Over the past decades, there has been a continuously increasing interest in MRI of tissues with very short transverse relaxation times T2 or T2* below 1 ms, such as e.g. bone, tendons, ligaments, lung, or myelin. Therefore, 3D radial centre-out sequences have been developed with ultra-short (1, 2) or even zero TE (3-7) which enable capturing short-lived signals efficiently. However, on the downside of increased short-T2 sensitivity also signals from hardware parts are detected, in particular from the RF coils (8, 9).

To avoid related image artefacts for techniques with ultra-short TEs of typically several tens of µs, coil designers and manufacturers often choose materials with particularly short T2 whose signals have decayed before data collection starts. However, with zero-echo-time (ZTE) imaging even these signals are detected, experience broadening, and are aliased if stemming from outside the FOV. This situation is aggravated when imaging short-T2 tissues at large bandwidths where the finite initial RF dead time introduces a considerable spherical gap in central k-space of ZTE data which leads to strong amplification of background signal (10) (Fig. 1). Thus, ZTE imaging may become two to three orders of magnitude more sensitive to background signal than UTE approaches.

The background issue can partly be addressed by overprescribing the FOV at large expense of scan time and unnecessarily high bandwidth. Hence, ideally, the undesired signal is avoided in the first place during coil design and production. This can be achieved relatively easily for simple geometries such as loop coils (9). However, resonator structures providing high uniformity are challenging to construct due to majorly altered considerations regarding materials, mounting, and geometry.

In this work, strategies are proposed to address these difficulties in the dedicated design of a volume RF coil suitable to image human joints at 7T. The first fully ZTE-compatible birdcage coil is presented and shown to be virtually free of proton signal.

Mechanical design: To build up the coil, only materials were used which are generally considered to be proton-free. In particular, for fixation glue was avoided and screw connections (brass) were employed instead. Cylinders made of borosilicate glass (Schott DURAN, Germany) served as formers for birdcage structure and shield, and were assembled with lids milled from “virgin” PTFE (Fig. 2). For birdcage conductors, bars of Cu-ETP were used and brazed to achieve mechanical robustness. The coil was cleaned with alcohol or acetone to remove in particular fat and solder flux.

Electrical design: A 7T shielded eight-rung band-pass quadrature birdcage transmit-receive coil was designed supported by simulations (Sim4Life, ZMT, Switzerland) and implemented with 180 mm diameter and length. The RF shield of diameter 250 mm was made of solid aluminium foil, as eddy currents are negligible in the ZTE sequence with quasi-continuous gradients.

Cables and traps: Cables with PTFE dielectric and FEP sheath (Leoni, Germany) were used. Bazooka cable traps were implemented with a dielectric made from glass tubes which were bent to fit into the coil casing (Fig. 2). The cable sheath was removed in the traps to reduce signal from FEP amplified by high local B1. Residual signal in traps and cables was spoiled by wrapping them loosely-spaced with thin ferromagnetic wire.

ZTE imaging was performed at 250 kHz bandwidth in a large FOV of 600 mm, covering the full gradient range, and a normal FOV of 240 mm adapted to sample and coil volume. Further parameters were: matrix size 128-352, full Nyquist angular encoding, flip angle 1-3°, TR 0.5-0.9 ms.

Fig. 3: Large-FOV images of the empty coil show that residual proton signal is below noise level. Only with high averaging, minor signal is seen close to conductors.

Fig. 4: Images acquired with normal FOV exhibit a very clean background even at an ambitious k-space gap of three Nyquist dwells, which is a strong improvement as compared with Fig. 1.

Fig. 5: In-vivo ZTE images of joints at 3D isotropic sub-millimetre resolution show high SNR, appropriate coverage, and good uniformity.

A virtually proton-free birdcage coil was designed and built, thus enabling short-T2 MRI with zero echo time free of background signal. Importantly, this cannot be achieved by using materials with ever shorter T2s. Very low residual signal was detected close to conductors, stemming probably from moisture in air (8) or glass. Nevertheless, glass was found to be clearly superior to perfluorinated polymers which are prone to contamination with protons (11).