Automation of HU-RED Calibration: Development, Validation, and Monthly Consistency Testing in Radiotherapy

Abstract

Radiotherapy delivers high doses of radiation to tumors, requiring careful optimization and quality management to maximize tumor control and minimize normal tissue damage. Quality Assurance (QA) protocols led by certified medical physicists are essential to maintain treatment accuracy within tight tolerances and prevent errors. In this study, we developed and validated an automated quality management procedure for the monthly calibration of Computed Tomography (CT) Hounsfield Unit to Relative Electron Density (HU-RED) conversion, as a case example of improving dose calculation accuracy in radiotherapy treatment planning through quality assurance optimization. A novel Python-based algorithm was implemented to extract CT calibration curves from phantom scans automatically, reducing human error and improving efficiency. The algorithm was tested on three different multi-material phantoms and compared its performance to manual analysis. The automated method significantly improved the accuracy of HU measurements and reproducibility of calibration, while reducing analysis time by over 90% compared to manual methods. All differences between automated and manual results were within 5 HU and statistically insignificant in terms of resulting electron density, with Pearson correlation R ≈ 0.999. Using the automated procedure, we established a monthly constancy test for CT calibration. Over 12 months of routine use, no significant drifts in the HU-RED relationship were detected, confirming the stability of the CT scanner’s calibration within ±1-2% dose impact following the automated method. These results demonstrate how a systematic automated QA program ensures consistent dose calculation accuracy, aligning with international recommendations for radiotherapy safety.