Program Information
An Open-Source Software Analysis Library for Linear Accelerator Quality Assurance
J Kerns1*, D Yaldo2 , (1) UT MD Anderson Cancer Center, Houston, TX, (2) Advocate Health Care, Park Ridge, IL
Presentations
SU-G-BRB-2 (Sunday, July 31, 2016) 4:00 PM - 6:00 PM Room: Ballroom B
Purpose: Routine linac quality assurance (QA) tests have become complex enough to require automation of most test analyses. A new data analysis software library was built that allows physicists to automate routine linear accelerator quality assurance tests. The package is open source, code tested, and benchmarked.
Methods: Images and data were generated on a TrueBeam linac for the following routine QA tests: VMAT, starshot, CBCT, machine logs, Winston Lutz, and picket fence. The analysis library was built using the general programming language Python. Each test was analyzed with the library algorithms and compared to manual measurements taken at the time of acquisition.
Results: VMAT QA results agreed within 0.1% between the library and manual measurements. Machine logs (dynalogs & trajectory logs) were successfully parsed; mechanical axis positions were verified for accuracy and MLC fluence agreed well with EPID measurements. CBCT QA measurements were within 10 HU and 0.2mm where applicable. Winston Lutz isocenter size measurements were within 0.2mm of TrueBeam’s Machine Performance Check. Starshot analysis was within 0.2mm of the Winston Lutz results for the same conditions. Picket fence images with and without a known error showed that the library was capable of detecting MLC offsets within 0.02mm.
Conclusion: A new routine QA software library has been benchmarked and is available for use by the community. The library is open-source and extensible for use in larger systems.
Contact Email: