ESTRO 2025 - Abstract Book

S2773

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

ESTRO 2025

of the heart were significantly lower in the robust method ( P < 0.001) compared to all the plans except the flash regarding the maximum dose, V 20Gy , and V 30Gy ( P > 0.310). The flash+MCO(O) had significantly lower left lung V 20Gy and V 30Gy , as well as right lung V 5Gy and right breast V 3Gy . Regarding the left and right lung mean doses, the flash+MCO(O) method had lower values ( P ˂ 0.016) except the robust method ( P > 0.051). The monitor unit values were significantly lower in the robust+MCO(P) method ( P < 0.001). Conclusion: Both of the robust+MCO treatment plans had higher PTV coverage. The flash+MCO(O) showed higher OARs sparing (except heart) compared to other plans, and regarding heart sparing, robust optimization can be considered a more appropriate VMAT optimization technique for patients having left-sided breast cancer.

Keywords: Robust optimization, Multi criteria optimization

References: [1] Craft D, Good M. A guide to using multi-criteria optimization (MCO) for IMRT planning in RayStation. Massachusetts General Hospital: Department of Radiation Oncology. 2011.

[2] Chan RC, Ng CK, Hung RH, Li YT, Tam YT, Wong BY, Yu JC, Leung VW. Comparative study of plan robustness for breast radiotherapy: Volumetric modulated arc therapy plans with robust optimization versus manual flash approach. Diagnostics. 2023 Nov 7;13(22):3395.

1596

Poster Discussion Adding degrees of freedom to oropharyngeal cancer treatment planning: is re-training of automated planning models necessary? Eva Onjukka 1,2 , Olubunmi Olumuyiwa 3 , Lars Södergren 1 , Julia Söderström 1 , Mimmi-Caroline Bolin 1 , Marianne Falk 1,2 , Anna Embring 4,2 , Vanessa Panettieri 5,6,7 1 Department of Nuclear Medicine and Medical Physics, Karolinska University Hospital, Stockholm, Sweden. 2 Department of Oncology Pathology, Karolinska Institutet, Stockholm, Sweden. 3 Department of Physics, Stockholm University, Stockholm, Sweden. 4 Department of Radiotherapy, Karolinska University Hospital, Stockholm, Sweden. 5 Department of Physical Sciences, Peter MacCallum Cancer Centre, Melbourne, Australia. 6 Sir Peter MacCallum Department of Oncology, University of Melbourne, Melbourne, Australia. 7 Central Clinical School, Monash University, Melbourne, Australia Purpose/Objective: Head and neck cancer (HNC) treatment has greatly improved with the introduction of intensity-modulated radiotherapy. However, the algorithms used for optimisation are based on methodology and technology implemented more than a decade ago [1]; these might not exploit the potential of modern delivery machines. A novel solution (RapidArc Dynamic-RAD) (Varian Medical Systems, Palo Alto, USA) was recently introduced to combine the conformality of VMAT with the modulation of IMRT-like static angle modulated ports, including the use of a dynamic collimator. This solution has the potential to further sculpt the dose for complex anatomical areas like the head and neck. However, clinics that rely heavily on established auto-planning models may hesitate to make the transition due to difference in optimisation nature. We compare the performance of RAD planning for oropharyngeal cancer (OPC) optimised with 1) a currently clinical auto-planning model trained with VMAT, and 2) a model re-trained with RAD. Material/Methods: Fifty validation OPC patients (68 Gy in 34 fractions) were retrospectively selected. These were initially planned with 2-arc VMAT using RapidPlan (RP-VMAT) trained on 111 historical cases of HNC. These training patients were re-