Evaluation of the RUBY modular QA phantom for planar and non‐coplanar VMAT and stereotactic radiations

Abstract Purpose This study evaluates the clinical use of the RUBY modular QA phantom for linac QA to validate the integrity of IGRT workflows including the congruence of machine isocenter, imaging isocenter, and room lasers. The results have been benchmarked against those obtained with widely used systems. Additionally, the RUBY phantom has been implemented to perform system QA (End‐to‐End testing) from imaging to radiation for IGRT‐based VMAT and stereotactic radiations at an Elekta Synergy linac. Material and Methods The daily check of IGRT workflow was performed using the RUBY phantom, the Penta‐Guide, and the STEEV phantom. Furthermore, Winston–Lutz tests was carried out with the RUBY phantom and a ball‐bearing phantom to determine the offsets and the diameters of the isospheres of gantry, collimator, and couch rotations, with respect to the room lasers and kV‐imaging isocenter. System QA was performed with the RUBY phantom and STEEV phantom for eight VMAT treatment plans. Additionally, the visibility of the embedded objects within these phantoms in the images and the results of CT and MR image fusions were evaluated. Results All systems used for daily QA of IGRT workflows show comparable results. Calculated shifts based on CBCT imaging agree within 1 mm to the expected values. The results of the Winston–Lutz test based on kV imaging (2D planar and CBCT) or room lasers are consistent regardless of the system tested. The point dose values in the RUBY phantom agree to the expected values calculated using algorithms in Masterplan and Monte Carlo engine in Monaco within 3% of the clinical acceptance criteria. Conclusion All the systems evaluated in this study yielded comparable results in terms of linac QA and system QA procedures. A system QA protocol has been derived using the RUBY phantom to check the IGRT‐based VMAT and stereotactic radiations workflow at an Elekta Synergy linac.


| INTRODUCTION
The process of radiation therapy from patient imaging for the treatment planning until the radiation at the linear accelerator (linac) requires seamless integration of multiple system components along the chain. Any errors that occur within any system components could affect the quality of the treatment plan delivered and hence the patient's clinical outcome. Therefore, stringent tests are designed and carried out either at regular intervals or on a patient-to-patient basis aiming at detecting these errors before the treatment begins and during the course of the treatment. There exist several guidelines, which recommend quality assurance (QA) protocols for specific system components, like the TG 142 1 for linac, the TG 179 2 for CTbased IGRT workflow, the AAPM practical guideline 5.a 3 for treatment planning system, and the TG 218 4 for the patient-specific plan verification.
In view of emerging technologies in radiation therapy, the complexity of the radiation therapy process has increased, partly due to the integration of additional system components into the treatment chain like onboard cone-beam CT (CBCT) imaging systems or optical body scanning systems as used for surface-guided radiation therapy (SGRT). Image-guided radiation therapy (IGRT) relies on the daily image acquisitions to identity errors in patient immobilization and positioning. The results from the image guidance routine are derived based on either an automated or a manual registration workflow between the daily and the reference images. The execution of couch corrections resulted from this procedure involves the communication between different system components, such as the imaging software, the Record and Verify (R&V) system, and the couch control system. Adaptive radiation therapy also relies heavily on the use of onboard imaging, but not only for the verification of patient positioning, but also for plan adaption based on daily anatomical changes assessed from these images. The QA of this workflow involving multicomponents is essential for correct patient treatment and in case of remote controlled couch system, a daily-based QA workflow is also recommended. Thereby, it is essential to implement check procedures to ensure the congruence of the isocenters of the imaging systems and the linac. This can be realized for example by performing the procedure introduced by Lutz et al 5

