Publikationsrepositorium - Helmholtz-Zentrum Dresden-Rossendorf

Purpose: The physical integration of MRI with proton therapy (PT) into an MR-integrated PT (MRiPT) system is expected to improve the targeting accuracy of PT. The purpose of this project was to develop a prototype system combining a low-field in-beam MRI scanner with a proton pencil beam scanning (PBS) beamline to enable a first MRiPT treatment. This contribution presents first results of the installation and commissioning of the MRiPT system where the positioning reproducibility, magnet shimming performance and image quality with a focus on geometric fidelity were analyzed.

Methods: The MRiPT setup consists of an open C-shaped 0.32 T MRI scanner (MRJ3300, ASG Superconductors SpA, Genoa, Italy) positioned in close proximity of the nozzle of a horizontal proton PBS beamline (Figure 1). The MRI scanner was encased in a custom-designed compact aluminum Faraday cabin. At the location of the beam exit window of the nozzle, a beam entrance opening was incorporated in the wall of the RF cabin, which was sealed by a thin (30 µm) aluminum foil to combine high RF attenuation and small lateral spreading of the traversing proton beam. The scanner and RF cabin were mounted on top of an air-cushion-based transport platform, allowing the assembly to be accurately positioned in the beam path exiting the nozzle. The maneuvering of the assembly into treatment position was thereby visually guided based on room lasers that intersect at the beam isocenter and project onto the outer wall of the cabin. The magnet was shimmed in treatment position close to ferromagnetic components of the nozzle where the B0 field homogeneity was measured using a magnetic field camera (MFC3045, Metrolab Technology SA, Geneva, Switzerland). During commissioning the MR image quality was assessed using the ACR Small MRI Phantom (American College of Radiology, Virginia, USA) with T1w spin echo (SE) imaging, and the CIRS MRI-LINAC Dynamic Phantom (Computerized Imaging Reference Systems Inc., Norfolk, USA) with a T1w 3D spoiled gradient echo (GFE) pulse sequence dedicated for patient positioning control in the MRiPT workflow. The phantom allows for distortion analysis with 4 grid layers consisting of 5 concentric circles.

Results: The positioning accuracy and precision of the mobile in-beam MRI system were below 1 mm. A peak-to-peak B0 field homogeneity of 43 ppm over a 25 cm diameter spherical volume (DSV) around the MR magnetic isocenter was achieved during shimming. The ACR QA protocol revealed a signal-to-noise ratio (SNR) of >80 and a geometric distortion of <1.5 mm over a 10 cm DSV around the magnetic isocenter. The CIRS distortion measurement using the 3D GFE sequence showed that geometric distortion increased up to <8 mm at 20 cm DSV and <11 mm at 24 cm DSV.

Conclusion: A 0.32 T in-beam MRI scanner was successfully installed and commissioned in front of a horizontal proton PBS beamline in preparation for the development of a first prototype MRiPT system. Large volume image distortion measurements showed the necessity for geometric distortion correction algorithms in order to facilitate an accurate image guidance in a future MRiPT treatment.