ESTRO 2024 - Abstract Book

S3648

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

ESTRO 2024

2312

Digital Poster

Fully automated dose planning with Eclipse scripting API

Thomas Henry 1,2 , Elin Svensson 1,2

1 Region Västra Götaland, Sahlgrenska University Hospital, Dept of Medical Physics and Biomedical Engineering, Gothenburg, Sweden. 2 Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Department of Medical Radiation Sciences, Gothenburg, Sweden

Purpose/Objective:

To automate the dose planning process without human input, producing a clinically acceptable treatment plan in just a few minutes.

Material/Methods:

The scripting capabilities in the TPS Eclipse together with SQL requests towards the Aria database were used to perform all the necessary tasks to ultimately produce a treatment plan. The following tasks, but not only, were included: fetching the list of patients waiting for a treatment plan and the prescribed dose, creating an optimization structure set and adequate optimization structures, creating and optimizing a dose plan and calculating the final dose distribution. This process was applied for rectal, prostate, and palliative treatments. For rectal and prostate patients, RapidPlan models were created and used for optimization. For palliative patients, fixed optimization criterias were used. The RapidPlan models for rectal and prostate were based on 62 and 60 patient plans treated at our hospital respectively. Automatically generated plans were compared to clinically accepted human made plans for 17 rectal patients and 20 prostate patients. The script, Rapidplan models and all tests were performed in a research version of Aria/Eclipse. The script was thereafter tested on a couple patients in our clinical version of Aria/Eclipse. The automatically generated plans gave similar dose distributions compared to the clinical plans but in a shorter period time. It took the script around 8-15 minutes to create a prostate or a rectal dose plan, without Multi Criteria Optimization calculation, which is three to five times faster than the time required for a dose planner to perform the same task for a standard dose plan. For rectal plans dose coverage was in average slightly better for the 17 automatically generated plans compared to the clinical plans done by a dose planner, PTV D 98% = 96.7 ± 0.4% vs 95.7 ± 0.5%. All bladder endpoints considered (V 21Gy , V 25Gy and mean dose) were within ±2 % of the clinical plans. The bowel doses were lower with the automatically generated plans for both V 10Gy (252.0 ± 166.4 cc vs 258.6 ± 158.2 cc) and V 18Gy (115.2 ± 84.9 cc vs 120.6 ± 86.0 cc) at the expense of the femoral head mean dose (12.9 ± 3.4 Gy vs 11.8 ± 3.0 Gy). V 23Gy for bowel was slightly higher 80 ± 66.8 cc vs 78 ± 64 cc due to the better coverage. Finally, the automatically generated plans tended to be in average a little more complex with a higher total number of Monitor Units, 1360 ± 226 MU vs 1130 ± 138 MU. Results: