S1976
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
Digital Poster Highlight 4045
Automated planning as a tool to reduce dose to organs-of-interest in oesophageal radiotherapy: how low can you go? Geraint J Lewis 1,2 , Sarah Gwynne 3,2 , Jonathan Helbrow 4,2 , Sarah Bridges 5 , Lisette Nixon 5 , Tom Crosby 6 , Philip Wheeler 1 1 Radiotherapy Physics, Velindre University NHS Trust, Cardiff, United Kingdom. 2 National RTQA, Radiotherapy Trials Quality Assurance (RTTQA) group, London, United Kingdom. 3 Clinical Oncology, South- West Wales Cancer Centre, Swansea, United Kingdom. 4 Clinical Oncology, Gloucestershire Hospitals NHS Foundation Trust, Cheltenham, United Kingdom. 5 Centre for Trials Research, Cardiff University, Cardiff, United Kingdom. 6 Clinical Oncology, Velindre University NHS Trust, Cardiff, United Kingdom Purpose/Objective: The purpose of this study was to calibrate and clinically-validate the use of a protocol-based automated iterative optimisation (PB-AIO) planning solution for use in the treatment of oesophageal carcinoma, using a sample of patients from SCOPE2: a UK phase 2/3 randomised-controlled trial with a prescription of 50Gy and 60Gy in 25 fractions as a comparator. Material/Methods: A site-specific ‘autoplan’ protocol was created and used to generate automated plans for 108 patients, which were compared against the plans originally- submitted to the trial. Analysis consisted of a Wilcoxon signed-rank test of paired samples. Clinical and automated plans for 10 patients were subject to blinded review by two experienced Consultant Clinical Oncologists. Results: The autoplan demonstrates improvements in V95% for PTV_5000 (p<0.001) and PTV_6000 (p<0.001) in the 60Gy/25# arm, and for PTV_5000 in the 50Gy/25# arm (p<0.001). Automated planning also showed significant improvements in the Paddick conformity index for PTV_5000 in both the 50Gy/25# (p=0.04) and 60Gy/25# (p<0.001) arms, while no significant difference in conformality was observed for PTV_6000 (p=0.1).
Point doses calculated on CBCT images differed from reference by 0.357-1.286%.Percentage difference of mean doses in VOIs from reference ranged 0.14 - 7.14 (head); 0.21 - 1.14 (thorax); 0.29 - 1.50 (pelvis) and 0 - 1.44 (head); 0.21 - 0.56 (thorax); 0 – 2 (pelvis) for CBCT scans aquired with aSi-1200 and Hypersight respectively. Assignment of thorax calibration curve to head and pelvis CBCT images yielded differences ranging 0.29 - 1.3, 0.34 - 1.85 and 0 - 1.17, 0.21 - 1.64 for aSi -1200 and HyperSight respectively.Pass-rate results of 3D gamma dose distribution comparison are presented in Tab. 1. Conclusion: Thorax calibration curve can be assigned universally to all CBCT modalities for both aSi-1200 and HyperSight detector as it matches CT-curve best and yeilds calculated dose distributions sufficiently accurate (difference < 2% for both point dose and mean dose in VOI and 2%, 2mm local 3D gamma pass rate > 99.3 for open and modulated field).CBCT images aquired with aSi-1200 detector provide comparable agreement with reference doses to HyperSight hence both can be utilized in emergency treatment planning and treatment dose accumulation control. References: 1. Duan J. et al., Assessing HyperSight iterative iterative CBCT for dose calculation in online adaptive radiotherapy for pelvis and breast patients compared to syntethic CT, J Appl Clin Med Phys. 2025;2. Wessels C. et al., Technical note: Phantom-based evaluation of CBCT dose calculatioin accuracy for use in adaptive radiotherapy, Med. Phys, 2024;3. Thomsen S.N. et al., Daily CBCT-based dose calculations for enhancing the safety of dose-escalation in lung cancer radiotherapy, Radiotherapy and Oncology, 2024;4. Gregg K. et al., Hounsfield Unit characterization and dose calculation on a C-arm linac with novel on-board cone-beam computed tomography feature and advanced reconstruction algorithms, J Appl Ckin Med Phys., 2025; Keywords: CBCT, dose calculation, HU-RED calibration curve
Made with FlippingBook - Share PDF online