Comparison of pretreatment VMAT quality assurance with the integral quality monitor (IQM) and electronic portal imaging device (EPID)

Abstract The purpose of this study was to compare pretreatment volumetric modulated arc therapy (VMAT) quality assurance (QA) measurements and evaluate the multileaf collimator (MLC) error sensitivity of two detectors: the integral quality monitor (IQM) system (iRT systems IQM) and the electronic portal imaging device (EPID) (Varian PortalVision aS1200). Pretreatment QA measurements were performed for 20 retrospective VMAT plans (53 arcs). A subset of ten plans (23 arcs) was used to investigate MLC error sensitivity of each device. Eight MLC error plans were created for each VMAT plan. The errors included systematic opening/closing (±0.25, ±0.50, ±0.75 mm) of the MLC and random positional errors (1 mm) for individual/groups of leaves. The IQM was evaluated using the percent error of the measured cumulative signal relative to the calculated signal. The EPID was evaluated using two methods: a novel percent error of the measured relative to the predicted cumulative signals, and gamma (γ) analysis (1%/1 mm, 2%/2 mm, 3%/3 mm and 3%/1 mm for Stereotactic Body Radiation Therapy plans). The average change in maximum dose obtained from dose‐volume histogram (DVH) data and change in detector signals for different systematic MLC shifts was also compared. Cumulative signal differences showed similar levels of agreement between measured and expected detector signals (IQM: 1.00 ± 0.55%; EPID: 1.22 ± 0.92%). Results from γ analysis lacked specificity. Only the 1%/1 mm criteria produced data with remarkable differences. A strong linear correlation was observed between IQM and EPID cumulative signal differences with MLC error magnitude (R = 0.99). Likewise, results indicate a strong correlation between the cumulative signal for both detectors and DVH dose (RIQM = 0.99; REPID = 0.97). In conclusion, use of cumulative signal differences could be more useful for detecting errors in treatment delivery in EPID than γ analysis.


| INTRODUCTION
Volumetric modulated arc therapy (VMAT), a form of rotational intensity-modulated radiation therapy (IMRT), has been widely implemented in radiotherapy as a tool to deliver heterogeneous dose distributions providing high doses to target volumes while sparing normal tissues. In VMAT, multiple overlapping arcs with simultaneous variation of the gantry speed, multileaf collimator (MLC) movement, and dose rate are used to create fields with modulated beam intensities. [1][2][3][4] Despite its numerous advantages, the implementation of VMAT adds complexity to treatment planning and delivery thereby increasing the sources of error in their workflows. 5,6 These added uncertainties in the VMAT process highlight the need for patient specific pretreatment quality assurance (QA) of treatment plans to verify the accuracy of dose calculations and detect clinically relevant errors in radiation delivery.
Currently, most pretreatment QA for IMRT and VMAT plans is measurement based. A number of techniques can be used for measurements with the workflow typically consisting of recalculation of the approved treatment plan on a dosimeter and subsequent irradiation in the same geometry. 5 The calculated and measured dose distributions are compared, and the plan is either approved or rejected for treatment based on in-house specific acceptance criteria. An advancement in the science of QA for radiotherapy has been the development of Portal Dosimetry. The electronic portal imaging device (EPID) is a digital MV imaging detector attached opposite to the treatment head of the linac gantry. It is primarily used for verification of patient positioning during treatment; however, its dosimetric properties have led to its utilization in patient-specific QA measurements. [7][8][9][10] This technique consists of comparing predicted portal dose distributions and acquired portal images using an evaluation method, typically the gamma (γ) index. 7,11 However, this methodology is intrinsically limited by numerous complications. Use of a portal dose prediction algorithm in place of the actual dose calculation algorithm is problematic and can mask errors, such as those related to the quality of the dose calculation model, that occur downstream from calculation of the fluence. 12 Furthermore, EPID measurements are subject to issues from the use of digital detectors, such as contributions from electronic noise. 8 Lastly, although the γ index is commonly employed in patient-specific QA, limitations associated with its use should also be considered. 13,14 Recent advances in transmission detector technology offer a potential alternative to the EPID. The integral quality monitor (IQM) is a novel transmission detector consisting of a large-area ionization chamber capable of performing high sensitivity charge collection measurements. The detector is mounted on the accessory tray of the linear accelerator treatment head and connects wirelessly to a controlling workstation via a Bluetooth transceiver. 15 Patient-specific VMAT QA can be performed by comparing chamber readings to calculated signal values. 16 The IQM directly measures signal from the approved treatment plan generated with the actual dose calculation algorithm in the treatment planning system (TPS). The system is robust with fewer moving parts requiring less maintenance and Qualified Medical Physicist time allocation. 17 Finally, IQM does not require use of the γ index for comparison of two-dimensional dose distributions.
Previous studies in the literature have investigated the use of the IQM system in the context of pretreatment QA. 16,[18][19][20] The study by Hoffman et al. showed that several types of IMRT errors can be found with the IQM system. 19 Razinskas et al. investigated the use of the IQM system for VMAT verification which is inherently more complex. 16

