Introduction

As recent studies indicate, MR-guided radiotherapy (MRgRT) harbors great potential as an alternative to conventional CT-guided radiotherapy¹. Alongside better soft tissue contrast, NMR provides an imaging method without ionizing radiation. Imaging between irradiation fractions or real-time gating could permit modifications to the treatment plan and lead to adaptive radiotherapy. This concept can also be applied to particle therapy, which has the potential to further reduce the dose exposure to the patient². Therefore a hybrid device comprising a treatment table in an MR scanner with access of the ion beam to the patient would be favorable. However, this configuration puts strict conditions on the MR system and its components. The ion beam must not be affected by the RF coil, since scattering or unknown attenuation may impair the treatment. Ideally, the patient table provides access from several angles to be more flexible in the treatment planning. Even for static ion beams, this could be realized with a rotatable patient capsule. Moreover, uniform and homogeneous RF transmit field characteristics over a large FOV are necessary for reliable tissue classification. In this work, an RF body coil with minimal water equivalent thickness (WET) was designed and constructed for a clinical MR scanner at 1.5T.

Methods

Coil: A single channel Tx/Rx 16-leg high-pass birdcage configuration^3,4 with total length of 520mm was chosen to achieve high RF field homogeneity. The coil is attached to the inner surface of an acrylic glass cylinder with an inner diameter of 530mm and consists of copper with a thickness of 35µm covered between two sheets of polyimide and multiple layers of acrylic glue. While the supporting components of the coil and the capsule can be built homogeneously and corrected for in the treatment plan, sharp edges near the conductor or at electronic components pose a bigger obstacle. For that reason, all conductors and other electronic devices (quadrature hybrid, Tx/Rx switch, cables,..) are located at the end rings. The WET of the conductor was calculated for protons and ¹²C⁶⁺ ions in the clinical energy range of (48-221)MeV and (88-430)MeV/u respectively via the Bethe-Bloch formula⁵ to estimate the beam interaction with the coil. MR scanner: All measurements were performed on a 1.5T whole-body MR system (Sola, Siemens Healthcare, Erlangen, Germany). B₁⁺ mapping was performed using the double-angle method⁶ with the following sequence parameters: 2 acquisitions with GRE sequence: TE=10ms, TR=10s, in-plane resolution=(3.75mm)², slice thickness=5mm, RF spoiling; 1st pulse voltage=150V; 2nd pulse voltage=300V. For comparison, an MR image was acquired with a pulse voltage of 200V with equal parameters except for a resolution of (3.2mm)²×(5mm). Additionally, the coil performance was examined with the ellipsoidal phantom rotated 30° within the capsule as detuning of the body coil could deteriorate its imaging capabilities. Phantom: A phantom of size 500×350×200mm³ is filled with a mixture of water, sodium chloride and polyvinylpyrrolidone (PVP) to imitate the mean permittivity (ε_r=47) and mean conductivity (σ=0.42$$$\frac{S}{m}$$$) of the human torso at 64MHz; these values were extracted from a human voxel phantom (Gustav) in simulation. Simulation: Crucial parts of the setup (see Fig. 1) were implemented in a simplified model to perform electromagnetic field simulations in CST Studio Suite 2020⁷ from 0 to 150MHz. Insulating materials were specified as acrylic glass (ε_r=2.8, tan(ẟ)=0.02 (1MHz)).The mesh for the finite element calculation covers the structure with approximately 20 million cells, including an enhanced resolution of 2mm in the vicinity of the RF coil. Imaging: The relationship between MR signal and relative RF field distributions is as follows: $$$S(\vec{x})\sim sin(α(\vec{x}))×B_{1,~rel}^{-}(\vec{x})$$$ with flip angle $$$α(\vec{x})\sim B_{1,~rel}^{+}(\vec{x})$$$. Given the transmit field and the MR signal for a specific flip angle, the receive field distribution (see Fig. 2) or vice versa the MR signal was calculated (see Fig. 3).

Results

For both protons and ¹²C⁶⁺ ions, the WET of the conductor in the RF coil is less than 215µm in the applied energy range and thus is deemed negligible for treatment planning. In Fig. 2 the transmit and receive fields are illustrated for a transversal slice of the phantom in the magnet center. Good field homogeneity was achieved for 0° and 30° rotation of the phantom with slight field asymmetries, especially at the edges of the phantom. However, the RF fields from measurement and simulation show similar distributions. The coefficient of variation (COV) values in Fig. 4 quantitatively confirm the uniformity of the B₁⁺ and B₁^- fields and the agreement between measurement and simulation. Fig. 3 shows MR images of the homogeneous phantom and the similarity between directly measured, scaled, and purely simulated data.

Discussion & Conclusion

The built RF coil provides a concept for MRgRT with particles while maintaining good imaging characteristics at 64MHz with little asymmetries comparable to those reported in literature^8,9. As the coil is capable of both transmitting and receiving MR signals, additional receive coils can be omitted, facilitating particle therapy by avoiding the effect of additional hardware on the ion beam. To achieve higher flexibility in therapy, a large range of rotation angles would be beneficial. Therefore, further investigation of the angular dependence of the RF field homogeneity will be performed in future experiments with the presented setup.

Acknowledgements

This work received financial support from the German Federal Ministry of Education and Research (BMBF, ARTEMIS project WP8, funding reference 13GW0436B)