Click here to

Session: Advances in Safety [Return to Session]

Using Phantom Results to Quantify Treatment Error Among the Radiation Oncology Community

S Edward*, C Peterson, R Howell, J Pollard-Larkin, S Kry, The University of Texas MD Anderson Cancer Center, Houston, TX


TU-H-BRA-5 (Tuesday, 7/12/2022) 2:45 PM - 3:45 PM [Eastern Time (GMT-4)]

Ballroom A

Purpose: Imaging and Radiation Oncology Core (IROC) phantom results show 8-17% of institutions fail end-to-end radiotherapy tests. This work quantifies the magnitudes of errors in the treatment process, which lead to these discrepancies in accurate treatment delivery.

Methods: IROC lung (n=40), head and neck (H&N) (n=43), and spine (n=63) irradiated phantom results were evaluated for quantifiable errors that occurred during treatment. We examined TPS commissioning errors, by recalculating plans using an independent algorithm with machine specific generic beam models, delivery uncertainties/errors by comparing planned and delivered doses using delivery log files, and beam calibration errors using contemporaneously conducted thermoluminescent dosimeter (TLD) based audits for verification of absolute beam calibration. These identified errors were compared to the phantom dose deviation, which is the overall difference between the institution’s treatment planning system (TPS) and measured in-phantom TLD doses. Phantom dose deviations above TLD measurement uncertainty of 3.2%, were considered substantial and evaluated separately.

Results: Phantom dose deviation results within measurement uncertainty (<3.2%), showed no quantifiable error, whereas those with overall phantom dose deviation >3.2% indicated the presence of all 3 error types. The largest sources of error found, and the correlation of their contribution with absolute phantom dose deviation, were output error, contributing 16% (Pearson correlation; r=0.54, p<.01) to total lung dose deviations, and dose calculation error, contributing 48% (r=0.72, p <.01) to H&N and 40% (r=0.59, p<.01) to spine dose deviations. The total percentage of absolute dose deviations caused by all error types, was 21%, 69% and 71% for lung, H&N, and spine respectively.

Conclusion: We quantified 21%, 69% and 71% of all substantial dose deviations among the lung, H&N, and spine phantom results respectively, with dose calculation and machine output errors being the highest contributors. Knowing these errors will help guide targeted solutions to improving radiotherapy performance.




Not Applicable / None Entered.

Contact Email