Synopsis

Keywords: In Silico, RF Arrays & Systems

The recently developed 60-channel Tic-Tac-Toe transmit coil was analyzed for use in either single or parallel transmit systems at 7T. With the 60-channel TTT coil, excellent homogeneity can be achieved and safe levels of SAR can be provided regardless of which system (sTx or pTx) is used.

Introduction

The Tic-Tac-Toe (TTT) family of transmit coils^1-4 has shown effectiveness at producing homogeneous and load-insensitive B₁⁺ for 7T MRI, a priority for single-channel transmit (sTx) systems. This study investigates the use of TTT coils for parallel transmit (pTx) systems; specifically, we are investigating B₁⁺ and SAR of a 60-channel TTT head coil optimized separately for sTx and pTx systems.

Methods

The 60-channel head coil has 15 TTT panels which each have 4 transmit channels. Simulations of the magnetic and electric fields were performed using in-house finite difference time domain (FDTD) software, which includes transmission line and capacitor models. Three anatomically detailed head models, shown positioned in the coil in Figure 1, were used from the IT’IS Virtual Family⁵: Duke (155lbs male), Ella (126lbs female), and Billie (75lbs female). RF shimming was performed in Matlab in order to optimize the phases and amplitudes of each channel. The ROI for SAR calculations was the whole head, and the ROI for B₁⁺ was the whole head excluding nasal cavities and ears, from the bottom of the cerebellum to the top of the brain.

For the single-channel mode (sTx), the coil was simultaneously optimized over the three head models in order to find a non-subject specific case. Amplitudes were constrained to implementable cases (a reasonable number of standard power splitters) and total power was constrained to 8kW as provided by the scanner in sTx mode. We aimed to raise the minimum B₁⁺ in the cerebellum and temporal lobe while not sacrificing homogeneity by using an objective function of coefficient of variation (CV) over minimum B₁⁺ (CV/min).

To simulate how the coil would perform in a pTx system, the fields were optimized independently for each head model and total power was constrained to 16kW output from the scanner. The cost functions utilized were coefficient of variation (CV) of B₁⁺, maximum B₁⁺ over minimum B₁⁺ (max/min), and peak SAR. Different weights were applied to each of the terms in order to provide varied shim cases. An optimal shim case was chosen for each head model based on peak and average SAR, CV, and visual assessment of homogeneity and symmetry of the B₁⁺ field distribution.

Results

Comparisons of B₁⁺ and SAR in sTx and pTx mode for each of the head models are displayed in Figures 2, 3 and 4. Each is scaled to its maximum value. A summary of relevant statistics is shown in Table 1, including mean, CV, and maximum over minimum (max/min) of the B₁⁺, peak/average SAR, and SAR efficiency (μT/√(W/kg) for each case. Using a 16kW pTx system, large improvements in CV and max/min of the B₁⁺ are seen as compared to the 8kW sTx mode, while, for Duke and Ella, peak/average SAR is lower with 8kW sTx. For all head models, SAR efficiency is similar in both pTx and sTx modes.

Discussion and Conclusion

The parallel transmit system improves the homogeneity of the B₁⁺ provided by the 60-channel TTT array as it allows for a unique shim case for each head and supplies more power. However, B₁⁺ efficiency (B₁⁺ per unit power) is reduced in pTx, but without having to combine the 60 transmit channels into one, less losses due to cables and splitters are expected. Additionally, regardless of sTx or pTx mode, SAR efficiency remains nearly the same. In any case, with the 60-channel TTT coil, excellent homogeneity can be achieved (CV as low as 0.157) and safe levels of SAR can be provided (peak/avg less than 3.3).

Acknowledgements

This work was supported by the National Institutes of Health under award numbers R01MH111265, R01AG063525, R56AG074467, and T32MH119168, and by the National Science Foundation Graduate Research Fellowship under Grant No. 1747452. This work was supported in part by the University of Pittsburgh Center for Research Computing through the resources provided.