Quality assurance of geometric accuracy based on an electronic portal imaging device and log data analysis for Dynamic WaveArc irradiation

Abstract The purpose of this study was to develop a simple verification method for the routine quality assurance (QA) of Dynamic WaveArc (DWA) irradiation using electronic portal imaging device (EPID) images and log data analysis. First, an automatic calibration method utilizing the outermost multileaf collimator (MLC) slits was developed to correct the misalignment between the center of the EPID and the beam axis. Moreover, to verify the detection accuracy of the MLC position according to the EPID images, various positions of the MLC with intentional errors in the range 0.1–1 mm were assessed. Second, to validate the geometric accuracy during DWA irradiation, tests were designed in consideration of three indices. Test 1 evaluated the accuracy of the MLC position. Test 2 assessed dose output consistency with variable dose rate (160–400 MU/min), gantry speed (2.2–6°/s), and ring speed (0.5–2.7°/s). Test 3 validated dose output consistency with variable values of the above parameters plus MLC speed (1.6–4.2 cm/s). All tests were delivered to the EPID and compared with those obtained using a stationary radiation beam with a 0° gantry angle. Irradiation log data were recorded simultaneously. The 0.1‐mm intentional error on the MLC position could be detected by the EPID, which is smaller than the EPID pixel size. In Test 1, the MLC slit widths agreed within 0.20 mm of their exposed values. The averaged root‐mean‐square error (RMSE) of the dose outputs was less than 0.8% in Test 2 and Test 3. Using log data analysis in Test 3, the RMSE between the planned and recorded data was 0.1 mm, 0.12°, and 0.07° for the MLC position, gantry angle, and ring angle, respectively. The proposed method is useful for routine QA of the accuracy of DWA.


| INTRODUCTION
Intensity-modulated radiotherapy (IMRT) is an effective method for achieving high dose conformity for the target in radiotherapy. 1 Volumetric-modulated arc radiotherapy (VMAT) has been developed to reduce the treatment time by allowing additional degrees of freedom, such as variations in the gantry speed and dose rate, as well as dynamically changing the shape of the field, 2-6 while noncoplanar VMAT has further improved the dose distribution. [7][8][9] Dynamic WaveArc (DWA), a new function incorporated into Ver-o4DRT instruments (MHI-TM2000; Mitsubishi Heavy Industries, Ltd., Hiroshima, Japan, and Brainlab AG, Feldkirchen, Germany), could be used to perform novel three-dimensional noncoplanar irradiation. 10,11 The Vero4DRT is composed of an O-ring gantry that is designed to rotate AE180°, and can itself also rotate AE60°around the vertical axis. The MLC design is a single-focus type, has 30 pairs of 5 mm thick leaves at the isocenter, and produces a maximum field size of 150 9 150 mm 2 . The DWA technique is a beam delivery method designed to maximize the versatility of the Vero4DRT by synchronizing the noncoplanar movement of the gantry/ring (G/R) with the optimization of the dynamic multileaf collimator (MLC). It is a similar technique to noncoplanar VMAT, and noncoplanar beam directions are capable of being selected from a list of preinstalled trajectories in the treatment planning system. Noncoplanar VMAT produces a high conformal dose distribution, but its delivery is inefficient because it involves rotating the patient couch. [12][13][14] One of the main characteristics of the DWA technique is continuous noncoplanar VMAT without the requirement to move the couch. To maximize the benefits of the DWA approach, the Vero4DRT system incorporates the following capabilities: variable dose rate, variable gantry speed, variable ring speed, and dynamic MLC movement, with the expectation that these will optimize dose conformity, delivery efficiency, accuracy, and reliability.
DWA is a complex irradiation technique, and it is important to ensure that the device is operating correctly. Several studies have reported the geometric accuracy for DWA irradiation. Sato et al. comprehensively examined machine-limiting accuracy during DWA irradiation in a number of situations using various dose rates, G/R angle positions, and speeds. 15 Burghelea et al. developed a novel evaluation method for measuring the accuracy of the G/R position using a cube phantom with a kilovolt x-ray imaging subsystem. 16 This procedure is effective for both commissioning and detailed verification. On the other hand, a simple verification method that can quickly measure and automatically analyze is required for routine quality assurance (QA).
The purpose of this study was to develop a simple verification method for the routine QA of DWA irradiation. Several studies reported that an electronic portal imaging device (EPID) and log data analysis had sufficiently verified the accuracy of the mechanical uncertainty, that is, MLC and gantry position as well as delivery error, during VMAT. [17][18][19] Therefore, we investigated the application of the QA method based on the EPID and log data analysis to DWA irradiation.
2 | MATERIALS AND METHODS 2.A | DWA QA using the EPID and log data analysis DWA is an extension of noncoplanar VMAT and its irradiation accuracy depends on a complex combination of various factors. With respect to the mechanical restriction on a characteristic Vero4DRT with DWA irradiation, the maximum dose rate, gantry rotational speed, ring rotational speed, and MLC speed are 400 monitor units (MU)/min, 6.0°/s, 2.5°/s, and 4.0 cm/s, respectively.
Our proposed method utilized EPID images and log data analysis.
Fluence profiles were evaluated using EPID images. EPID calibration of the Vero4DRT was performed in the manner specified by the manufacturer, by acquiring a flood and a dark-field image. The amorphous silicon EPID on the Vero4DRT has a 180 9 180 mm 2 detection area with a matrix size of 1024 9 1024; that is, 0.18 mm/pixel at the isocenter plane. EPID images were acquired at a rate of 1.75 frames/s. EPID images were analyzed by using relative values. The performance of the machine during DWA irradiation was also analyzed using two sets of log data: the G/R control log and the MLC control log. The G/R control log captured the accumulated MU, dose rate, gantry, and ring angles. Meanwhile, the MLC control log recorded the MLC positions (defined at the isocenter). They were recorded every 50 ms with the same time stamp by using the same controller. Analysis software based on Matlab version 8.6 (MathWorks Inc., Natick, MA, USA) was developed to evaluate the machine's performance automatically. All of the following tests irradiated the EPID and recorded log data that were then analyzed by the in-house software.

