Benchmarking daily plan adaptation on the Unity MR-Linac
OC-0087
Abstract
Benchmarking daily plan adaptation on the Unity MR-Linac
Authors: Thyrza Jagt1, Tomas Janssen1, Anja Betgen1, Lisa Wiersema1, Rick Verhage1, Sanne Garritsen1, Tineke Vijlbrief-Bosman1, Peter de Ruiter1, Femke Peters1, Peter Remeijer1, Corrie Marijnen1, Jan-Jakob Sonke1
1Netherlands Cancer Institute, Radiation Oncology, Amsterdam, The Netherlands
Show Affiliations
Hide Affiliations
Purpose or Objective
The Unity MR-Linac offers an adapt-to-shape (ATS) workflow for daily plan adaptation. This workflow entails a full plan re-optimization on the daily MRI using new delineations. The adaptation optimizes the dose distribution using the objective functions of the reference plan generated on the CT scan. To reduce calculation times we do not manually tweak the objective functions in the daily ATS workflow. The question arises whether the reference plan quality is reproduced in the adapted plan when using the reference plan objectives without any manual tweaking. This study therefore benchmarks the daily ATS workflow against a daily automated full plan optimization.
Material and Methods
Eight rectal cancer patients were included in this study with 5 daily MRIs per patient with a total of 40 daily MRIs. A reference MR-Linac plan was generated for all patients prescribing 25Gy in 5 fractions to the rectum and lymph nodes using anisotropic CTV-PTV margins (5-8 mm). The CTVs and main OAR were redelineated on all daily MRIs.
The research Elekta-mCycle treatment planning system allows for automated plan optimization based on a wishlist holding planning constraints and objectives. A wishlist was designed such that the mCycle plans mimic the clinical reference plans.
For each daily MRI, two plans were generated:
- ATS: The conventional Monaco objective functions used to obtain the reference plan on the planning CT scan were used to optimize a new plan on the daily MRI.
- mCycle: Using the wishlist, fully optimized new plans were automatically generated on the daily MRIs.
Results
Median differences between the mCycle plans generated on the planning CT scans and the clinical reference plans were less than 1Gy (Figure 1). The largest interquartile range (2.23Gy) was seen for the lumbar skin D2, most others were below 1Gy. This shows that the mCycle plans successfully mimic the clinical treatment plans and mCycle can hence be used to benchmark the ATS workflow.
Figure 2 shows the obtained values for the daily ATS and mCycle plans. 27/40 of the ATS plans and 40/40 of the mCycle plans obtained a PTV D98>95%. The daily criteria PTV D98_daily > D98_reference-0.5Gy was achieved for 39/40 and 40/40 plans for ATS and mCycle, respectively.
The Dmean and D2 values of the bowelarea+bladder were similar for the ATS plans and the mCycle plans. Similar to the results in Figure 1, the lumbar skin was spared slightly better in the fully optimized mCycle plans than in the ATS plans.
Overall the differences in OAR doses between ATS and mCycle plans were similar to those observed on the planning CT scans, suggesting that they primarily originate from differences between the reference plans and the wishlist rather than the quality of the ATS workflow.
Conclusion
This study shows that the relatively simple ATS workflow can robustly perform high quality adaptations, maintaining the quality of the reference plan throughout the treatment fractions for rectal cancer MR guided radiotherapy.