Survey results of 3D‐CRT and IMRT quality assurance practice

Abstract Purpose To create a snapshot of common practices for 3D‐CRT and intensity‐modulated radiation therapy (IMRT) QA through a large‐scale survey and compare to TG‐218 recommendations. Methods A survey of 3D‐CRT and IMRT QA was constructed at and distributed by the IROC‐Houston QA center to all institutions monitored by IROC (n = 2,861). The first part of the survey asked about methods to check dose delivery for 3D‐CRT. The bulk of the survey focused on IMRT QA, inquiring about treatment modalities, standard tools used to verify planned dose, how assessment of agreement is calculated and the comparison criteria used, and the strategies taken if QA fails. Results The most common tools for dose verification were a 2D diode array (52.8%), point(s) measurement (39.0%), EPID (27.4%), and 2D ion chamber array (23.9%). When IMRT QA failed, the highest average rank strategy utilized was to remeasure with the same setup, which had an average position ranking of 1.1 with 90.4% of facilities employing this strategy. The second highest average ranked strategy was to move to a new calculation point and remeasure (54.9%); this had an average ranking of 2.1. Conclusion The survey provided a snapshot of the current state of dose verification for IMRT radiotherapy. The results showed variability in approaches and that work is still needed to unify and tighten criteria in the medical physics community, especially in reference to TG‐218's recommendations.

of the medical physics community to ensure that progress is being made toward more robust and uniform management of IMRT QA.
Nelms and Simon surveyed several hundred sites about their planar IMRT QA methods and analysis, which helped identify how IMRT QA was evolving. 17 However, the scope of their survey was narrow and is now 12 years old. 17 Similarly, a survey of eight vendors was included in the TG-218 study to gain information on what was the current state of IMRT QA practice; but again, the scope was limited to specific aspects of IMRT QA, for example, gamma analysis and calculation (absolute vs relative), following the focus of the Nelms and Simon study. 18  To this end, a survey was created to broadly assess the current practice of IMRT QA, and at the same time capture other QA information, including MU verification for 3D-CRT. With this survey including multiple facets of QA, a baseline of common practices can be made known to the community. Our survey examined these facets of current QA practices and thereby provides a snapshot of the QA world for the medical physics community.

| ME TH ODS
The IROC Houston QA Center monitors sites that participate in NCI-sponsored trials through several audit processes that include annual output checks and anthropomorphic phantom irradiations. To track changes in each site's personnel, machines, and treatment modalities, IROC maintains an electronic Facility Questionnaire. This questionnaire is sent annually (or more often as needed) to every institution to allow for updates to the institution's status.

A QA survey was created and included in IROC-Houstons Facility
Questionnaire that was open from August 2011 to January 2018.
Our data includes only those sites that had updated their Facility Questionnaire in 2017 in order to capture the most up-to-date information. The survey had two main sections: 3D-CRT MU verification and IMRT QA. For all sections, standard/common answers were pro- vided, while open-ended "other" options were also available for less common/unexpected answers.  Respondents who indicated that they used point measurements indicated that they primarily used an ion chamber (84.7%), with approximately equal distribution of several chamber volumes: radius less than 0.1 mm (33.1%), 0.1-0.2 mm (29.0%), or 0.6 −0.7 mm (27.7%). Diodes (24.5%) and TLD (6.3%) were also used, but less often. 7.7% of respondents indicated an "other" device was used, which was most commonly a MOSFET.
For array detectors, respondents were asked if they evaluated their measurement in the phantom geometry or if they mapped it onto the patient dataset (ie DVH analysis). The vast majority of the responses indicated that they evaluated their measurement in the phantom geometry as shown in Table 3.
When making QA measurements, 62.6% of sites deliver beams at the planned gantry angle, while 37.0% delivered beams at the same fixed angle. Of these fixed angle deliveries, the plans were either assessed through composite measurements (60.2%) or field- The vast majority of sites using planar measurements to assess agreement utilize absolute dose (87.6%) as compared to relative dose (16.1%), with almost every site using gamma analysis (99.4%) as the evaluation metric. Very similarly to planar evaluations, the vast majority of sites using 3D/Volumetric measurements to assess agreement utilize absolute dose (90.5%) as compared to relative dose (9.5%), with almost every site using gamma analysis (95.1%) as the test. Table 5 provides the number of participants that used a certain passing criteria for gamma analysis. The loosest criteria was 5%/5 mm with ≥ 90% pixels passing and the tightest criteria was 1%/3 mm with ≥ 97% pixels passing indicating, again, a large variability in passing criteria. Again, the most common criteria used was 3%/3 mm (84.1%). For percentage of pixels passing, 95% (54.2%) and 90% (37.1%) also shared the majority of responses from sites.
If IMRT QA did not pass, we provided nine possible follow-up steps to choose from in our survey. Sites were given the opportunity to rank them on a scale of one to nine, with one denoting the first strategy used. These strategies are ordered in Table 6 according to the order of their average rank (for places employing that strategy).
The highest average rank selection was to remeasure with the same setup, which had an average position ranking of 1.    Looking into the data further, we analyzed the second and third choices selected given a particular first choice. When the most common strategy, remeasure with the same setup, was selected with rank one, the top two most common selections for rank two were to move to a new calculation point and remeasure, followed by replan.
In addition, replan was also the most common selection for rank three. This may indicate that these three strategies tend to be the most commonly used in the medical physics community. Additional strategies and their prevalence are shown in Table 6. Overall, a large majority selected at least one (98.8%) or two (97.0%) strategies, with a substantial fall off occurring at four (46.8%) to five (19.6%) strategies. Seventeen sites did not select any strategies.

