Session Item

The Imaging and Radiation Oncology Core (IROC) head and neck (H&N) and lung phantoms, used as an end-to-end test of an institution's radiotherapy treatment process, has an average failure rate of 14%. This work investigates whether delivery log files predict the observed dose deviations in these phantom irradiations.

Delivery log files were collected for IROC lung and H&N phantoms irradiated by various institutions from January 2019 to present. We evaluated these files in Mobius FX to determine the difference between the TPS-calculated and machine-delivered doses (as a perturbation of the TPS). We evaluated whether this perturbed dose more accurately predicted the actual delivered dose, measured by the thermoluminescent dosimeter (TLD). A total of 41 lung phantoms (82 TLDs) and 36 H&N phantoms (216 TLDs) were evaluated. The maximum root-mean-square (RMS) errors in the MLCs, jaws, collimators, and gantry angles during treatment delivery were also analyzed for correlations with delivered dose.

Compared to the average dose deviation (TPS vs TLD) of 2.4% (max = 11.6%) for the H&N phantom, and 2.6% (max = 9.6%) for the lung phantom, the dose perturbations predicted by log files were relatively small. The delivery errors ranged between -0.73% to 1.20% for the H&N phantom. Delivery played an even smaller role for the lung, which had a range of -0.75% to 0.17%. Minimal correlation between dose deviation and delivery error was found: r = .13 (p = .04) for H&N, r = .03 (p = .7) for lung.

Delivered dose was not strongly correlated with jaw, collimator, or gantry angle maximum RMS errors (-0.2 < r < ± 0.2). MLC RMS error results for the two Varian machine classes: Truebeam (including Edge) and the older Base model (iX, EX & Trilogy), showed that the Truebeam machine MLCs performed better among both phantoms. This difference, however, was not great enough to influence phantom irradiation accuracy (Fig. 1).

Analysis of IROC H&N and lung phantom delivery log files showed minimal contributions to the total TLD dose deviations seen among these phantoms. This indicates that few delivery errors are present, or that these errors are not properly captured within the delivery log files. Future work will investigate the ability of this delivery data to catch/identify irradiation errors in IROC phantoms.