Purpose: Owing to improved palliative treatment options and increased survivorship, a growing number of patients require multiple distinct radiotherapy treatments. This presents unique planning challenges, and it is now commonplace to use image registration and biologically-effective dose (BED) methods for lifetime organ-at-risk dose estimation. However, quality assurance (QA) options are limited, and often involve spot-checks. In this work, we present a simple and universal method and software implementation for QA of dose transformation methods like BED.
Methods: The equivalent dose in 2 Gy fractions (EQD2) paradigm is based on the linear-quadratic cell survival model. How dose is transformed in clinical scenarios depends on the fractionation and α/β, which is tissue- and outcome-specific. Inverting the EQD2 model provides a way to estimate α/β across the entire body, voxel-by-voxel, given the fractionation, EQD2 dose, and original dose. Histograms of α/β-volume can be extracted for individual regions-of-interest and analyzed automatically via percentile extraction, or maps can be visually inspected to survey the entire body quantitatively. Model inversion is sensitive to human transcription and selection errors involving the original fractionation, ensuring appropriate courses were used, α/β settings, selection of appropriate regions-of-interest, and also that the anticipated EQD2 model was selected. A free and open source implementation of EQD2 model inversion is provided. Simulated errors were tested to confirm detectability. Results of being used clinically in a supplementary role to spot-check QA for 21 plans are provided.
Results: Errors resulting from incorrect region-of-interest selection or overlap handling, use of incorrect EQD2 models, incorrect fractionation and course selection can all be detected. Clinically, instances of incorrect α/β were detected for two plans with EQD2 model inversion that were not detected using spot-checks (invalid region-of-interest selection and invalid α/β).
Conclusion: Model inversion is an effective approach for independent QA of BED-based dose transformations.
Not Applicable / None Entered.