Use of plan complexity metrics for quality assurance in SBRT radiotherapy trials
Rushil Patel,
United Kingdom
MO-0306
Abstract
Use of plan complexity metrics for quality assurance in SBRT radiotherapy trials
Authors: Rushil Patel1, Angus Main2, Patricia Díez1, Catharine Clark3,4,5
1National Radiotherapy Trials Quality Assurance Group, Mount Vernon Cancer Centre, Radiotherapy Physics, London, United Kingdom; 2UCL, Department of Medical Physics and Biomedical Engineering, London, United Kingdom; 3National Radiotherapy Trials Quality Assurance Group, UCLH, Radiotherapy Physics, London, United Kingdom; 4UCLH NHS Foundation Trust, Radiotherapy Physics, London, United Kingdom; 5National Physical Laboratory, Metrology for Medical Physics, London, United Kingdom
Show Affiliations
Hide Affiliations
Purpose or Objective
Complexity metrics (CM) can be used to quantify the complexity of radiotherapy treatment plans. As part of a national program for the implementation of spinal SBRT, accreditation was required for centres prior to treating clinically. The quality of spine SBRT planning was assessed with a benchmark case and safe and accurate delivery ensured through an external dosimetry audit. A pre-outlined benchmark case had to be planned to specifications set in the UK SABR Consortium Guidelines. Once approved, a treatment plan was created for an anthropomorphic spine phantom, which would be delivered in an end-to-end dosimetry audit. This study aims to use CMs to evaluate whether the level of complexity is comparable between benchmark and audit plans produced at each centre and to assess whether certain metrics could be used in the QA process to flag outliers and plans which could perform poorly in the dosimetry audit.
Material and Methods
29 UK centres’ benchmark and dosimetry audit plans were analysed using PlanAnalyzer, software developed by Hernandez et al. Some centres completed QA using multiple TPSs; in total 26 Eclipse, 13 Pinnacle, 9 Monaco, 9 Raystation and 1 Oncentra plans were evaluated. CMs were computed and five were selected for comparison: plan monitor units (MU), Modulation Complexity Score (MCS), Gantry Speed Modulation (GSM), Median Dosimetric Leaf Gap (MDLG) and Tongue and Groove Index (TGI). A Wilcoxon sign rank test was performed to test whether the two groups (benchmark and audit plans) were statistically different.
For Eclipse users, a comparison was made of CMs against alanine and film dosimetry results, to determine any correlation.
Results
A comparison of MU, MCS, GSM, MDLG and TGI for benchmark and audit plans is displayed in Figure 1 (NB- 3 centres changed TPS between benchmark submission and audit). A Wilcoxon sign rank test only showed statistical significance for MDLG (p-values were p=0.649, p=0.524, p=0.668, p<0.001 and p=0.309 for MU, MCS, GSM, MDLG and TGI, respectively).
For Eclipse users only TGI correlated with the dosimetry audit data, where a higher TGI gave poorer QA results (see Figure 2). The Pearson correlation coefficients of alanine and film dose vs TGI were r = 0.491 and 0.564 respectively.
Conclusion
In general, the complexity of benchmark and audit plans was found to be comparable, with no statistically significant difference in complexity observed for 4 of the 5 CMs evaluated. MDLG was significantly higher in the benchmark plans, which may be attributed to the larger target volume in this plan. This indicates the audit programme is testing plans of comparable complexity to the clinical benchmark case.
TGI showed moderate correlation with the dosimetry audit results. This may be used to assess the dosimetry audit plan before measurements are undertaken to ensure that potential outliers are identified before audit measurements are undertaken.