Room: Track 3
Purpose: To determine the role, degree, and predominant parameters of beam modeling errors on IMRT dose delivery to understand the causes of IROC phantom failures.
Methods: We administered a beam modeling parameter survey to all IROC service participants to identify the variance in TPS model development. Next, we performed a sensitivity study of these parameters, based on the variability seen, on IMRT plans using the RayStation and Eclipse treatment planning systems. Finally, we retrospectively evaluated 337 IROC phantom results with concurrent survey responses to evaluate the relationship between atypical beam modeling characteristics (e.g. 10th or 90th percentile) and dose delivery accuracy. IROC head and neck phantom results were also re-evaluated for principal attributes of phantom performance (i.e. setup errors, systematic dose errors).
Results: Survey results based on 2818 clinical beam models indicated that substantial variations exist among parameters representing MLC characteristics (e.g. MLC transmission factor or offset) for machines of the same type, energy, and MLC configuration. For Eclipse, modifying the dosimetric leaf gap (DLG) according to the clinically observed range generated systematic changes up to 6% in the dose calculation. Likewise, varying RayStation’s MLC offset and MLC transmission parameters induced systematic changes up to 11% and 7%, respectively. Other modeling characteristics, based on community reported data, produced changes of ±1%.
In a review of phantom cases with concomitant survey responses, the reported DLG (Eclipse AAA) was significantly correlated with TLD agreement (r=0.337, p<0.0001). Qualitatively, use of an atypical DLG or MLC offset resulted in dose deviations consistent with systematic dosimetric errors – the most common error (69% of phantom failures) observed by IROC.
Conclusion: Dosimetric characteristics of IROC phantom failures are often directly associated with atypical beam modeling, particularly with respect to parameters modeling the MLC offset. This work can help guide the radiotherapy community to improve IMRT accuracy.
Funding Support, Disclosures, and Conflict of Interest: This work is supported by Public Health Service Grants CA180803 and CA214526, awarded by the National Cancer Institute, US Department of Health and Human Services. Glenn is supported by the Rosalie B. Hite and American Legion Auxiliary Fellowships awarded by MD Anderson Cancer Center UTHealth Graduate School of Biomedical Sciences.