179.
However, even if QA tasks within each system component are passed, the correctness of the complete radiation therapy process is still not guaranteed as the outcome necessitates the faultless interplay between different system components due to their underlying dependencies. Therefore, a unified system QA, also referred to as End-to-End testing, that covers the entire process from the beginning to the end is desired to identify these system dependencies and any flaws within the whole radiation therapy process. The system QA should also ensure the data integrity when information is passed between various system components. During the execution of the system QA, the tasks should be carried out by the same clinical team members, who are performing these tasks clinically, so that possible faults caused by user's interactions with these system components can be pinpointed.
No universal protocols for system QA are available due to the diversity of the clinical systems in use and institutional specific workflows. The system QA should be designed to reflect every aspect and system component along the chain of a radiation therapy process as clinically realistic as possible. This execution of the system QA can be considered as a dry-run of the whole process, only without the patient. This suggests the need to use a patient surrogate, typical a phantom or ideally, an anthropomorphic phantom.
In this work, the clinical use of a new modular phantom (RUBY, PTW Freiburg, Germany) has been evaluated as a universal system QA phantom for VMAT irradiations as well as IGRT-specific QA in terms of geometry accuracy checks. The performance of the new phantom is compared to an established head anthropomorphic phantom 6 (STEEV, CIRS, Norfolk, USA) used for system QA of intracranial stereotactic radiation, a widely used IGRT QA phantom 7 (Penta-Guide, Modus Medical Devices Inc, London, Canada); and a ballbearing system (Elekta, Stockholm, Sweden). Eight VMAT plans with 6 MV flattened beam were created using Oncentra Masterplan (Elekta, Stockholm, Sweden) (see Table 1), which were optimized and calculated using the collapsed cone algorithm. The studied treatment plans include six coplanar radiations with dose per fraction between 1.8     T A B L E 1 Overview of the treatment plans studied. Six coplanar treatment plans with dose per fraction between 1.8 Gy and 6 Gy. Two noncoplanar treatment plans with dose per fraction of 6 Gy. Target volume sizes are between 4.8 and 486 cm 3 .

2.B.4 | Ball-bearing phantom
The ball-bearing phantom (Elekta, Stockholm, Sweden) consists of a ball bearing with diameter of 8 mm made of steel that is embedded in an acrylic wand. The wand is mounted on a three-dimensional positioning stage that can be adjusted using the built-in micrometers, with which the ball bearing can be positioned at the imaging isocenter with the help of planar images or CBCT. Subsequently, the room lasers can be tuned to align with the markers on the phantom.

2.C | Detectors
The PinPoint 3D (type 31022) ionization chamber and microDiamond The difference between measurement and TPS calculation was determined using the following formula: where a positive difference value indicates that the measurement value is larger than the TPS calculation and vice versa.

2.D | Linac quality assurance
The purposes of linac quality assurance using the phantoms evalu- In case of the Penta-Guide phantom, the "Grey values (T + R)" preset had to be used, since no bone structures are embedded in the phantom.  In the next step, the patient treatment plans in Table 1 were  Table 2 for the three phantoms investigated and the difference between expected and detected shift is shown in Fig. 2. For CBCT imaging, the calculated shifts in lateral and vertical directions agree within 1.6 mm to the expected values. For the longitudinal direction, the deviation is up to 3 mm for all phantoms. The standard deviation for CBCT imaging is less than 0.5 mm.

2.D.1 | Geometry accuracy
In case of MV planar imaging, the agreement between the calculated shifts and expected values is up to 3.7 mm. It is noteworthy that the image registration of the MV planar images was performed manually, whereas the image registration of the CBCT images was automated based on the implemented algorithm in the software.

3.A.2 | Geometry accuracy
The calculated diameters of the isospheres resulted from the Win- with RUBY phantom and ball-bearing phantom are presented in Table 3 for the three methods of positioning.
The calculated offset positions are plotted in Fig. 3. The uncertainty of the positioning method is different in each case. According to the results in Table 2 Table 2, where all three phantoms showed a deviation in the longitudinal direction.    F I G . 6. Differences between measurements in RUBY phantom with System QA insert or STEEV phantom with microDiamond insert and TPS calculations using collapsed cone (upper panel), pencil beam (middle panel), and Monte Carlo (lower panel) algorithms. Measurements were performed with microDiamond and PinPoint 3D ionization chamber in case of RUBY and microDiamond only in case of STEEV. Before the measurements, the IGRT workflow was performed using CBCT imaging.

3.B | System (End-to-End) QA
from charge imbalance in the conductive detector components. [9][10][11] The air-filled PinPoint 3D chamber underresponds in small fields due to the volume-averaging effect and the low-density sensitive volume. 12,13 Nevertheless, no correction factors have been applied to the measurements due to the lack of these factors for composite fields. Despite the small target volumes of plan 3 to 8 that correspond to the typical situations of stereotactic radiations, the consistency between the two detectors and the good agreement to TPS calculations indicate that both detectors are suitable for such point dose measurements within the scope of system QA. In these cases, the uncertainty due to the detector-specific field size-dependent dose response falls within the acceptance criteria of 3% of the system QA. Nevertheless, extra precautions must be taken when these tests are performed for treatment plans of target volumes with size comparable to the dimensions of the used detectors. Since system QA is not intended to substitute treatment plan verification, but more to test the system dependencies and to identify flaws within the radiation therapy process, independent 2D or 3D dose measurements using detector arrays are still recommended for individual plans to allow for more comprehensive comparisons between actual and calculated dose distributions.

CONFLI CT OF INTEREST
Daniela Poppinga is an employee of PTW Freiburg, Germany.