The integration of MR imaging and MV beam generation achieved with the Elekta Unity MRL (Elekta, Stockholm, Sweden) design presents quality assurance challenges. Following the on-site construction of the system, Elekta personnel perform device acceptance tests (DAT) that replace conventional linac acceptance tests. Commissioning and beam modelling validation measurements are also performed by Elekta personnel, with Philips staff responsible for MR image quality testing. Prior to clinical go-live, a period of quality assurance follows during which in-house physicists perform baseline quality assurance and validation measurements across both MR and MV modalities. Elekta tests incorporate specialised QA devices and analysis software that are not necessarily commercially available. Furthermore, due to the presence of the magnetic field, conventional equipment available to the clinic may not be compatible with the Unity system.
The design of the Unity system has been comprehensively discussed by other investigators, [1–5] and the reader is referred to these works for further information. In short, the Elekta Unity MRL is a combination of a modified Philips Ingenia 1.5 T MRI with a split-coil superconducting magnet and a straight-through linear accelerator. The beam generation system, producing a single 7 MV FFF x-ray source, is mounted on an annular gantry that is free to rotate around a cylindrical cryostat containing the static-field MR coils. The axis of gantry rotation and the central axis of the coils are coincident, with the static magnetic field (B0) in the negative Y direction (IEC61217) as shown in Fig. 1. For all gantry angles the beam passes through the cryostat and is perpendicular to B0. The angular dependent beam transmission through the cryostat (aluminium annular structure containing liquid helium) is referred to as the cryostat characterisation and will vary between Unity systems mostly due to differences in construction of the cryostat annulus; however, a small component will be due to differences in helium fill [3]. A modified Elekta Agility® Beam Limiting Device (BLD) shapes fields ranging from 0.8 × 0.5 cm2 to 57.4 × 22.0 cm2 at isocentre, with a fixed dose-rate at this point of 425.0 MU/min. The patient positioning system (PPS) is capable of longitudinal movement only and the isocentre is 14.0 cm above the PPS, 143.5 cm from the source. A fixed EPID panel, now called the mega-voltage imager (MVI) [5], diametrically opposite the x-ray source on the gantry is capable of MV portal imaging a maximum field size of 22.0 × 9.5 cm2, for QA purposes only. A schematic of the MR-linac is shown in Fig. 1, courtesy of Elekta.
The Elekta Unity MRL in conjunction with the Monaco v5.40 treatment planning system, enables real-time adaptive radiotherapy planning. Within Monaco, the GPU Monte Carlo dose (GPUMCD) algorithm is used for fast, Monte Carlo dose calculations in the presence of a static 1.5 T magnetic field [6, 7]. Dose calculations are performed on voxelized models of patient or QA phantom datasets with relative electron density (RED) assigned per voxel based on a CT specific, user defined CT number to RED table. For calculations performed on MR datasets, RED information is assigned on a per structure basis with user specified values; typically, the mean RED for a structure is calculated from a reference CT scan. REDs are converted to mass densities, with the result mapped to chemical composition using Patient, Phantom or Couch material look-up tables. Users can specify a dose calculation grid resolution and statistical uncertainty to control the accuracy and duration of calculations. For plan adaptation two workflows are available, Adapt-to-Position (ATP) or Adapt-to-Shape (ATS). ATP involves repositioning of pre-treatment (reference) plan isocentre based on the rigid registration of that plan and image dataset with a daily MR image. The pre-treatment plan can be recalculated or reoptimized on the reference dataset to reproduce or improve target dose coverage. ATS allows for plan adaption based on anatomical changes as shown on the daily MR image. Contours can be automatically deformed to match the daily anatomy, with optional user-based adjustments. The reference plan is recalculated or reoptimized using reference plan constraints [8]. For more information on the replanning options the reader is referred to other works [6–8].
Published work focused on commissioning of the Elekta Unity system is limited [3–5, 9]. Recently, the work of Roberts et al [3] provides tolerances and frequencies for common QA tests and Woodings et al [5] for acceptance procedures on the linac component. Since this technology has only recently been clinically introduced (results reported in this work were obtained with the first Unity in the Southern Hemisphere), there is currently no region-specific guidance for implementation and safe use of the MRL, as is the current situation for conventional treatment systems [10]; however, position papers are unlikely to vary significantly from the recommendations of the above investigators [3, 5]. It is the intention that our work will aid clinics introducing an Elekta Unity and provide benchmarks for test results not previously reported.
Although numerous commissioning tests may be performed on an Elekta Unity, only a subset are presented below. Adaptions of the Elekta supplied MV alignment phantom (Fig. 2(a)) enabled independent verification of the gantry angle accuracy and reproducibility and the MV isocentre, whereas development of in-house phantoms was required for measurements of beam attenuation due to the anterior imaging coil and End-to-End (E2E) testing. Difficulties were encountered when measuring beam quality and output, and when performing IMRT commissioning, with commercial equipment that required adaptions to standard methodologies and as such are also presented below. During installation there was limited time to facilitate customer selected measurements prior to magnet ramp up. A spontaneous quench and planned ramp down events enabled selected commissioning tests to be repeated with B = 0 T. More common commissioning measurements, for example mechanical behaviour of the system, MR-to-MV isocentre offset confirmation, relative dosimetry measurements, a thorough investigation of the treatment planning system (TPS) and the commissioning of the MR system are not provided in this work for conciseness. That work, together with MLC characterisation, mutual interference of MR and MV systems and a radiation survey within the treatment room will be published separately.
