ESTRO 2025 - Abstract Book

S3512

Physics - Optimisation, algorithms and applications for ion beam treatment planning

ESTRO 2025

2812

Mini-Oral Coping with CT number inaccuracies in CBCT-based online-adaptive IMPT planning Michelle Oud 1,2 , Sebastiaan Breedveld 1 , Kelvin Ng Wei Siang 1,2 , Roberto Cassetta 3 , Steven Habraken 4,2 , Zoltán Perkó 5 , Ben Heijmen 1 , Mischa Hoogeman 1,2 1 Department of Radiation Oncology, Erasmus MC Cancer Institute, Rotterdam, Netherlands. 2 Department of Medical Physics & Informatics, HollandPTC, Delft, Netherlands. 3 Varian Proton Systems, Varian, a Siemens Healthineers company, Baden, Switzerland. 4 Department of Radiation Oncology, Leiden University Medical Center, Leiden, Netherlands. 5 Department of Radiation Science and Technology, Delft University of Technology, Delft, Netherlands Purpose/Objective: CBCT-based online-adaptive proton therapy is challenging because of CT number errors in CBCTs, even with modern correction approaches. We propose and evaluate an approach for coping with CT number uncertainties by enlarging Range Robustness Settings (RRS) in daily CBCT-based online-adaptive planning. We compared this approach to our trigger-based offline (TB-Offline) adaptive approach, and to diagnostic in-room CT-on-rails-based online-adaptive planning. Material/Methods: For 23 head-and-neck cancer patients, one CT-on-rails and iteratively reconstructed CBCT were acquired in a single fraction within 3 minutes, using a single immobilized patient setup. CT-on-rails contours were rigidly propagated to the CBCT. Five treatment plans were generated on each CBCT: with 3, 6, 8, 10, and 12% RRS, each with 1 mm setup-RS. Subsequently, these plans were recomputed on the corresponding ground-truth CT-on-rails (forward dose computation). These dose distributions were compared to 1) the treatment plan resulting from the TB-Offline approach generated on a regular planning-CT or repeat-CT (3 mm/3%), also recomputed on the CT-on-rails, 2) the treatment plan generated directly on the CT-on-rails (1 mm/3%), serving as reference for the maximum dosimetric potential with online-adaptive IMPT. CTV coverage (voxelwise-minimum) and grade ≥II normal tissue complication probabilities were compared between strategies. Results: Mean absolute errors in CBCT CT-numbers were 145 ± 27 HU (± SD over the patients) compared to the CT-on-rails. From RRS=3% to RRS=10%, the population 90th percentiles of CTV 7000 V 94% improved from 92.8% to 96.4%, and CTV 5425 V 94% from 89.6% to 96.4% (Fig.1). Large coverage losses (V 94% <95%) with CBCT-based online-adaptive and RRS=10% were observed in 1/23 evaluated patient-fractions for CTV 7000 and 2/23 for CTV 5425 (Fig.1). This was an improvement compared to 3/23 and 4/23 with TB-Offline. For RRS=10%, the average risk of xerostomia was reduced with 2.4 ± 1.7%-point (average ± SD) compared to TB Offline, while differences in risk of dysphagia were non-significant. Xerostomia reduction was 3.6 ± 1.8%-point using the treatment plan generated directly on the CT-on-rails (Fig.2). This means that with CBCT-based online-adaptive planning using RRS=10% (xerostomia reduction 2.4%-point), 67% of the potential maximum xerostomia reduction was achieved.