Helical Tomotherapy DQA with ArcCHECK: Sensitivity to Possible Delivery Errors
A Templeton*, W Sensakovic, J Chu, J Turian, Rush University Medical Center, Chicago, ILSU-E-T-68 Sunday 3:00:00 PM - 6:00:00 PM Room: Exhibit Hall
Purpose: To test the ability of the SunNuclear ArcCHECK QA device to detect errors artificially introduced in a helical Tomotherapy plan.
Methods: Delivery sinograms from clinically-used plans were extracted from the Tomotherapy treatment planning system (TPS), and modified to reflect several delivery errors: leaf pair incorrectly always open, leaf pair always closed, leaves slow to close once open, and desynchronization between leaf pattern and projection number. QA plans were created using the Tomotherapy TPS with the high dose area aligned to the annular region containing the diodes, and delivered to the ArcCHECK device. The position of the device remained constant between deliveries of the altered sinograms. Measured data were compared to calculated dose using the ArcCHECK dose analysis tools with a 3%/3mm ? tolerance.
Results: ArcCHECK was able to detect delivery differences except for peripheral leaves remaining open for the entirety of the procedure (the % agreement for measurement points increased in this case). Altering the sinogram by one projection (~7°) or increasing leaf open time by 50ms resulted in <80% agreement. Increasing leaf open time by 250ms, fully opening or closing the central two leaves, or creating a ~100° gantry starting position error yielded passing rates below 30% and were clearly detectable.
Conclusions: For many years the gamma passing rate has been used as the standard for comparing measured and calculated dose distributions. However, arbitrary shifting of the dose distribution may mask delivery errors, especially in a less intuitive measurement pattern. Having the ability to simulate possible errors should provide the medical physicists with tools to evaluate the sensitivity and response of a particular QA device. Investigation into the types and magnitudes of detectable errors may lead to more robust QA.