Characterization of radiotherapy component impact on MR imaging quality for an MRgRT system

Abstract Radiotherapy components of an magnetic resonnace‐guided radiotherapy (MRgRT) system can alter the magnetic fields, causing spatial distortion and image deformation, altering imaging and radiation isocenter coincidence and the accuracy of dose calculations. This work presents a characterization of radiotherapy component impact on MR imaging quality in terms of imaging isocenter variation and spatial integrity changes on a 0.35T MRgRT system, pre‐ and postupgrade of the system. The impact of gantry position, MLC field size, and treatment table power state on imaging isocenter and spatial integrity were investigated. A spatial integrity phantom was used for all tests. Images were acquired for gantry angles 0–330° at 30° increments to assess the impact of gantry position. For MLC and table power state tests all images were acquired at the home gantry position (330°). MLC field sizes ranged from 1.66 to 27.4 cm edge length square fields. Imaging isocenter shift caused by gantry position was reduced from 1.7 mm at gantry 150° preupgrade to 0.9 mm at gantry 120° postupgrade. Maximum spatial integrity errors were 0.5 mm or less pre‐ and postupgrade for all gantry angles, MLC field sizes, and treatment table power states. However, when the treatment table was powered on, there was significant reduction in SNR. This study showed that gantry position can impact imaging isocenter, but spatial integrity errors were not dependent on gantry position, MLC field size, or treatment table power state. Significant isocenter variation, while reduced postupgrade, is cause for further investigation.

treatment planning, inaccurate dose calculation and poor patient alignment. [12][13][14] Many components of the LINAC and beam delivery systems have the potential to further reduce the accuracy of the integrated MRI system. The impact of gantry position on radiation-MR isocenter coincidence has been investigated on MR-60 Co and MR-LINAC systems, and shown variations in the MR isocenter depending on gantry position. 15,16 Utilizing an MLC for beam shaping introduces mobile metallic material, which is often the RT component closest to the magnet, that may also impact the magnetic field. A computational study by Kolling et al investigated MLC impact on field homogeneity including source-to-isocenter distance, field size, magnetic field strength, and other properties and found that MLC field size caused dynamic changes in field distribution. 17 A third component impacting MRI system performance is the treatment couch, specifically the motors used to move the patient in and out of the bore and accurately align them to treatment position. These motors introduce electric current at the head and foot of the MRI bore. In MRgRT system design, attempts to reduce the impact of the gantry and LINAC systems on the MRI have been made. These changes include integrating the cryostat into the Faraday cage, utilizing carbon fiber components for RF absorption, placing electronic components outside of the radiation window, utilizing magnetically shielded rings between the MRI and LINAC, and physically moving the LINAC components away from the MRI. 18,19 Unfortunately, interactions between the MRI system and LINAC system are still present.
The authors present a phantom-based assessment of radiotherapy component impact on the MRI system by tracking changes in imaging spatial integrity and imaging isocenter. Images were acquired pre-and postupgrade to the 2.0 system of an institutional MR-LINAC system with RT system components in various positions.

| MATERIALS AND METHODS
Magnetic resonance imaging isocenter variation was quantified using a phantom-based method. Images were acquired in MRI QA mode for image center-based quantification and assessed spatial integrity changes. Gantry position, MLC field size and treatment table power state were investigated with this image center-based method. All experiments were conducted on the 0.35T ViewRay MRIdian MR-LINAC system (Mountain View, CA) pre-and postupgrade to the 2.0 hardware system and 5.3.0 software system.

2.A | MRIdian system
The MRIdian system utilizes a single energy 6-MV-flattening filter free LINAC on a gantry, with six drums containing electrical and RF components, positioned between the two halves of a 0.35T split bore Siemens MRI scanner. The LINAC system uses a double-stacked doublefocused 138-leaf MLC for radiation field definition. The upgrade consisted of major changes to the LINAC hardware, including the waveguide and MLC system, and system software, but did not make hardware changes to the magnet, however, significant software updates were implemented to improve shimming algorithms. The waveguide position was altered to run further from the interior of the magnet, and the MLC leaf speed was increased from 1.5 to 4.0 cm/s. The altered waveguide position is intended to reduce eddy currents produced during imaging. MLC upgrades included an improved drive mechanism with 303 stainless steel drive screws to increase leaf speed. CAD images of the gantry ring pre-and postupgrade, as well as the upgraded MLC components are shown in Fig. 1.

2.C.3 | Treatment table
For assessing the impact of treatment table power state on imaging isocenter and spatial integrity the image center-based method was used on the postupgrade system. The treatment table is automatically turned off for all image acquisitions in RT mode, requiring the MRI system and TDS to be disconnected for this test so that table power shut off could be prevented.

3.B | MLC impact on 3D MRI
Volumetric 3D images were acquired of the SI phantom with the MRI system disconnected from the TDS. The mean error for F I G . 2. Spatial Integrity Phantom. Fluke 76-907 uniformity and linearity water phantom used for spatial integrity and image centerbased MRI isocenter variation.  Table 1.

3.C | Treatment table impact on 3D MRI
When the table was powered on the mean spatial integrity error increased to 0.48 ± 0.017 mm from 0.42 ± 0.02 mm with the power on. The isocenter shift was 0.03 mm with the table powered on compared to when powered off. In addition to increased spatial integrity mean error, the SNR was significantly reduced when

| DISCUSSION
The change in MRI isocenter at multiple gantry angles, MLC field including studies using an identical phantom to that used in this study. 16,[19][20][21][22] In addition, the phantom used in this study cannot replicate the magnetic susceptibility or imaging volume of patient anatomy. The larger size of a patient and increased magnetic susceptibility could further increase the inhomogeneities of the main magnetic field, gradient fields, and RF pulses.
This work demonstrates that the MRI isocenter remains dependent on gantry angle after a significant vendor upgrade due to the significant effects of the ferromagnetic gantry on the MRI system altering field homogeneity. Isocenter deviation remained consistent in magnitude between the standard and tune-up shimming methods postupgrade indicating that reacquisition of the 3D phase map at each gantry angle was not sufficient to correct for magnetic field inhomogeneities. MLC field size and treatment table power did not significantly impact the imaging isocenter or spatial integrity of the MR system.

| CONCLUSION
MRI isocenter variation and spatial integrity variations were charac- The MRI isocenter variation postupgrade is under 1 mm but warrants continued investigation for sources and improvement.

ACKNOWLEDG MENTS
Benjamin Lewis contributed to data collection, analysis, and wrote the manuscript. Bruce Gu contributed to the design of the data F I G . 5. Image quality with table power. Spatial integrity phantom images acquired with the table off (a), and with the table on (b).
collection and analysis methods. Ryan Klett and Rajiv Lotey contributed to the design of data analysis and analysis tools. Olga Green contributed to conception of the study design and supervision of the project. Taeho Kim contributed to data collection, analysis, analysis tools, and supervision of the project. All authors discussed the results and contributed to the final manuscript.