2.A.1 | Electronic portal imaging device
The PortalVision as1200 flat-panel EPID used in this study consists of an amorphous silicon (a-Si) photodiode detector array attached to a Varian TrueBeam (V2.7) linear accelerator gantry through a robotic arm. The detector features an active area measuring 40 cm × 40 cm with an 1190 × 1190-pixel array with pixel pitch of 0.336 mm. Each pixel is composed of a metal plate covering a scintillator (Lanex Fast Back) over an a-Si layer with embedded photodiode and thin film transistor on glass substrate. 21,22 EPID pretreatment QA of IMRT and VMAT plans consists of comparing acquired portal images to predicted portal dose images commonly using the γ index. Portal images are acquired by the detector measuring the fluence of each field and subsequently using analytic software to correlate the response to dose delivered. 23 The relationship between the EPID readout signal per pixel and measured portal dose is determined by the imager calibration. 8 The predicted portal dose distribution is calculated by the standalone algorithm portal dose image prediction (PDIP) (V13.6), which has an internal calculation resolution of 0.39 mm. 24 Quantitative comparison of predicted and measured portal dose distributions is typically performed using the γ index developed by Low et al. 11

2.A.2 | Integral quality monitor
The IQM detector is comprised of three aluminum alloy plates  Patient-specific VMAT QA can be performed using the IQM system by evaluating agreement between the expected and measured signals for a given treatment field. 16 This can be done in real time or after the field has been delivered. For a beam segment of U MU, the signal (C IQM ) across the area of the chamber is given by

2.C | Evaluation metrics
Agreement between expected and measured values for the two systems was compared using the cumulative signal, γ index, and dosevolume histogram (DVH) data. The γ index is not an appropriate tool for evaluation of IQM measurements as the system does not provide a two-dimensional spatial response. Hence, a novel evaluation metric, the EPID cumulative signal (C EPID ), was developed in addition to the γ index to directly compare the two systems. A comparison of detector response to DVH data was also performed for plans with MLC errors.

2.C.1 | Cumulative signal
The IQM and EPID were evaluated using the cumulative signal difference (CSD) given by the deviation of the measured cumulative signal (C meas ) relative to the reference cumulative signal (C ref ) for each field: A lower threshold equivalent to 1% of the maximum pixel value was applied to remove signal contributions from electronic noise.
This threshold value was found to minimize noise and improve the signal-to-noise ratio. Lower threshold values did not remove enough of the signal noise, while higher threshold values eliminated too much data from the analysis.
Detector error sensitivity (S error ) was also evaluated using the cumulative signal difference, but with the reference value (C ref ) substituted by measured cumulative signal of the original plan (C meas,original ): In Eq. (4), C meas,error is the cumulative signal measured for the field with the MLC error.

