Purpose:

To improve the transmit (Tx) performance and brain coverage of a human head array coil at 9.4 T (400 MHz), a novel dipole antenna array design was developed, constructed, and evaluated.

Introduction:

Clinical application of ultra-high field (UHF, >7 T) MRI, is still hindered due to technical hurdles associated with imaging of large (comparable to the RF wave length) human objects (body, head). Resulting issues include a significant increase of the maximum local SAR, and a strong inhomogeneity of the Tx RF magnetic field, B₁⁺ characterized by a pattern with strongly decreased peripheral values (~2 times for a head (1) at UHF). Both effects are associated with a so-called “dielectric resonance” or standing wave. Importantly, this effect is seen along all three directions and considerably limits whole-brain coverage at UHF. Improvement of the B₁⁺ homogeneity and coverage can be achieved by using multi-loop Tx-arrays, e.g. single-row 8-element (1x8) arrays (2,3), or double-row 16-element (2x8) arrays (2,4-6). However, an adequate whole-brain coverage was demonstrated only by using multi-row arrays (e.g.2x8) capable of 3D RF shimming (7). In this work, we developed a novel human head 9.4 T transceiver (TxRx) 1x8 array consisting of bent folded dipole antennas, which substantially simplifies the Tx-coil design but provides a whole-brain coverage that is better than that of a more complex 2x8 loop array.

Methods:

Using electromagnetic (EM) simulations we demonstrate that in the presence of the RF shield, the bent dipole array not only produces its own circularly polarized (CP) mode but also couples to the intrinsic transverse electric (TE) mode (“dielectric resonance”) of the sample (Fig.1), i.e. a phantom or a human head. The TE mode resonates near 400 MHz and is characteristic of tangential electric field, E_t, near the center of the sample (Fig.1). We show that the mode mixing, firstly, improves the longitudinal coverage of the Tx-array. Secondly, it helps to reduce the maximum local SAR. This, however, can also decrease the array Tx-performance since the TE mode also possesses H_z component of the magnetic field. Thus, mode mixing has to be optimized by adapting the dipole geometry. EM simulations of the local SAR and B₁⁺ were performed using CST Studio Suite 2017 (CST, Darmstadt, Germany) and the time-domain solver based on the finite-integration technique. Two models were used, i.e. an elliptical phantom (ε=58.6, σ=0.64 S/m), and the virtual family model “Duke”. After numerical optimization, we constructed and tested the dipole TxRx-array consisting of eight 30-mm bent folded dipoles (30 mm – the folded portion length). Array measured 200 mm x 230 mm in clearance, and 175 mm in length. We compared the performance of the new array to that of 1x8 (3) and 2x8 (6) surface loop TxRx-arrays with the same clearance. While the 2x8 array had the same length, the 1x8 array was shorter (100 mm). All arrays were shielded (distance to the RF shield - 40 mm). All data were acquired on a Siemens Magnetom 9.4 T human imaging system. Experimental B₁⁺ maps were obtained as previously described (8).

Results and Discussion:

Fig.2 shows the effect of mode mixing on the cylindrical phantom in the presence of an RF shield. While straight dipoles (Fig.2C) reveal only a minor quantitative difference for shielded and unshielded arrays, bent dipoles (Fig.2D) demonstrate a large qualitative and quantitative difference in both B₁⁺ and SAR distributions. Firstly, shielded bent dipoles extend the B₁⁺ map longitudinally. Secondly, appearance of the non-zero SAR at the center of the phantom indicates coupling to the TE mode. Fig.3 shows the effect of bending and folding dipoles on the Duke voxel model. Clearly, mixing of modes growths with increasing the folded portion length, which is indicated by the stronger H_z component (Fig.3B) and SAR increase near the brain center (Fig.3C). Mode mixing also improves the brain coverage (Fig.3D). Thus, changing the folded portion length provides an easy way of optimizing the Tx-performance. Table 1 summarizes results of numerical optimization. The 30-mm bent folded dipoles provide the best <B₁⁺>/√SAR value. Finally, Fig.4 shows an experimental comparison of the new dipole array performance to that of the 1x8, and 2x8 surface loop arrays. As seen from the figure, the dipole array provides better coverage than both loop arrays.

Conclusion:

We developed a novel 9.4T human head TxRx-array consisted of 8 optimized 30-mm bent folded dipole antennas. Due to the asymmetrical shape of dipoles (bending), the array couples to the TE mode of the human head. Adjustment of the folded length provides for a simple way of reducing the maximum local SAR and improving the longitudinal coverage.

Acknowledgements

SG acknowledges a support by Government of Russian Federation (Grant 08-08) and Ministry of Education and Science of the Russian Federation (No. 3.2465.2017/4.6). Funding by the European Union (ERC Starting Grant, SYNAPLAST MR, Grant Number: 679927) is gratefully acknowledged by AN and RL.

References

1) Vaughan JT, Garwood M, Collins CM et al. Magn Reson Med 2001;46:24-30. 2) Avdievich NI, Giapitzakis IA, Pfrommer A et al. Magn Reson Med 2018;79(2):1200–1211. 3) Avdievich NI. Appl Magn Reson 2011;41(2):483-506. 3) Adriany G, Van de Moortele P-F, Ritter J et al. Magn Reson Med 2008;59:590-597. 4) Gilbert KM, Belliveau J-G, Curtis AT et al. Magn Reson Med 2012;67:1487–1496. 5) Shajan G, Kozlov M, Hoffmann J et al. Magn Reson Med 2014;71:870–879. 6) Avdievich NI, Giapitzakis IA, Pfrommer A et al. NMR in BioMed 2018; 31(9):1-13. 7) Hoffmann J, Shajan G, Scheffler K et al. Magn Reson Mater Phy 2014;27(5):373-386. 8) Yarnykh VL. Magn Reson Med 2007;57:192-200.