S2252
Physics - Intra-fraction motion management and real-time adaptive radiotherapy
ESTRO 2026
difference. MLC updates occurred on average every 158±80ms. Experimental evaluation of dose-guided ART was performed by delivering IMRT treatment plans to a Quasar MRI 4D phantom (IBA Quasar, Canada) emulating sinusoidal target motion (20mm peak-to-peak amplitude, 5s period) and slow drift motion (0-10mm over treatment delivery). Motion was tracked using dose-guided ART, geometry-guided MLC tracking, and no tracking for single target and multi- target phantom setups (Figure 1). The dosimetric accuracy of each tracking method was evaluated by measuring the delivered dose to radiochromic film and comparing to a static reference. The end-to-end latency was evaluated by comparing the positions of the target and the MLC leaves in EPID images.
envelope expansion.
Keywords: MR-LINAC, PTV margins, simulations
Proffered Paper 3564
The first experimental implementation of real- time dose-guided adaptive radiation therapy
Emily A Hewson 1 , Pim TS Borman 2 , Bas W Raaymakers 2 , Paul J Keall 1 , Martin F Fast 2
1 Image X Institute, The University of Sydney, Sydney, Australia. 2 Department of Radiotherapy, University Medical Center Utrecht, Utrecht, Netherlands Purpose/Objective: Motion during radiation therapy decreases the accuracy of radiation dose deposition, compromising patient outcomes. Current intrafraction motion management strategies have been limited to geometry-based adaptations. However, dose-guided adaptive radiation therapy (ART) optimises to the most clinically relevant metric, with dose highly correlated with local tumour control and toxicities [1,2]. Real-time dose-guided ART has been developed [3] but until now, has been limited to simulation studies. This study presents the world-first experimental implementation of real-time dose-guided ART. Material/Methods: Real-time dose-guided ART was integrated with a 1.5T Unity MR-linac (Elekta, Sweden) in a research configuration. The dose-guided ART system used MRI to determine the target location, which was input into a fast, simplified dose calculation algorithm to perform motion-inclusive dose accumulation to estimate delivered dose. The difference between the delivered dose and the planned dose at each update timepoint was calculated and the next MLC leaf positions were optimised to deliver a dose that would minimise this
Results: The average gamma-pass rates using a 3%/3mm pass criteria for single target tracking was 95.2%, 98.1%, and 55.1% using dose-guided ART, geometry-guided MLC tracking, and no tracking. The average pass rates using a 3%/3mm criterion for multi-target tracking were 93.8% for dose-optimised ART, 91.1% for geometry-guided MLC tracking, and 63.6% for no tracking (Figure 2). The end-to-end latency was 545±20ms for dose-guided ART and 354±20ms for geometry-guided MLC tracking which was compensated for by a linear-ridge prediction model.
Made with FlippingBook - Share PDF online