Optimization of RF system for homogenous, consistent, and safe neuro imaging at 7T MRI
Tales Santini1, Sossena Wood1, Tiago Martins1, Nadim Farhat1, Salem Alkhateeb1, Howard J. Aizenstein1, and Tamer S. Ibrahim1

1University of Pittsburgh, Pittsburgh, PA, United States


This work presents two shimming cases for homogenous B1+ at 7T. The Tic-Tac-Toe system was optimized using FDTD simulations and experimentally verified. In the homogenous shim case, the array was capable of delivering a homogeneity of 16.6% (measured) with a SAR efficiency of 1.40μT/√(W/kg) (simulated). In the B1+ efficient shim case, a measured homogeneity of 18.0% and SAR efficiency of 1.55μT/√(W/kg) (simulated) was obtained. The B1+ field was measured over the whole head above and including the cerebellum and excluding the nasal cavities. The coil performance was compared with the TEM coil and experimentally verified with TSE and T2-SPACE sequences.


The Tic-Tac-Toe (TTT) transmit RF coil design has been used for imaging several parts of the body, such as head1 (currently utilized in 11 patient studies at 7T), breast2, foot3, and body. In this work, two RF shimming cases of the 16-channel TTT head coil for 7T is presented.


The TTT RF coil design is composed of eight square-shape transmission lines connected to each other in a Tic-Tac-Toe fashion. Each coil has 4 excitation ports, 4 matching rods, and 4 tuning rods. The TTT head coil is composed of 4 sets of TTT coil positioned around the head, totaling 16 channels (Figure 1a). It creates 4 levels of excitation in Z (static magnetic field) direction (Figure 1d).

The RF behavior of the coil was simulated using an in-house developed FDTD package with an accurate transmission line model. The simulated coil was tuned and matched by variating the lengths of the inner rods of the transmission lines. The parameters used in the simulation were: spatial resolution ~1.59mm, temporal resolution ~3ps, running for 100,000 time-steps (enough to achieve steady state).

The implemented RF shimming cases were obtained using numerical optimizations with the coefficient of variation as the cost function and constraining the mean B1+ over the region of interest (defined as the whole head above and including the cerebellum and excluding the nasal cavities). The amplitude of the channels was chosen according to the Eigenmodes of the RF coil, described in 1. The shimmed cases were implemented in the TTT head coil using power splitters and phase shifters (see Figure 1b and 1d).

An extensive comparison between the 4-channel 16-element TEM coil and the 16-channel TTT array was performed. In this case, quadrature/pseudo quadrature, amplitude-and-phase, and phase-only RF shimming were used.

B1+ maps were acquired in-vivo using the turbo FLASH sequence, with the following parameters: resolution 2.0 × 3.2 × 3.2mm3, TE/TR = 1.09/3000 ms, 6 flip angles. The images were acquired using the TSE product sequence (TE/TR = 61/10060ms, resolution 0.37x0.37x1.5mm3, GRAPPA 2, acquisition time 3:32min, 36 slices) and the SPACE sequence (TE/TR = 367/3400 ms, resolution 0.6 isotropic, GRAPPA 3, acquisition time 8:11 min, 224 slices, and bias corrected using SPM.)


Figure 2 show FDTD simulations and in-vivo acquired B1+ maps, the main statistics values, and the profiles from central slices in the simulations and acquisitions for the two RF shimming cases. Figure 3 compares the 16-channel Tic-Tac-Toe design with the 4-channel 16-element TEM design for 5 different head shapes derived from the Duke model. Figure 4 shows coronal slices of the in-vivo TSE sequence acquisition, demonstrating the B1+ homogeneity of the RF array in regions that are challenging at 7T MRI, such as the cerebellum. Figure 4 also shows the details of the hippocampal structure in the zoomed image. Figure 5 shows sagittal slices of the T2-SPACE sequence (animated GIF format) acquired with the 16-channel TTT RF system.

Discussion and conclusion

The TTT RF array presents excellent homogeneity at 7T, demonstrated in the B1+ maps, T2 weighted image acquisitions. As the coil was tested on the single transmit mode, the differences between the measured and simulated B1+ power efficiency is due to the RF splitters (4 levels of 2- and 4-way splitters)/plugs/cables/coils losses which amounted for 40-45% RF power loss. Figure 2 shows the flexibility of the design with the tradeoff between homogeneity and efficiency of the array (both experimentally and using RF simulations). Figure 3 also shows the TTT design is subject insensitive, with small variations of the coil performance with different head shapes/volumes when compared with the TEM coil. While the TTT design presents lower B1+ power efficiency when compared to quasi static-based RF coil designs, it achieves higher SAR efficiency.


This work was supported by NIH R01MH111265 and by the CAPES Foundation, Ministry of Education of Brazil, 13385/13-5.


[1] Santini, T.,et al. (2018). “In-vivo and numerical analysis of the eigenmodes produced by a multi-level Tic-Tac-Toe head transmit array for 7 Tesla MRI.” In press. PloS one.

[2] Kim, J., et al. (2016). "Experimental and numerical analysis of B1+ field and SAR with a new transmit array design for 7T breast MRI." J Magn Reson 269: 55-64.

[3] Santini, T. et al. (2018). “A new RF transmit coil for foot and ankle imaging at 7T MRI”. Magn Reson Imaging 45: 1-5.


Figure 1: The 16-channel Tic-Tac-Toe RF system. In a), the FDTD model of the transmit array with the Duke model as the load; In b), the power splitter configuration to drive the RF array; In c), the assembled RF system; In d), the excitation points and Z-levels of the transmit array.

Figure 2: RF shimming optimization cases for the 16-channel transmit array. The array presents high levels of homogeneity and low levels of SAR for a 7T system.

Figure 3: comparison between the 16-channel transmit Tic-Tac-Toe array and the 4-channel 16-element TEM coil for 5 head sizes derived from the Duke model.

Figure 4: 0.375x0.375x1.5mm resolution 2D TSE acquisition at 7T MRI using the Tic-Tac-Toe RF system. The TE/TR = 61/10060ms, acceleration factor 2, and acquisition time 03:32min.

Figure 5: Animation showing the T2 SPACE sequence acquired with the 16-channel TTT system at resolution of 0.6 mm isotropic, TE/TR = 367/3400 ms, acceleration factor 3, and acquisition time = 8:11 min.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)