A Pfrommer

Evaluation of transmit efficiency and SAR for a tight fit transceiver human head phased array at 9.4 T

Ultra‐high field (UHF, ≥7 T) tight fit transceiver phased arrays improve transmit (Tx) efficiency (B1+/√P) in comparison with Tx‐only arrays, which are usually larger to fit receive (Rx)‐only arrays inside. One of the major problems limiting applications of tight fit arrays at UHFs is the anticipated increase of local tissue heating, which is commonly evaluated by the local specific absorption rate (SAR). To investigate the tradeoff between Tx efficiency and SAR when a tight fit UHF human head transceiver phased array is used instead of a Tx‐only/Rx‐only RF system, a single‐row eight‐element prototype of a 400 MHz transceiver head phased array was constructed. The Tx efficiency and SAR of the array were evaluated and compared with that of a larger Tx‐only array, which could also be used in combination with an 18‐channel Rx‐only array. Data were acquired on the Siemens Magnetom whole body 9.4 T human MRI system.

Combination of surface and "vertical" loop elements improves receive performance of a human head transceiver array at 9.4 T

Ultra‐high‐field (UHF, ≥7 T) human magnetic resonance imaging (MRI) provides undisputed advantages over low‐field MRI (≤3 T), but its development remains challenging because of numerous technical issues, including the low efficiency of transmit (Tx) radiofrequency (RF) coils caused by the increase in tissue power deposition with frequency. Tight‐fit human head transceiver (TxRx) arrays improve Tx efficiency in comparison with Tx‐only arrays, which are larger in order to fit multi‐channel receive (Rx)‐only arrays inside. A drawback of the TxRx design is that the number of elements in an array is limited by the number of available high‐power RF Tx channels (commonly 8 or 16), which is not sufficient for optimal Rx performance. In this work, as a proof of concept, we developed a method for increasing the number of Rx elements in a human head TxRx surface loop array without the need to move the loops away from a sample, which compromises the array Tx performance. We designed and constructed a prototype 16‐channel tight‐fit array, which consists of eight TxRx surface loops placed on a cylindrical holder circumscribing a head, and eight Rx‐only vertical loops positioned along the central axis (parallel to the magnetic field B0) of each TxRx loop, perpendicular to its surface. We demonstrated both experimentally and numerically that the addition of the vertical loops has no measurable effect on the Tx efficiency of the array. An increase in the maximum local specific absorption rate (SAR), evaluated using two human head voxel models (Duke and Ella), measured 3.4% or less. At the same time, the 16‐element array provided 30% improvement of central signal‐to‐noise ratio (SNR) in vivo relative to a surface loop eight‐element array. The novel array design also demonstrated an improvement in the parallel Rx performance in the transversal plane. Thus, using this method, both the Rx and Tx performance of the human head array can be optimized simultaneously.

Analytical modeling provides new insight into complex mutual coupling between surface loops at ultrahigh fields

Ultrahigh‐field (UHF) (≥7 T) transmit (Tx) human head surface loop phased arrays improve both the Tx efficiency (B1+/√P) and homogeneity in comparison with single‐channel quadrature Tx volume coils. For multi‐channel arrays, decoupling becomes one of the major problems during the design process. Further insight into the coupling between array elements and its dependence on various factors can facilitate array development. The evaluation of the entire impedance matrix Z for an array loaded with a realistic voxel model or phantom is a time‐consuming procedure when performed using electromagnetic (EM) solvers. This motivates the development of an analytical model, which could provide a quick assessment of the Z‐matrix. In this work, an analytical model based on dyadic Greens functions was developed and validated using an EM solver and bench measurements. The model evaluates the complex coupling, including both the electric (mutual resistance) and magnetic (mutual inductance) coupling. Validation demonstrated that the model does well to describe the coupling at lower fields (≤3 T). At UHFs, the model also performs well for a practical case of low magnetic coupling. Based on the modeling, the geometry of a 400‐MHz, two‐loop transceiver array was optimized, such that, by simply overlapping the loops, both the mutual inductance and the mutual resistance were compensated at the same time. As a result, excellent decoupling (below −40 dB) was obtained without any additional decoupling circuits. An overlapped array prototype was compared (signal‐to‐noise ratio, Tx efficiency) favorably to a gapped array, a geometry which has been utilized previously in designs of UHF Tx arrays.

