Measurement validation of treatment planning for a MR‐Linac

Abstract Purpose The magnetic field can cause a nonnegligible dosimetric effect in an MR‐Linac system. This effect should be accurately accounted for by the beam models in treatment planning systems (TPS). The purpose of the study was to verify the beam model and the entire treatment planning and delivery process for a 1.5 T MR‐Linac based on comprehensive dosimetric measurements and end‐to‐end tests. Material and methods Dosimetry measurements and end‐to‐end tests were performed on a preclinical MR‐Linac (Elekta AB) using a multitude of detectors and were compared to the corresponding beam model calculations from the TPS for the MR‐Linac. Measurement devices included ion chambers (IC), diamond detector, radiochromic film, and MR‐compatible ion chamber array and diode array. The dose in inhomogeneous phantom was also verified. The end‐to‐end tests include the generation, delivery, and comparison of 3D and IMRT plan with measurement. Results For the depth dose measurements with Farmer IC, micro IC and diamond detector, the absolute difference between most measurement points and beam model calculation beyond the buildup region were <1%, at most 2% for a few measurement points. For the beam profile measurements, the absolute differences were no more than 1% outside the penumbra region and no more than 2.5% inside the penumbra region. Results of end‐to‐end tests demonstrated that three 3D static plans with single 5 × 10 cm2 fields (at gantry angle 0°, 90° and 270°) and two IMRT plans successfully passed gamma analysis with clinical criteria. The dose difference in the inhomogeneous phantom between the calculation and measurement was within 1.0%. Conclusions Both relative and absolute dosimetry measurements agreed well with the TPS calculation, indicating that the beam model for MR‐Linac properly accounts for the magnetic field effect. The end‐to‐end tests verified the entire treatment planning process.


| INTRODUCTION
Integrated MRI guided radiation therapy (MRgRT) systems are emerging radiation therapy (RT) techniques that combine a MR-scanner with either a linear accelerator (MR-Linac) [1][2][3] or Co-60 teletherapy system. 4 The main magnetic field of the MRI is oriented either longitudinal or transverse to the central axis of the radiation beam.
Owing to MRI's capability to provide excellent soft-tissue contrast images and biological/functional information, these systems are expected to enable adaptive RT and provide real-time, high quality image guidance during delivery with the potential to significantly improve RT outcomes. However, the presence of the magnetic field The dosimetric effect of a transverse magnetic field (TMF) in phantom and patients has been studied extensively. [5][6][7][8][9][10][11][12][13][14][15][16] Compared to conventional external beam RT without magnetic field, the magnetic field can generally lead to an altered buildup depth, and a larger, slightly shifted (asymmetric) penumbra arising from tilting of the dose kernel. 5 The dose deposited at tissue-air interfaces can increase due to the electron returning effect (ERE), 6 which can also be affected by the surface orientation where the photon beam enters or exits. 8 The dosimetric response of QA devices [17][18][19][20][21][22][23][24][25][26][27] (especially the detectors used for beam calibration) in magnetic fields have also been studied, focusing mainly on the absolute dose response of the detectors. The effect of the magnetic field on the detectors is demonstrated 26 not only through the trajectory path deflection of the secondary electrons in the medium surrounding and inside the volume of the gas-filled or solid detector, but also through the change of the intrinsic properties of detectors such as charge carrier versus lattice defect recombination, polarity effect, etc. For the ionization chamber, however, when the field size is sufficiently larger than the detector's diameter, an additional magnetic field and beam quality correction factor 18 can be applied to the original calibration coefficient, obtained without the magnetic field, to obtain the correct absolute dose calibration in the presence of the magnetic field.
The effective point of measurement (EPOM) of the ionization chamber 27 is shifted not only in the beam direction but also laterally, perpendicular to both the beam and the magnetic field direction.
According to O'Brien et al. 27 , the measured lateral shift in the dose distribution was independent of depth and field size from 2 × 2 cm 2 to 10 × 10 cm 2 . The depth of maximum dose had little dependence on field size in the presence of the magnetic field. The output factors measured at the point of the peak intensity in the cross plane profile are more consistent than those measured at the central axis (CAX).
The MR-Linac system developed by Elekta AB (Stockholm, Sweden) in cooperation with Philips Healthcare (Best, Netherlands) consists of an Elekta linear accelerator with a nominal 7MV flattening filter-free photon beam (160-leaf MLC oriented in fixed superior-inferior direction), and a Philips 1.5 T integrated wide-bore MRI scanner. With this system, the magnetic field is oriented transverse to the irradiation field. To match the MR-Linac, the treatment planning system (TPS) must incorporate the effects of magnetic fields. A prototype TPS (Monaco research version v5.0.19.03, provided by Elekta AB) developed for the MR-Linac employs a graphics processing unit (GPU)-based Monte Carlo dose calculation algorithm 28 and considers a uniform magnetic field with strength and orientation as input. 29 The main purposes of this study were to: (a) validate the beam models in the TPS by comparing the model predictions with comprehensive dosimetric measurement data from the MR-Linac and (b) verify the performance of the entire planning and delivery process by carrying out a series of end-to-end tests on the MR-Linac.

| MATERIALS AND METHODS
All the measurement data were collected on a preclinical MR-Linac   The field size is defined at the isocenter.

