ESTRO 2024 - Abstract Book

S3612

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2024

The performance of each plan was evaluated using a probabilistic evaluator, also implemented in RayStation via the integrated python scripting interface. Monte Carlo simulations of 1000 courses were performed for each plan (with sampled systematic error for each course and random error per fraction), using the static dose cloud assumption. For each case, the 90th centile for each clinical goal was calculated and compared between the 3 plans.

Results:

Total combined systematic and random geometric uncertainties of 1.2 and 1.5 mm were found from the audit, confirming the clinically used PTV margin of 4 mm and validating the use of 4mm uncertainty in the robust optimisation. Figure 1 shows the relative change for robust and probabilistic compared with margin plans for the salient dosimetric parameters of interest. Both robust and probabilistic plans maintained the same CTV D95% coverage as margin plans under uncertainty evaluation, but with significantly reduced total V95%, V50% and mean brain dose. Probabilistic plans also demonstrated a significant improvement in dose to CTV in the region of overlap with OARs that was not evident for robust plans. However, probabilistic plans did not always achieve OAR sparing at the required 90% confidence (OAR tolerance was exceeded for 4 of 42 overlapping OARs in probabilistic plans, whereas all OAR tolerances were maintained for margin and robust plans). Figure 2 shows the irradiated V95% for each plan type against PTV volume. This demonstrates that a large proportion of the reduction in V95% for Robust and Probabilistic plans can be attributed to the non-conformality of the 95% isodose in the margin plan with the PTV; however even if the margin plans were completely conformal, there would still be an advantage to both alternative plan types in terms of achieving a smaller irradiated volume.

Figure 1: Box plot comparison of robust and probabilistic (“Prob”) plans with clinical margin plan for salient dosimetric parameters. Each data point shows the relative change in the trial plan against the clinical margin plan for a given patient (90th centile of 1000 simulations per plan).