Our objective is to design, build, and test an RF head coil for demonstrating feasibility of human head imaging in vivo at 10.5T. To this end, we evaluate a number of RF coil designs, beginning with the dipole array reported here. Dipole elements have been used in a variety of transmit array applications^1-3 and show promise in high field MRI. In this study, we present an end-loaded dipole array for 10.5T head imaging. An array of these elements is compared to the loop array and birdcage coil at 3T, 7T, and 10.5T.

Until IRB and IDE approval for 10.5T human studies is received, we demonstrate imaging feasibility with an approximately human-sized, anesthetized pig under an approved IACUC protocol.

For this experiment, a dipole element composed of a pair of radiating and end-loading segments was designed and simulated⁴ that would be self-resonant at 10.5T (447MHz). Dimensions for all coils were restricted to 16cm length, 25cm ID, and 31cm OD (Figure 1). The end-loading segments (located on the end rings) were parametrized in XFdtd 7.3 (Remcom, USA) to determine their optimal size for self-resonance at 10.5T. The radiating element was constructed from a pair of copper tubes 80mm(L)x6.35mm(OD). The end-loading segments are 20mm wide copper strips subtending an angle of 36° (Figure 2a). This element required minimal impedance matching at 10.5T, and was tuned to 3T and 7T with lumped inductors located between the radiating and end-loading segments (Figure 2b). This dipole design was inspired by the top-loaded monopole antennas used by non-directional beacons for aviation navigation. Top-loading increases the radiation efficiency of the antenna⁵.

For comparison purposes, a shielded 8-rung high-pass quadrature-driven birdcage coil and an 8-loop decoupled array were also constructed from the same materials to the same dimensions. A cylindrical phantom 32cm long and 16.8cm wide was used for testing the coils. The phantom solution (51.1% Sucrose, 47.6% H₂O, 1.3% NaCl) has a ε_r = 49 and σ = 0.6 S/m⁶. The radial distance from coil elements to the phantom was 41mm.

For 3T and 7T, the dipole elements were matched on the bench via a pair of series inductors at the feed point and tuned via a pair of inductors located between the radiating segments and end-loading segments (Figure 2b). At 10.5T, the length of the end-loading segment was trimmed to fine tune the resonant frequency. When resonant at 10.5T, the parasitic inductance and capacitance at the feed point were enough to achieve an impedance match without lumped components.

The dipole array, loop array, and birdcage coil were all tested on a Siemens MAGNETOM Trio (3T) and 7T system. Fixed phases were used at 3T to produce circular polarization (CP). The coils were interfaced to the system via an 8-channel T/R switch box (Virtumed LLC, USA). At 7T and 10.5T the B₁ field was shimmed for B₁⁺ efficiency at the center of the phantom (coinciding with coil center)^7-10. This produced a CP-like B₁ field distribution. Two proton-density gradient-echo images (α=60°, 2α=120°) were collected and B₁⁺ maps were calculated via the double-angle method¹¹. Signal-to-noise ratio (SNR) maps were calculated from the α=60° image and a noise scan¹².

The dipole array was used to acquire the first 10.5T in vivo images of a porcine head (Figure 3). The experimentally obtained B₁⁺ maps at 3T, 7T, and 10.5T are shown in Figure 4 with SNR and B₁⁺ values at coil center plotted across field strength in Figure 5. The transmit efficiency of the dipole array sees marked improvement as field strength increases, but never out performs the comparison coils. The SNR is approximately equal for the dipole and loop arrays at 7T. The high-pass birdcage coil failed to resonate at 10.5T and was excluded. SNR results at 10.5T are currently unavailable.

The dipole array presented is a very simple design because it is self-resonant and does not rely on lumped components. Unfortunately, while the dipole array performs well in terms of SNR, it is inferior to both the loop array and birdcage coil in terms of transmit efficiency. Although the high-pass birdcage coil failed to resonate at 10.5T, a band-pass birdcage coil can be made resonant at higher frequencies and may provide another point of comparison.Feasibility of successful in vivo head imaging has been demonstrated for the first time at 10.5T. The end-loaded dipole array provides a viable alternative to traditional transmit arrays at 10.5T, while remaining inferior in terms of transmit efficiency. Future work will compare the dipole array to the band-pass birdcage coil and other line element arrays.