Feasibility of an In-House Python Script for Automated Epid-Based Machine Quality Assurance Analysis
Abstract
Purpose
This study evaluates the feasibility of an in-house Python script for analysis of routine quality assurance (QA) procedures that utilize an electronic portal imaging device (EPID) as a low-cost, efficient QA option compared to commercially available software. Demonstration of the implementation of open-source python scripting could prove valuable to medical physics learners and to more financially strained clinics.
Methods
Starshot, Winston–Lutz, and multileaf collimator (MLC) picket fence tests were performed on a Varian TrueBeam using vendor-provided MLC files and portal imaging. Script analysis results were compared with those for current methods. For picket fence tests, manual analysis utilized the measurement tool within Varian Aria offline review while the script analyzed the same exported images. The starshots were analyzed via film with manual analysis in Adobe Acrobat, film with the script, and EPID images with the script. Winston–Lutz images from four photon energies over six months were included.
Results
For Starshot tests, the minimum tangent circle diameter for film and EPID images, respectively, was 0.4±0.1 mm and 0.3±0.1 mm for manual analysis. Analysis using the script yielded 0.1±0.1 mm and 0.1±0.1 mm. Static picket fence results for are 1.5±0.1 cm and 1.5±0.1 cm for the manual and script analysis, respectively. Similarly, for the RapidArc test results are 1.5±0.1 cm and 1.5±0.1 cm. Manual Winston–Lutz test results for 6X, 6X-FFF, 10X, and 10X-FFF are 0.4±0.2 mm, 0.5±0.2 mm, 0.5±0.2 mm, and 0.4±0.2 mm, respectively. Script results agree well with observed values 0.4±0.2 mm, 0.4±0.2 mm, 0.4±0.2 mm, and 0.5±0.2 mm, respectively.
Conclusion
We have demonstrated the reliability of an in-house python script to analyze QA tests captured on an EPID. Expanding use to other QA tests and further streamlining the workflow to accelerate QA procedures is under investigation.