2.B | Calibration method for misaligned EPID geometry during DWA irradiation
In general, the alignment of the megavoltage treatment beam and EPID changes during gantry rotation due to gantry and/or detector sag. In the case of the Vero4DRT, the gimbaled x-ray head and EPID are mounted on the rigid O-ring structure; therefore, misalignment of the beam axis is reduced. Moreover, the beam axis of each angle is sufficiently accurate owing to beam axis correction using a gimbal head [20][21][22] (Fig. 1). However, to evaluate MLC positional and output accuracy using the EPID during G/R rotation, the misalignment between the center of the EPID and beam axis needs to be corrected. Therefore, an automatic calibration method for the misaligned EPID geometry was developed. The outermost MLC pairs, which are not usually used for treatment, form narrow slits and these were fixed during irradiation. The EPID images were captured in continuous mode during 360°clockwise rotation of the gantry and AE40°rotation of the ring. The O-ring angle of DWA trajectory was limited to approximately AE40°due to gantry-couch collision. After that, the position of the center of mass (COM) in the narrow slit was detected in each frame.
According to the detected COM, the misalignment for each frame of the EPID image was then corrected. Then, all EPID images were converted to an integrated fluence map (Fig. 2).
In addition, to evaluate the detection accuracy of the EPID images, the five x-ray slit fields in a static gantry position were irradiated by introducing an intentional error in the MLC slit width within the range 0.1-1 mm. Mean absolute error (MAE) and standard deviation (SD) of peak positions of the slit width on the exposed EPID and log data were analyzed by the in-house software and compared to those taken without the intentional value. To eliminate the possibility of error other than the MLC position, this test was performed at static gantry.
2.C | Development of a QA procedure for DWA irradiation Ling et al. reported a step-by-step approach that examines the functional ability to deliver accurate treatments using the complex irradiation method of VMAT. 23 In this work, the QA procedure for DWA irradiation was determined with reference to the reported method.

2.C.1 | Test 1: Accuracy of the MLC position
To assess the accuracy of the dynamic MLC leaf position, a picket fence test was performed at a stationary gantry angle of 0°and during DWA irradiation. The picket fence test consisted of five narrow bands with a slit width of 2 mm and spaced at intervals between two central positions of the leaf gap of 29 or 30 mm. For DWA irradiation, the gantry was rotated by 360°in a clockwise direction, while the ring was rotated by AE40°. The dose rate, gantry speed, and ring speed were 400 MU/min, 1.67, and 0.83°/s, respectively.   to provide the same output to the four strips (see Table 1). These combinations were determined based on the following considera-

2.C.3 | Test 3: Accurate control of MLC leaf speed
The output constancy was assessed as a function of the MLC leaf speed during DWA irradiation. This test used four different combinations of the dose rate, gantry speed, ring speed, and MLC speed to give the same output to the four strips (see Table 2). These combinations were also determined based on the above considerations.
The width of each of the strips was 30 mm, and the size of the inter-strip gaps was 5 mm. The DWA profiles were normalized by the maximum value of the stationary gantry position. The averaged central output profile agreement between the stationary gantry angle of 0°and DWA irradiation was also inspected using in-house software and evaluated according to the RMSE, MAE, and SD in the exposed field, except for 5 mm from the field edge (MLC number 15). The stationary gantry angle in test 3 referred to same field using dynamic MLCs. The RMSE of the MLC and G/R motion was evaluated by log data analysis.

| DISCUSSION
The proposed method showed that the beam fluence detected by the EPID images was combined with log data used to assess output rather than log data. 19 These results showed that log data must be carefully used to assess MLC position. In this study, we compared MLC slit widths in the integrated EPID image and ones in log data by using the static picket fence test without the intentional error. These results represented that the difference of MLC slit widths between EPID and log data was small and the reasonable agreement with EPID. Therefore, we confirmed that the detection accuracy of log data-based analysis was less than 0.1 mm, which is same for EPID detection accuracy. Less than 0.1 mm detection accuracy of log analysis was sufficient in the clinical.
The picket fence test was successfully adapted for use in the QA of DWA irradiation in Test 1. By comparing irradiation profiles in the static gantry and DWA modes, the effect of simultaneous G/R rotation and leaf position accuracy was assessed. The difference in MLC width in the static gantry and DWA modes was less than 0.2 mm, which is less than 0.27 mm with the maximum MLC position deviation reported by Jørgensen et al. 26 The RMSE in the MLC position based on log data was 0.03 mm, which is less than the 0.5-mm deviation of the MLC position suggested by Ling et al. 23 The displacement of the MLC between slits was <0.3 mm during DWA irradiation. The position which showed the largest error in the MLC width between slits coincided with the offset position. This is comparable to studies that showed that MLC position errors between slits were less than 0.5 mm at the offset position. 18

CONF LICT OF I NTEREST
The authors have no relevant conflicts of interest to disclose.