| DISCUSSION
Our survey captured many facets of MU verification for 3D-CRT, and IMRT QA. For IMRT QA, our survey evaluated the use of different tools and methods to assess agreement between planned and delivered dose, and explored the strategies used to deal with failing plans. The results of our survey showed that a wide variety of tools and techniques are used for assessing agreement between measured and planned doses. One important note is that while we sometimes specify manufacturers that are used, we do not state any usefulness or correctness in their utilization. We only report these results as a snapshot of the community practice.
As with every survey, any biases must be understood before formulating conclusions. With the structure of the survey, sites were able to answer secondary questions without answering the primary question. For example, a site could select absolute dose or relative dose for a planar technique for agreement assessment, without actually selecting the planar option to begin with. This sometimes caused a discrepancy between our totals; however, these discrepancies were minimal. Another potential bias is the period over which the  (2) 1.0% (12) 0.0% (0) 1.9% (22) The gamma distance-to-agreement ranged from 2 to 5 mm while the gamma dose percent difference ranged from 1 to 5%. Percentage based on number of sites that provided answers for the criteria (N = 1143). *Acceptable criteria based on recommendations for TG-218 T A B L E 5 Response to survey questions: 3d/volumetric analysis secondary questions for passing criteria for gamma analysis.
Gamma distance-to-agreement The gamma distance-to-agreement ranged from 2 to 5 mm while the gamma dose percent difference ranged from 1 to 5%. Percentage based on number of sites that provided answers for the criteria (N = 391). *Acceptable criteria based on recommendations for TG-218. survey was available. IROC-Houston's Facility Questionnaire contains more information than just this survey, and the specifics of what was updated in the Facility Questionnaire was not tracked (just the binary fact that it was updated). We have assumed that when a facility updated their Facility Questionnaire, they did this comprehensively, as is requested of the institutions. However, it is possible that this was not always done, which could lead to some institutional results being descriptive of a previous year. Finally, while the number of survey participants was large, our selectivity of participants in the survey only included sites that were actively participating in NCI-sponsored trials, which focuses primarily on, and is therefore reflective of practice in, the United States and Canada. In addition, it is of note that access and support for QA tools and practices for sites that participate in NCI-sponsored trials could vary greatly with sites that do not participate in these types of trials.
The results of our survey can be compared with best practice recommendations from the AAPM TG-218 report, although it is important to remember that the TG-218 report does not present the only acceptable solution. From our survey, approximately two of three sites use a composite measurement while the remainder utilize a field-by-field measurement. From TG-218, the recommended delivery method for IMRT QA is a "true composite" followed by "perpendicular field-by-field". 18 The use of the perpendicular field-by-field is only recommended when true composite cannot be utilized because of the error-prone nature of field-by-field evaluation. 18 8,9 meaning that if the plan is truly acceptable, IMRT QA devices will claim it is good. Plans that fail IMRT QA should therefore be managed with care. More generally, our survey shows that the strategies to handle failure of criteria is nonuniform, suggesting that the community as a whole is struggling with overcoming QA failures. Improving sensitivity and specificity of QA processes and devices could help alleviate some of the strain caused by these QA failures and create a more uniform approach to IMRT QA. recommendations from the AAPM TG-218 report. The survey found that recommendations for patient-specific IMRT QA were not universally implemented. 17.9% of sites still utilize a perpendicular composite delivery method to test IMRT QA, which is known to hide errors. In addition, a large proportion of sites still utilized looser criteria for all aspects of assessing agreement than what is recommended by TG-218. 18 Furthermore, our survey showed that strategies normally used when IMRT QA does not pass could potentially fall outside of current recommendations and that new strategies should be examined and implemented for better IMRT QA results.