3149 - The Impact of the Accuracy of Independent Dose Calculation on Patient-Specific QA for Online Adaptive Radiotherapy
Presenter(s)
G. Tang, W. Donahue, S. C. Lin, K. Episcopia, S. Xing, Y. Yang, M. Savacool, M. Aristophanous, S. Liu, D. K. Mynampati, L. I. Cervino, J. Yang, P. Zhang, Y. C. Hu, and S. B. Lim; Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY
Purpose/Objective(s): To investigate the feasibility of replacing routine patient-specific QA measurements for online adaptive radiation therapy using independent dose calculation software with high dosimetric accuracy.
Materials/Methods: The CBCT-based online adaptive radiotherapy platform at our institution is a closed-loop ecosystem that consists of a ring-gantry linac, online adaptive treatment planning system (OnA_TPS), and a built-in independent dose calculation software that is based on collapsed-cone convolution-superposition (Indp_CCC) algorithms. With the intent of replacing routine patient-specific QA measurements, Indp_CCC was commissioned through a comprehensive and rigorous process as per the recommendations of AAPM-MPPGs 5b, 9b, and TG-219. This involved comparing open field calculations to water tank measurements (field size = 1, 2, 3, 4, 6, 8, 10, 20, 28 cm) and evaluating a basket of IMRT plans across various disease sites and complexities. For a total of 19 plans including prostate-whole pelvis, kidney, and head and neck cases, calculations were compared to film measurements. The gamma criteria used for the IMRT plans was ?(3%,2mm) per AAPM-TG-218 recommendation. The same comparisons were performed for the calculations generated by OnA_TPS and a Monte Carlo based independent dose calculation software (Indp_MC) that is used for non-adaptive treatments in the department. Except for OnA_TPS (which has restricted access for beam modeling), the beam models for both Indp_CCC and Indp_MC were adjusted to best match water tank and IMRT film measurements, which represent the ground truth.
Results: For simple open fields, OnA_TPS, Indp_CCC and Indp_MC fulfilled the recommended requirements from MPPGs 5b and 9b. Compared to film measurements of the IMRT plans, the average gamma passing rate was 99.2% [90.0%-100%], 97.8% [84.3%-99.9], and 99.5% [91.9%-100%] for OnA_TPS, Indp_CCC, and Indp_MC calculations, respectively. While the average gamma passing rate met TG218 recommendation, some of the cases failed with ? <90% for Indp_CCC. Upon inspecting the distribution of dose differences, local dose errors of >10% were observed in both high and low dose regions. Despite optimizing the beam modelling parameters such as the dosimetric leaf gap, Indp_CCC yielded unsatisfactory results for the IMRT plans per AAPM-TGs 218 and 219.
Conclusion: The dedicated built-in independent dose calculation software for an online adaptive radiotherapy system demonstrated inadequate agreement with measurements and was not suitable to replace patient-specific QA measurements. Adoption of an alternative independent dose calculation software was therefore necessary to achieve an acceptable level of accuracy as a measurement surrogate, which strengthens the safety and efficiency of the online adaptive treatment workflow.