Evaluation of an improved picket fence style test for routine MLC positional QA.

Abstract

Purpose: The aim was to develop and evaluate an EPID-based MLC positional test that addresses known weaknesses of the picket fence test and has sufficient accuracy so that the AAPM MPPG 8.b. MLC position action limit of ± 0.5 mm can be implemented.

Methods: Weaknesses and inaccuracies in the picket fence test were identified and a new test plan and analysis algorithm named stakitt was developed. Stakitt was evaluated for repeatability and for sensitivity on the Varian TrueBeam linac with both Millennium MLC and HDMLC and on a Varian Clinac with Millennium MLC. Sensitivity was tested via deliberate introduction of errors into the test plan of magnitude: ± 0.1, ± 0.5, ± 1.0 and ± 1.5 mm. Measured sensitivity was compared to published sensitivity of the picket fence test. Additionally, a use case was presented based on results from a TrueBeam linac to highlight the effect of MLC backlash on MLC positions at non-zero gantry angles.

Results: Repeatability was observed to within 0.04 mm (3 SD) with the TrueBeams being more repeatable than the Clinac. The deliberately introduced errors were accurately measured to within 0.28 mm and were comparable to the traditional picket fence. Reduced accuracy was recorded for the HDMLC small leaves, which was attributed to an apparent variation in measured leaf width across the range of travel, which impacted the measurement of the leaf tip position. The clinical use case demonstrated variability in MLC leaf positions between gantry 90° and gantry 270° that were of the magnitude of the MLC backlash.

Conclusion: The stakitt test addresses the weaknesses of the picket fence test and has accuracy appropriate for implementation of a ± 0.5 mm action limit. However, such an action limit may not be currently practical at non-zero gantry angles due to the impact of MLC leaf backlash.