On the Contribution of Curl‐Free Current Patterns to the Ultimate Intrinsic Signal‐to‐Noise Ratio at Ultra‐High Field Strength

The ultimate intrinsic signal‐to‐noise ratio (SNR) is a coil independent performance measure to compare different receive coil designs. To evaluate this benchmark in a sample, a complete electromagnetic basis set is required. The basis set can be obtained by curl‐free and divergence‐free surface current distributions, which excite linearly independent solutions to Maxwells equations. In this work, we quantitatively investigate the contribution of curl‐free current patterns to the ultimate intrinsic SNR in a spherical head‐sized model at 9.4 T. Therefore, we compare the ultimate intrinsic SNR obtained with having only curl‐free or divergence‐free current patterns, with the ultimate intrinsic SNR obtained from a combination of curl‐free and divergence‐free current patterns. The influence of parallel imaging is studied for various acceleration factors. Moreover results for different field strengths (1.5 T up to 11.7 T) are presented at specific voxel positions and acceleration factors. The full‐wave electromagnetic problem is analytically solved using dyadic Greens functions. We show, that at ultra‐high field strength (B0⩾7T) a combination of curl‐free and divergence‐free current patterns is required to achieve the best possible SNR at any position in a spherical head‐sized model. On 1.5‐ and 3T platforms, divergence‐free current patterns are sufficient to cover more than 90% of the ultimate intrinsic SNR.

Decoupling of a tight-fit transceiver phased array for human brain imaging at 9.4T: Loop overlapping rediscovered

To improve the decoupling of a transceiver human head phased array at ultra‐high fields (UHF, ≥ 7T) and to optimize its transmit (Tx) and receive (Rx) performance, a single‐row eight‐element (1 × 8) tight‐fit transceiver overlapped loop array was developed and constructed. Overlapping the loops increases the RF field penetration depth but can compromise decoupling by generating substantial mutual resistance.

The ultimate intrinsic signal-to-noise ratio of loop- and dipole-like current patterns in a realistic human head model

The ultimate intrinsic signal‐to‐noise ratio (UISNR) represents an upper bound for the achievable SNR of any receive coil. To reach this threshold a complete basis set of equivalent surface currents is required. This study systematically investigated to what extent either loop‐ or dipole‐like current patterns are able to reach the UISNR threshold in a realistic human head model between 1.5 T and 11.7 T. Based on this analysis, we derived guidelines for coil designers to choose the best array element at a given field strength. Moreover, we present ideal current patterns yielding the UISNR in a realistic body model.

Decoupling of a double-row 16-element tight-fit transceiver phased array for human whole-brain imaging at 9.4 T

One of the major challenges in constructing multi‐channel and multi‐row transmit (Tx) or transceiver (TxRx) arrays is the decoupling of the arrays loop elements. Overlapping of the surface loops allows the decoupling of adjacent elements and also helps to improve the radiofrequency field profile by increasing the penetration depth and eliminating voids between the loops. This also simplifies the design by reducing the number of decoupling circuits. At the same time, overlapping may compromise decoupling by generating high resistive (electric) coupling near the overlap, which cannot be compensated for by common decoupling techniques. Previously, based on analytical modeling, we demonstrated that electric coupling has strong frequency and loading dependence, and, at 9.4 T, both the magnetic and electric coupling between two heavily loaded loops can be compensated at the same time simply by overlapping the loops. As a result, excellent decoupling was obtained between adjacent loops of an eight‐loop single‐row (1 × 8) human head tight‐fit TxRx array. In this work, we designed and constructed a 9.4‐T (400‐MHz) 16‐loop double‐row (2 × 8) overlapped TxRx head array based on the results of the analytical and numerical electromagnetic modeling. We demonstrated that, simply by the optimal overlap of array loops, a very good decoupling can be obtained without additional decoupling strategies. The constructed TxRx array provides whole‐brain coverage and approximately 1.5 times greater Tx efficiency relative to a transmit‐only/receive‐only (ToRo) array, which consists of a larger Tx‐only array and a nested tight‐fit 31‐loop receive (Rx)‐only array. At the same time, the ToRo array provides greater peripheral signal‐to‐noise ratio (SNR) and better Rx parallel performance in the head–feet direction. Overall, our work provides a recipe for a simple, robust and very Tx‐efficient design suitable for parallel transmission and whole‐brain imaging at ultra‐high fields.

On the superlinear increase of the ultimate intrinsic signal-to-noise ratio with regard to main magnetic field strength in a spherical sample

In this study, the increase of the ultimate intrinsic signal-to-noise ratio (UISNR) with regard to main magnetic field strength B0 is investigated. A simplified spherical phantom of human head size is used. In the center of the sphere, the UISNR grows more than quadratically. Within the volume, in which the distance to the center is smaller than 85% of the spheres radius, the UISNR increases superlinearly. At the surface, the UISNR grows only sublinearly. The SNR of curl-free current patterns grows more than cubically in the center, whereas the SNR of divergence-free current patterns increases quadratically. However, this does not imply, that curl-free modes result in higher SNR than divergence-free modes.

Optimal arrangement of finite element loop arrays for parallel magnetic resonance imaging in the human head at 400 MHz

Parallel magnetic resonance imaging is limited by the signal-to-noise ratio (SNR) of the MR-signal detected by an antenna array. To fully exploit the SNR of circular surface loops surrounding a spherical head phantom, we developed an optimization routine to minimize the arrays noise enhancement. We optimized the position of each element and a common loop radius. As a result we show optimal configurations for 8, 16 and 32 array elements at 400 MHz with different acceleration factors. The importance of proper element alignment with regard to the acceleration direction(s) is shown by a comparison between optimal and poor positioning.