S2385
Physics - Quality assurance and auditing
ESTRO 2026
Melbourne, Australia. 5 Imaging and Radiation Oncology Core, MD Anderson Cancer Center, Houston, USA. 6 Australian Clinical Dosimetry Service, Australian Radiation Protection and Nuclear Safety Agency, Melbourne, Australia. 7 Department of Radiation Oncology, King Khalid University Medical City, Abha, Saudi Arabia. 8 Department of Radiation Oncology, St George Cancer Care Centre, Kogarah, Australia. 9 Centre for Medical Radiation Physics, University of Wollongong, Wollongong, Australia. 10 Metrology for Medical Physics Centre, National Physical Laboratory, Teddington, United Kingdom. 11 Department of Medical Physics and Bioengineering, University College London, London, United Kingdom Purpose/Objective: To identify those beam model and treatment delivery parameters that most impact single-isocentre multi- target SRS dose distributions. Material/Methods: A dataset of single-isocentre multi-target SRS treatment plans from seven patients with 2–9 targets was used. For each patient, four plans were developed with different optimisation techniques in the Eclipse treatment planning system using a Varian TrueBeam linac with Millennium MLC and a 1mm GTV-PTV expansion. The four optimization modes were (i) the original (OG) plan as used for treatment with the Eclipse aperture shape controller set to medium, (ii) the Eclipse aperture shape controller changed to very high (ASC), (iii) monitor units limited to 70% of the original plan (70MU), and (iv) both the aperture shape controller set to very high and the 70% MU limit (70MU_ASC). All 28 plans met all clinical constraints and dose sparing limits. Each plan was then recalculated with 53 different beam model errors or treatment delivery errors at clinically realistic magnitudes (Figure 1). This included errors in collimator angle, gantry angle, setup position, MLC position, treatment planning system model parameters (MLC offset, MLC transmission), and machine output. Changes in DVH parameters for targets and OAR were assessed.
Figure 1. Descriptions of introduced errors. *Several magnitudes included Results: The robustness of individual plans to introduced errors varied dramatically based on error mode and on the specific plan. For example, the change in PTV D99% for a 1-degree couch rotation ranged from +1.6% to -19.3% (Figure 2). This was the case with most errors where some plans experienced minimal dose impact while others had a significant reduction in target coverage. However, significant difference between optimisation modes (p<0.05) was only found for errors in MLC position, MLC offset and MLC transmission, where the 70MU_ASC plans showed the smallest mean reduction in PTV D99%. The most impactful errors were found to be changes in MLC position with all plans experiencing some effect on target coverage, followed by large changes in couch and collimator angle.
Made with FlippingBook - Share PDF online