2.C.2 | γ analysis
Portal images were further evaluated using γ analysis to provide a comparison to current methodology. 11 The improved γ algorithm in the ARIA RTM Portal Dosimetry application (V13.6, Varian Medical Systems, Palo Alto, CA, USA) was used to calculate γ passing rates for each field. 11,26 Evaluation was performed using global normalization and absolute dose. The region of interest (ROI) was defined by the maximum MLC opening plus 0.5 cm. The dose threshold was set to 10% to exclude low-dose areas with minimal clinical relevance which could significantly bias results. 5 The following dose difference (DD) and distance-to-agreement (DTA) criteria were used, DD/DTA: 3%/3, 2%/2, and 1%/1 mm. SBRT plans were also evaluated with the 3%/1 mm criteria.

2.C.3 | DVH comparison
Correlation between detector response and treatment plan dose was evaluated for MLC error plans. Maximum doses for target volumes and organs at risk were obtained from DVH data from the original and error plans in the TPS. The average relative change in dose caused by each error was calculated and compared to the IQM and EPID cumulative signal difference.

2.C.4 | Effects of different field parameters on detector response
The following field parameters were evaluated to determine their effect on IQM and EPID detector response: modulation index (MI) and field size. The MI indicates the degree of modulation for each arc and was defined as the ratio of MU to dose (cGy) as specified by the TPS. The field size was defined by the opening of the collimator jaws. For arcs with jaw tracking, the field size was defined by the maximum opening of the jaws throughout the arc.

2.D | Statistical analysis
Linear correlations between different variables reported were evaluated using the Pearson correlation coefficient (R). The correlation has a value between +1 and −1, where +1 indicates a total positive linear correlation and −1 indicates a total negative linear correlation.
To determine whether the correlation was statistically significant, the probability (p-) value was calculated and compared to a significance level (α) of 0.01. Correlations with p-values ≤ α were determined to have a correlation that is different from 0 and statistically significant within a 99% confidence interval.   Mean CSD values and γ passing rates for different treatment sites are provided in Table 2. No correlation was observed between IQM and EPID cumulative signal differences with respect to plan type. In some instances, such as chest plan measurements, plans having the highest levels of agreement when measured using the IQM, were found to exhibit the greatest deviation from the calculated reference signal when measured using the EPID. Furthermore, GHAFARIAN ET AL.

| 169
there was no visible correlation between γ passing rates and IQM or EPID cumulative signals when evaluated by plan type.
The effects of different field parameters on detector response were evaluated. A moderate positive correlation (R = 0.50) was observed between the IQM cumulative signal difference and MI.
There was no evidence indicating EPID cumulative signal differences were affected by the degree of field modulation. Additionally, field size and MU were not found to be statistically significant predictors of the cumulative signal difference for the detectors. Although, there was a visible negative correlation between γ passing rates and field size with coefficients of R = −0.24, R = −0.47, and R = −0.58 for the 3%/3, 2%/2, and 1%/1 mm criteria, respectively.

3.B | MLC error sensitivity
In the next step, ten of the clinical VMAT plans (23 arcs) were modified to include MLC errors and evaluated in the same way. Table 3 summarizes detector error sensitivity (S error ) and Portal Dosimetry γ passing rates by MLC error type and magnitude. A strong correlation was observed between detector response and the magnitudes of systematic MLC errors. Correlation coefficients for the IQM and EPID T A B L E 1 Cumulative signal differences (CSD) and γ passing rate frequency distribution for 20 VMAT plans (n = 50 arcs).   and ≤90%, respectively. The γ index did not detect any of the errors in the modified plans when the 3%/3 mm criteria were used. Moreover, results from analysis using the 2%/2 mm criteria showed that passing rates did not meet tolerance and action limits for 11% and 3% of fields, respectively. Results from the 1%/1 mm analysis showed at least one field failing to meet the action limit for each type of error. For these criteria, approximately 68% of fields were within watch limits and 46% of fields failed to meet the ≥90% acceptance criteria.

CSD (%)
T A B L E 2 IQM and EPID pretreatment VMAT QA results by treatment site. Cumulative signal differences (CSD) and γ passing rates are presented (n = 50 arcs).

CONFLI CT OF INTEREST
No conflicts to disclose.

AUTHOR CONTRI BUTION STATEMENT
All authors satisfy the authorship requirements and do not have anything to disclose.