ESTRO 2025 - Abstract Book

S3278

Physics - Intra-fraction motion management and real-time adaptive radiotherapy

ESTRO 2025

In this phantom study, we demonstrate the feasibility of a PET-guided MR-linac workflow, utilizing an in-house developed PET motion-correction framework to reduce motion-induced artefacts, PET-MOTUS [1], and an in-house developed MR-guided MLC tracking workflow [2] to mitigate motion-induced inaccuracies during delivery. Material/Methods: PET/CT images were acquired of the QUASAR MRI 4D motion phantom (IBA QUASAR, London, ON). The phantom consists of a static anthropomorphic body with movable insert. During the PET/CT acquisition, the insert was moved with a cos 4 () motion (Apeak-to-peak=20mm, T=4s), or kept statically as reference. A spherical volume with a diameter of 3 cm within the insert was filled with 1.5MBq/mL F 18 -FDG and 8:1 sphere-to-background ratio. Acquiring raw list mode scans allowed static, motion-blurred, and motion-corrected reconstructions via PET-MOTUS [1] (Figure 1a). A 15-beam 8x7.5Gy lung IMRT MR-linac plan was created with automatic 40% SUV max delineations including 3mm PTV margins for the static, motion-blurred and motion-corrected PET/CT images. (Figure 1b). These plans were delivered to a static and moving phantom with a film insert, with and without MLC-tracking[2]. EBT3-film was used for dosimetry and for the comparison of PET/MR-linac motion correction strategies through dose differences and 3%/3mm global gamma pass-rates relative to the static reference.

Results: Motion-blurred PET-based delineations showed larger target volumes (24cc) compared to motion-corrected and static references (18cc), owing to the motion correction during target definition(Figure 1). Dosimetric results revealed reduced geometric and dose differences for motion-corrected PET, and dose differences are further reduced using MLC-tracking. Gamma pass rates of 43.5% and 78.4% were obtained for conventional delivery, and 48.2% and 95.6% for MLC-tracking, for motion-blurred and motion-corrected PET respectively (Figure 2). This indicates that, for these experiments, motion correction during target definition has more dosimetric impact than during delivery.