2.B | Radiochromic film measurements
Radiochromic film (GAFchromic EBT-3) was used as a secondary validation of measurements obtained from other detectors, particularly for small field sizes, for example, in the range of 1 × 1 to 5 × 5 cm 2 .
The EBT3 film was sandwiched between slabs of solid water and positioned on the couch so that the film was oriented coronally at depth 1.5 cm with SSD 131.2 cm. The cross-plane beam profiles were extracted for field size 1 × 1, 2 × 2, 3 × 3, and 5 × 5 cm 2 using a film processing software (Radiological Imaging Technology, Inc).
T A B L E 1 Summary of the field size (FS) and depth for percent depth dose (PDD) measurements with three different types of detectors (Farmer ion chambers (IC), Diamond detector and Micro IC).

PDD
The k Qmsr B was chosen with reference to

2.E | Data calculated from beam models
The dosimetric data in the same conditions as in the measurements described above were generated using the prototype Monaco TPS with the existing MR-Linac beam model (7.0FFF + cryostat) using a calculation grid of 3 mm × 3 mm × 3 mm. The data were extracted from 3D plans of single beams irradiating onto a rectangular water phantom similar to the water tank. The PDDs along the CAX and the beam profiles for each field size at corresponding depths were extracted from the corresponding 3D dose distributions.

2.F | End-to-end tests with an MR compatible 3D diode array
For the end-to-end tests, three single-beam 3D plans based on CT data of an MR-compatible cylindrical 3D diode array (ArcCheck,

2.G | Dose verification in inhomogeneous phantom
To absolute difference for most measurement points was < 1% (range ±2%). In general, the measurements with the diamond detector lead to smaller deviations from those by the beam models as compared to those by the micro IC. The depth of maximum (i.e., d max ) decreases with decreasing field size, from approximately 1.4 cm for field size 10 × 10 cm 2 to 1.0 cm for field size 1 × 1 cm 2 , as shown in both measured and calculated data in Table 3.

3.B | Cross-plane and in-plane beam profiles
Since the magnetic field affects dosimetry mostly in transverse planes, the majority of the data were collected in the cross-plane direction. The large field cross-plane beam profiles (≥5 × 5 cm 2 ) were measured with the Farmer IC and IC profiler TM . Those for small field sizes, for example, 1 × 1, 2 × 2, 3 × 3, and 5 × 5 cm 2 , were measured with diamond detector, micro IC, and radiochromic film.  Table 4. The in-plane beam profiles for field sizes: 1 × 1, 2 × 2, 3 × 3, 5 × 5, and 10 × 10 cm 2 were measured with the micro IC. These in-plane measurements at depth 1.5 cm along with the cross-plane profiles measured with the same chamber are shown in Fig. 4. calculated with the beam model (see Table 4). As expected, no lateral shifts in in-plane beam profiles were observed and the dose distribution is symmetric and peaks at CAX for both measured and calculated data.

3.C | Reference dose verification
Based on the type of ion chamber and its orientation with respect to the irradiation and magnetic fields, a k Qmsr B of 0.994 was used for the reference dose verification. 18 The absolute dose at the measure-

3.D | End-to-end tests
The three single-beam plans with a 5 × 10 cm 2 field at gantry 0°, 90°, and 270°were delivered to the ArcCheck and the measured dose was compared with the TPS calculations. Figure 6 displays dose comparison between a planned and ArcCheck measured 5 × 10 cm 2 rectangular field at gantry 0°using SNC Patient TM software. The test beams at three gantry angles all passed with more than 99% in absolute mode using gamma analysis and clinic criteria. These excellent agreements indicate that the ArcCheck was aligned properly.
The two step and shoot IMRT QA plans, with 6 and 5 beam angles, respectively, were delivered to the ArcCheck on MR-Linac.

| DISCUSSION AN D CONCLUSIONS
The beam model to account for ERE for a high-field MR-Linac was validated with measurements of relative and absolute dosimetry. The ERE is visible in penumbra especially for small fields which is correctly modeled by the TPS. Small dosimeter such as the micro ion chamber and the diamond detector used in this work is found to be acceptable for small field dosimetry for the MR-Linac. The diode array, such as the IC profiler, may be used for more frequent beam profile QA purposes for field sizes ≥ 5 × 5 cm 2 , while radiochromic film can be used for frequent small field size < 5 × 5 cm 2 beam profile checks.