Effect of beam profile measurement on arc therapy plan quality assurance: a case study

Abstract We present an example when profile measurement and modeling of an Elekta Agility multileaf collimator (MLC) had a large effect specifically on arc therapy plan quality assurance (QA) results using ArcCheck. ArcCheck absolute dose measurements of these plans were systematically lower than planned by 3–10%. Failing QA results were seen even with unmodulated static and conformal arcs. Furthermore, the effect was found to be dependent on collimator angle, with worse results associated with near‐zero collimator angles. In contrast, step‐and‐shoot QA results were not affected. Changing the beam model to match steeper profile measurements obtained using a different measurement device resolved the problem. This case study demonstrates that conventional gamma index analysis can be sensitive to small profile modeling changes.

Introduction of volumetric modulated arc therapy (VMAT) using 6 MV photons was planned at both locations, and new beam measurements were acquired for modeling. After the TPS change at our Varian site, patient-specific VMAT quality assurance (QA) using Arc-Check (Sun Nuclear Corporation, Melbourne, FL, USA) typically gave gamma index pass percentages in the high 90 s at 3%/3 mm tolerances in absolute dose mode using a 10% low-dose threshold.
At our Elekta site, however, testing showed VMAT QA gamma index pass percentages typically in the 45-85% range. These poor results were attributable to ArcCheck absolute dose measurements being systematically below Pinnacle planned doses (3-10% lower depending on the plan), although relative dose distributions agreed well. The low measured doses were seen only with arc deliveries and were observed for unmodulated conformal and static arcs as well as VMAT. In contrast, 10 step-and-shoot IMRT plans delivered to the ArcCheck yielded acceptable pass rates (93 AE 3%, mean AE 1 SD). Ion chamber measurements at the center of the ArcCheck confirmed that delivered doses for arcs were lower than planned (1-6% lower), though not to the same degree as the Arc-Check measurements (3-10% lower).
Notably, QA pass percentages were correlated with collimator angle. Arcs with collimator angle near 0 (IEC 61217 scales, i.e., leaf motion parallel to gantry motion) exhibited worse pass percentages and larger dose differences than arcs with collimator angle near 90 degrees, where pass rates generally satisfied the 90% standard (Fig. 1).  (Fig. 2). We speculate the difference in profiles obtained with the two instruments was due to different detector sizes contributing to different dose volume averaging. 1 The steeper profiles obtained with the LDA were considered closer to "reality", and we decided to partially remodel the beam on the basis of this older data.

| METHODS
We partially remodeled our beam to match the penumbra and tails measured using the LDA. Only the six "Out of Field" parameters were modified. Initially, the MLC "Offset Calibration" table was modified as well to match the LDA data, as the field edges appeared shifted~0.2 mm outward relative to the CC04 data.
However, the LDA profiles had been acquired along the beam axes, i.e., at the interleaf gap rather than the middle of the leaf, resulting in the apparent field size increase. Radiochromic film measurements of the MLC field edge confirmed a 0.2 mm difference between leaf center and leaf gap. As the vendor recommendation is to measure at leaf center, we ultimately chose not to modify the MLC offset calibration table to match the LDA field sizes.

| RESULTS AND DISCUSSION
ArcCheck results for VMAT using the revised model showed excellent agreement between TPS and measurement ( Fig. 3), while stepand-shoot IMRT results remained as good or better compared to the original model. Prior to remodeling, our mean pass rate was 88 AE 11% (1 SD) for 27 VMAT plans. Most of these plans had a collimator angle of 90°, which had been identified as yielding the highest pass rates. After remodeling, our mean pass rate for 67 VMAT plans was 98 AE 1% with no restriction on collimator angle. Over a 5-month investigation, with vendor support, we had explored many possible causes but had been slow to suspect our beam model for a few reasons. Our model had matched our measured data well, and validation tests for static fields all the way to complex step-andshoot deliveries had all shown excellent agreement. We were hardpressed to explain, on beam modeling grounds, the observed collimator angle effect and the poor results from even barely modulated arcs. Also, the same measurement devices and modeling techniques had been used in a successful implementation of VMAT at another facility. We do not know why we did not encounter the same issue when modeling our Varian machine. We speculate the difference could be because of the use of both jaw-and MLC-shaped fields when modeling the Varian. The Varian has both x-and y-jaws, and its MLC is a tertiary collimator. In contrast, the Elekta MLC is a secondary collimator, and there is no x-jaw.
How can profile measurement and modeling manifest only in arc therapies at certain collimator angles? At near-zero collimator angle, when leaves are oriented along the gantry motion, for any individual ArcCheck detector, dose from the entire penumbra and tail of a leaf tip is seen and integrated at least twice as the leaf gap sweeps across the detector whether the leaves themselves are moving or not (Fig. 4a). Particularly for the small leaf separations characteristic of VMAT, this can result in a systematic overestimation of dose by the TPS if the modeled penumbra and tail are too wide. For example, with the same central dose, the integrated dose difference for a 1 9 1 field was 9% between our two models. F I G . 2. Profiles for 2 9 2, 3 9 3, and 5 9 5 fields acquired with LDA and CC04 ion chamber.
Even for a static arc, where the MLC leaves are not moving, this effect is seen because of the position of the ArcCheck's diode detectors, which are arranged on a cylinder of radius 10.4 cm around the gantry axis. In contrast, for a centrally located ion chamber, the fully-integrated penumbra and tail dose will be seen only if the central leaves dynamically travel over the central axis, leading to our observation that a dose discrepancy may be seen at isocenter, but not of the same magnitude as seen by the more peripheral ArcCheck detectors. For arcs with collimator angle close to 90, leaf motion is perpendicular to gantry motion. Either the side or the tip of leaves may or may not partially or completely sweep a detector, mitigating any consistent, systematic dose difference (Fig. 4b). For step-and-shoot deliveries, a detector will only see penumbra and tail dose if a leaf tip happens to coincide with it, and even then at only one point along the dose profile.

| CONCLUSION
Small differences in beam profile measurement and modeling can have large effects on VMAT QA results under certain, perhaps unexpected, conditions. Conventional gamma index analysis appears sensitive to small differences in profile modeling.

ACKNOWLEDG MENT
We thank Geoffrey Ibbott for reading through the manuscript and for his suggestions.

CONFLI CT OF INTEREST
There are no conflicts of interest.