ESTRO 2025 - Abstract Book

S4391

RTT - Treatment planning, OAR and target definitions

ESTRO 2025

Purpose/Objective: To investigate if autogenerated radiotherapy treatment plans for prostate cancer treatments are clinically acceptable, time saving, and equivalent to manually created treatment plans. Material/Methods: A prospective study including 100 patients with prostate cancer was performed. Autogenerated treatment plans were compared with plans made manually. The autoplans were created fully automatically outside of working hours without human input using scripting capabilities in Eclipse and with the help of RapidPlan. These plans were reviewed by a treatment planner and reoptimized if necessary. The manual plans were created by an independent treatment planner. Both plans were reviewed by oncologists and medical physicists in a blind chart round to select plan for treatment. The time to make manual plans from scratch was estimated for 21 independent patient cases. The time to make the manual plan versus evaluating the autoplan (including eventual reoptimization) was compared. Dose-volume metrics for targets and OARs were also compared. Results: The average time ± 1 standard deviation to evaluate an autoplan without reoptimizing was 6 minutes ± 1.5 (n=30) versus 19 minutes ± 7.5 (n=70) when reoptimizing was needed (Figure 1). The major reason for reoptimization was that the RapidPlan model had a tendence to press too much on the dose to the OARs resulting in a loss of target coverage. This can, however, be adjusted in the model to mitigate the number of reoptimizations in the future. The average time ± 1 standard deviation to create a plan manually was 55 minutes ± 44 (Figure 1). The autoplans were deemed clinically acceptable in 89% (n=89) of the cases and the manual plans in 75% (n=75). For 7 patients (7%) none of the plans were deemed clinically acceptable during the chart round. The autogenerated plans were blindly chosen for treatment in 63% (n=63) of the cases (Figure 2). Of these cases, 65% (n=41) presented a higher PTV D 98% and 78% (n=49) had a lower V 90% in Rectum compared to the plans made manually.

Conclusion: Time can be saved using autogeneration of treatment plans. Thanks to the fully automated feature of the script, time can be saved even if the autoplans need to be reoptimized before the chart round. Furthermore, in a majority of the cases studied, the target dose coverage was superior for the autoplans at the same time as the dose to rectum was lower. This led to a higher number of autoplans being chosen for treatment.

Keywords: Autoplanning, Treatment Planning