Due to the construction of the gantry ring, the use of an inclinometer to determine gantry angle is limited to 270.0° [5]. As part of routine QA for conventional linacs it is common to determine gantry angle accuracy for multiple positions and to assess the reproducibility of gantry rotation. A unique phantom design has been proposed for such measurements on the Elekta Unity MRL [11]; however, in the absence of such a phantom we propose the use of a vendor supplied equipment, Fig. 2(a) and Fig. 2(b), that can be adapted to investigate gantry angle positioning.
During device acceptance tests (DAT) an Elekta supplied Cryostat Characterisation Tool (CCT), see Fig. 2(c) is used with the MVI to obtain a series of images of a ballbearing for MV isocentre determination. These images are forwarded to Elekta for analysis using Radiation Isocentre Tool (RIT) software v6.6.64 (Radiological Imaging Technologies, Colorado Springs, Colorado, USA). Alternatively, Snyder et al [4] described the use of the MV alignment phantom, Fig. 2(a), to obtain MVI images of the centrally located ballbearing; however, image analysis relied on access to RIT software. In this work we demonstrate the use of the same phantom and central ballbearing to determine the MV-isocentre with in-house software analysis, and a comparison of this this methodology with those mentioned above is provided.
For all treatments on the Unity the anterior imaging coil is present above the patient and should be fully characterised in terms of it is dosimetric impact on treatment beams [5, 9]. The use of an in-house phantom to determine coil attenuation for a selection of gantry angles is described. Furthermore, a default coil model is provided in the Monaco TPS and a comparison of calculated and measured attenuation is presented.
The beam quality specifier for the Elekta Unity MRL is the TPR20,10 [9, 12-14], consistent with the recommendations of the TRS-398 protocol, and due to its insensitivity to the magnetic field [9, 13, 14]. Due to the difficulties with measuring PDDs, a direct measurement of TPR20,10 is preferred and with the presence of the cryostat, slight angular dependence of beam quality could be expected. With the quality dependent factors for reference output, such as \({k}_{{B}_{\parallel },Q}\) [13, 14] and \({k}_{Q,{Q}_{0}}\), a determination of this angular dependence, or at minimum an assessment of the change in these factors with beam quality, should be evaluated.
During DAT on the recommendation of Elekta the MRL was calibrated to give 1.000 Gy per 100 MU to isocentre, at 5.0 cm depth in water, for a 10.0 × 10.0 cm2 field, G90. Measuring from G90 is the preferred methodology due to the constancy of the helium fill at this angle compared to acute anterior angles, like G0, where the output may vary depending on the level of helium. However, other users report a calibration depth of 10.0 cm to maximize the delivered dose-rate [4] thereby reducing treatment times.
The PTW 1D water tank (PTW, Freiburg, Germany) is often used for output determination with the dosimetry protocol proposed by van Asselen et al [13]. When using the current version of the PTW 1D tank, a direct measurement of the output at isocentre for a chamber depth of 5.0 cm in water cannot be achieved at G90. Future versions of the PTW tank will address this issue; however, at present an adaption of the formalism in van Asselen et al [13] is required for clinics calibrating the Unity consistent with the Elekta advice. Calibration in a plastic water phantom was avoided due to variation in response of the chamber due to the presence of air gaps generating and the electron return effect (ERE) [14, 15].
As part of the Elekta beam validation procedure, nine vendor IMRT plans, based on AAPM TG-199 [16] guidance, were imported to the Monaco treatment planning system (TPS) and delivered on the Unity system. Elekta recommends gamma criteria of 3.0 % (global dose), 3.0 mm DTA during beam validation with the ArcCheck®-MR device (Sun Nuclear, Melbourne, Fl, USA). This device was not available during commissioning and in its absence the Octavius 1500MR (PTW, Freiburg, Germany) and Gafchromic™ EBT3/EBT-XD film (Ashland ISP Advanced Materials, NJ, USA) were utilized. The methodology for planning system calculations and the various measurement techniques are provided below.
Elekta provides phantoms for ATP and ATS E2E testing on the Unity during Physics Validation; however, a clinic may want to perform their own measurements for routine QA and, for example, during treatment site commissioning. Purchasing specific phantoms for individual site development can be expensive and furthermore, a centre may not have ready access to commercial phantoms for adaptive radiotherapy E2E testing. The use of in-house developed 3D printed phantoms can alleviate these issues. 3D printed materials have been extensively used in conventional radiotherapy applications [17–19] and their use in MR-guided radiotherapy (MRgRT) systems has seen much development [20, 21]. An in-house designed phantom with 3D printed components was used for E2E measurements for baseline comparisons on the Unity in this work.
Clearly, Unity users encounter challenges performing independent routine determination of common QA tests, such as those listed above. Due to the complexity of the Unity and the variety of possible tests, it is likely that physicists will not find comparative values in current publications for all commissioning work performed locally. The significance of our work is in the novel measurement techniques that describe alternative uses for vendor supplied phantoms and the use of in-house developed phantoms. Additionally, this work provides useful baseline results for future Unity users.