁷ A similar idea was found relevant for MRI at 10.5T⁸. Recently, it was demonstrated for a 16-channel rectangular DRA that by using a combined loop-dipole coupling scheme, a substantial B₁⁺ efficiency gain (35% in the center) can be achieved vs. dipole-only⁹. That study provided a preliminary evidence that a dipole antenna alone might not be the most favorable coupling mechanism for brain MRI at 7T. Therefore, the goal of this study was to: a) determine which type of RF feed (loop-only, dipole-only or loop-dipole) for a rectangular DRA can provide the best performance in human brain MRI at 7T, and b) how these findings can be translated into novel multi-feed, loop-dipole combined DRA arrays.

Methods

Electromagnetic field simulations in a spherical phantom (radius=85mm, ε_r=56, σ=0.66S/m) and human voxel model Duke were conducted using Sim4Life (Zurich Medtech,Switzerland). 12 different, 16-channel rectangular DRA arrays were studied. For a single block: length a=150mm, width b=70mm while height d varied as follows: 0.125b=8.75mm, 0.25b=17.5mm, 0.5b=35mm, 0.75b=52.5mm. For such defined block geometries, a set of ε_r (75-300) was investigated. Loss tangent was kept constant in each simulation. Three types of RF feed were considered: loop-only, dipole-only and loop-dipole (Fig.1). The distance between the block and the phantom was constant for each simulation (10mm). In TX, each array was driven in circularly polarized (CP) mode with a phase increment 45º/element. A referential 8-channel bow-tie antenna array was reproduced from previous report⁷. Finally, multi-feed, loop-dipole combined approach was demonstrated in a 24-channel configuration (3 RF feeds per block: 1 loop element and 2 dipole antennas for ε_r=275 (d/b=0.25) and compared in RX mode with a 16-channel counterpart. For 24-channel array, 2 dipole antenna elements were positioned in a way that each outer arm of the antenna was 10 mm from the edge of the dielectric block. Transmit field efficiency was defined as B₁⁺/√P, where P is the input power, and SAR efficiency defined as B₁⁺/√SAR_10g, where SAR_10g is the maximum SAR averaged over 10g. Signal-to-noise ratio (SNR) was evaluated using an implementation of the Roemer’s algorithm¹⁰ which was based on the S-matrix formalism proposed earlier¹¹.

Results

In all of the cases, the highest B₁⁺ and SAR efficiency in a spherical phantom was observed for the loop-only coupling scheme, while the highest SNR was observed for the loop-dipole coupling scheme (Fig.2). The highest B₁⁺ efficiency (loop-only) in the center was obtained for ε_r=275 (0.79μT/√W): TE_21δ mode. The highest SAR efficiency (loop-only) for: ε_r=75, d/b=0.75 (1.37 μT/√(W/kg)): TE_11δ mode. The highest SNR (loop-dipole) was found for: ε_r=175, d/b=0.75 (1.38a.u.). For Duke, the highest B₁⁺ efficiency in the center of the head was higher for all three of loop-dipole combined arrays (loop-only used for TX): 1.53- (ε_r=150 and 175) and 1.49-fold (ε_r=275) vs. 8-channel bow-tie. The most apparent gain in SAR efficiency vs. bow-tie array was observed for ε_r=150 (1.23-fold). SNR in the center was higher for all three 16-channel loop-dipole combined arrays: 1.78- (ε_r=175), 1.71- (ε_r=275) and 1.63-fold (ε_r=150). Multi-feed, loop-dipole approach was showed for a 24-channel array (ε_r=275; d=0.25b). Scattering parameter matrix for the 24-channel array was evaluated demonstrating acceptable level of coupling (Fig.4). SNR for the 24-channel array in the center of the phantom was negligibly higher vs. 16-channel: 1.32 vs 1.31(a.u.) but with a noticeable SNR increase along profiles in two other planes: XZ and YZ (Fig.5).

Discussion

This work demonstrates for the first time that multi-feed, loop-dipole combined approach can be used to substantially enhance transmit (and receive) performance of rectangular dielectric resonator antennas for human brain MRI at 7T. The data indicate that loop-only coupling scheme should be used in TX mode to achieve the highest B₁⁺ and SAR efficiency, while the loop-dipole should be the most suitable in the receive mode to obtain the highest SNR in spherical samples which are similar to human head in terms of size and electrical properties. Note that in previous studies, DRAs were fed using just a single element (either loop or dipole) per block, and a multi-feed, loop-dipole approach enabled using the total number of 24 channels instead of 8 as reported before. Moreover, for the 24-channel rectangular DRA, inter-element coupling was found to be acceptable and no additional decoupling methods were necessary. Our goal is to develop a head-adjustable, 8-channel TX and 24-channel RX DRA array for brain MRI at 7T which could find its use in multi-modal imaging, e.g. electroencephalography (EEG) combined with functional MRI.

Acknowledgements

The authors acknowledge access to the facilities and expertise of the CIBM Center for Biomedical Imaging, a Swiss research center of excellence founded and supported by Lausanne University Hospital (CHUV), University of Lausanne (UNIL), Ecole polytechnique fédérale de Lausanne (EPFL), University of Geneva (UNIGE) and Geneva University Hospitals (HUG).