S2244
Physics - Intra-fraction motion management and real-time adaptive radiotherapy
ESTRO 2026
each to the appropriate instantaneous geometry and mapping via the inverted DVF to accumulate motion corrected dose on the planning CT. Dose statistics (D95%, D5%) were computed for the GTV, PTV and sub-PTVs to examine sup-inf, ant-post and left-right biases in dosimetric impact. Results: PTV D95% showed a systematic effect due to max- exhale dose planning, with mean reduction across patients of 24.3% prescription dose, biased towards underdose inferiorly and overdose superiorly. Left- right and anterior-posterior effects were observed to be patient specific, consistent for a given patient, and occasionally strong (range 0-18% prescription dose), but largely independent of target size or location.
Proffered Paper 3074
Dosimetric impact of motion and max-exhale planning on liver SBRT delivery, via intrafraction volumetric motion prediction from planar-kV images Sean J Rooney, Michael A Biddlecombe, Marcus Tyyger, Bashar Al-Qaisieh, Michael G Nix Radiotherapy Physics, Leeds Teaching Hospitals NHS Trust, Leeds, United Kingdom Purpose/Objective: Liver SBRT is an effective treatment option, but technically challenging to deliver due to (primarily respiratory) motion during treatment. Various mitigations are common, including abdominal compression and max-exhale dose planning. However, motion remain significant and unquantified during conventional RT dose delivery. We used deep-learning motion predictions, based on kV planar imaging, to estimate in-treatment motion as instantaneous deformation vector fields (DVFs) with 182ms time resolution. Motion compensated dose (MCD) was computed using these DVFs to yield an estimate of delivered dose and examine the impact of motion. Material/Methods: 4 retrospective liver SBRT patients with planning 4DCT and 10 sets of kV projection images (from pre- and post- treatment CBCT) were used for MCD. Deep- learning motion-models, trained on a subset of this data, predicted instantaneous DVFs from kV images.
Interplay effects with VMAT segments were minimal, likely due to the plan covering numerous breathing cycles (> 20) and the open segment nature of the SBRT plans. Specifics of patient breathing and max-exhale planning dominated observed differences. Conclusion: Motion-compensated RT dose distributions, based on motion predicted from kV planar imaging were presented for the first time. Max-exhale planning for liver SBRT resulted in anticipated gross superior overdoses and inferior underdoses. Patient-specific effects were observed axially. Impact on OARs will depend on specific target location and patient breathing pattern, suggesting max-exhale planning may sometimes be suboptimal, particularly where OARs are not inferior to target, or significant lateral breathing motion components exist.This work can inform liver SBRT planning and on- or off-line adaptive strategies. References: Sharma, M., et al. Radiotherapy & Oncology 2022, Jan; 10.1016/j.radonc.2021.11.022Poulsen PR et al. Int J Radiat Oncol Biol Phys. 2014 Jun;10.1016/j.radonc.2014.05.007.Wu QJ et al. Med Phys. 2008 Apr; 10.1118/1.2839095 Xu Z et al. Phys Med. Sci Rep 15. 2025; 10.1038/s41598-025-19218- wHardcastle N et al. Phys and Imag in Radiat Oncol. 2023 Jan; 10.1016/j.phro.2022.12.004Zakeri, A et al. Comput Methods Programs Biomed. 2024 Jun;
10 synthetic GTVs were generated per- patient, using a statistical shape model to ensure plausible target shapes, sizes and locations, and wereplanned as VMAT SBRT.MCD was computed (figure 1) by splitting VMAT plans into ~450 3D conformal beamlets, based on segment end locations and motion states, delivering
Made with FlippingBook - Share PDF online