Introduction

The ultra-high field MRI poses a huge interest for clinical and research studies and demands new instrumentation solutions, including coil design^1,2. The development of a new antenna involves in-vivo tests to verify the construction correctness and thus must ensure patient safety. One of the crucial requirements is a low specific absorption rate preventing body overheating. The demand for global and local SAR restrictions becomes more exigent for higher fields where elevated RF energy and strongly inhomogeneous SAR take place^2,3. Since the SAR distribution depends on coil configuration, in-vivo tests need numerous preliminary SAR simulations and checks on phantoms and human-head models with subsequent consideration by local (IRB) and, in some countries, national ethics commissions. Human studies can reveal the need to redesign the coil and prompt a new verification cycle for refined geometry. Thus, this process is highly resource- and time-consuming; it deprives time for experimentation and remodeling and slows down the coil development stage. To accelerate the design process, the so-called “restricted SAR” (rS) protocol has been proposed⁴. This work extends the previous version enabling the parameter variation and new sequence features. The new rS protocol offers better B₀ mapping coverage, optimal B₁⁺-map estimation, and compatible MR image contrasts at higher spatial resolutions.

Methods

The rS protocol constraints rely on energy conservation and guarantee the respect of the IEC guidelines for global and local (10-gram averaged) SAR values⁵ by limiting the transmitted RF power. Even in a hypothetical worst-case situation where the total input power is absorbed by a single 10-gram piece of biological tissue, local and global SARs remain below regulatory limits. Under these conditions, for head studies, the maximum power supply averaged over 6 minutes becomes 0.1 W. The short-term limit for 10 seconds is 0.2 W. The protocol calculates the energy in the corresponding time interval for given sequence settings and compares it to the limit. The user can change the parameters keeping the output power within the safe range. The calculation of the previous protocol limits assumes an uneducated user, which will possibly apply the maximum available reference voltage. This hypothesis freezes numerous sequence parameters such as TR, flip angle, and interleaved multi-slice mode limiting the sequence capabilities. Now the sequences automatically adjust the range of available parameters depending on the required reference voltage, similar to the usual unrestricted mode.
The extended rS protocol was implemented for three sequence source codes (Fig. 1): gradient recalled echo (GRE), used for localizer, B₀ shimming and T₂*-weighted images, XFL providing B₁⁺-map, and echo-planar imaging (EPI). Sequence testing was performed on a Magnetom 7T MRI scanner (Siemens Healthineers, Erlangen, Germany) with a SC72 whole-body gradient (maximal amplitude 100 mT/m and slewrate 200 T/m/s) using Head Coil 1Tx/32Rx (Nova Medical, Willington, USA) in single-channel receive mode. After phantom verification, we tested the old and new protocols on a healthy volunteer in accordance with local IRB rules. The new rS protocol used the most optimal available parameters for each sequence (Fig. 2). For a more accurate B₁⁺-map analysis with the new protocol, the XFL parameters were tuned to minimize the uncertainty of the targeted flip angle. The noise-induced FA error was estimated for different α/β (see Fig. 1) ratios⁶ considering the SNR dependence on the excitation angle β and the proportionality of the α and β distributions (Fig. 3). Actual Flip angle Imaging (AFI) applied in the non-restricted mode provided the gold-standard B₁⁺-map.

Results

Both protocols were used to obtain a localizer, a B₀-map, and a B₁⁺ distribution map. The FA uncertainty analysis gave the optimal saturation and excitation XFL flip angles of α=60°-80° and β=4°-7°. The resulting FA distributions for XFL were compared with the AFI map to evaluate their precision (Fig. 4). The new protocol version additionally provided EPI and T₂*-weighted images (Fig. 5).

Discussion

The extended rS protocol allows additional experimental options including both interleaved and sequential multi-slice modes for GRE and XFL, fat/water suppression, and magnetization preparation for GRE and EPI. This adjustability allows shorter TA and better image quality. In addition, the same sequence allows for different types of contrasts (e.g., rS_GRE code works for T₂*-weighted imaging and B₀-map). The new version yields B₀-shimming with improved coverage (Fig. 5). The B₁⁺-map comparison shows that the old version provides a biased FA and stronger sensitivity to B₀ inhomogeneities due to the long rectangular saturation pulse used. The current protocol permits different pulse types including SLR, which delivers a more precise B₁⁺-map closer to the AFI results. The appropriate parameters for a more accurate FA distribution analysis allow a better assessment of the coil transmission profile. Moreover, for experiments such as T₂*-weighted imaging, the extended rS protocol enables high single-slice resolution (0.5x0.5x4.0 mm³). This image quality gives reason to think about expanding the protocol applications for other research purposes requiring severe SAR limitations.

Conclusion

The rS protocol ensures the safe use of the RF coil prototype in vivo without complex prior validation, which significantly saves time and resources. The extended protocol version offers more accurate antenna characterization and new experimental opportunities. This simplicity and adjustability push forward the RF coil development at ultra-high field.

Acknowledgements

This study has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement No 952106 (M-ONE project) and from Leducq Foundation (Large Equipement ERPT program, NEUROVASC7T project). We also thank Bruno Pinho Meneses for his idea of an extended rS protocol.