Quality assurance of volumetric modulated arc therapy: evaluation and comparison of different dosimetric systems.

PURPOSE: To compare and evaluate different dosimetric techniques and devices for the QA of VMAT plans created by two treatment planning systems (TPSs).
METHODS: A total of 50 VMAT plans were optimized for treatment of anatomical sites of various complexities by two TPSs which use rather different approaches to VMAT optimization. Dosimetric plan verifications were performed both as part of commissioning and as patient specific QA of clinical treatments. Absolute point doses were measured for all plans by a micro ion chamber inserted in a dedicated water-filled cylindrical phantom. Delivered dose distributions were verified by four techniques based on different detectors: radiographic and gafchromic films, two systems based on 2D diode arrays and an ion chamber array. Gamma index analysis with various tolerance levels (3%, 3 mm and 3%, 2 mm) was used to analyze differences between calculated and delivered doses. Sensitivity to possible delivery errors was also evaluated for three of the considered devices introducing +/-3 mm shifts along the three directions and a 3 degrees gantry offset.
RESULTS: Ion chamber measured point doses were within 3% of calculated ones for 48 out of 50 values. For delivered dose distribution, the average fraction of passed gamma values using 3% and 3 mm criteria was above 95% for both TPSs and all detectors except gafchromic film which yielded on average of 91.4%. For 49 out of 50 plans, a pass-rate above 94% was obtained by at least one of the four techniques. Shrinking the tolerance to 3% and 2 mm, the average pass-rate by all detectors (except film) was still above 95% for one of the two TPSs, but lower for the other one. The detector sensitivity to 3 mm shifts and to gantry angle offset was strongly plan and partially detector dependent: the obtained pass-rate reduction ranged from 2% to 30%.
CONCLUSIONS: The presented results for VMAT plans QA assess the reliability of the delivered doses for both TPSs. The slightly lower pass-rate obtained for one of the considered TPS can be attributed to a higher level of complexity of the optimized plans. The results by different dosimetric techniques are coherent, apart from a few measurements by gafchromic films. The detector sensitivity to delivery errors, being strongly plan dependent, is not easy to evaluate.