S1844
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
target distribution (CTD)) relies on a central hypothesis related to the level of correlation between voxels, often expressed as either fully dependent or independent. However, current formulations of such CTDs yield a strong voxel size dependency with respect to probabilistic TCP computations, which is an unwelcome feature (TCP should not depend on voxel size for a given dose distribution). In this communication, we demonstrate why it is the case, and we propose a novel formulation which enforces invariance to voxel geometry. Material/Methods: Tumor extent, P(x), and decreasing density, ρ i(x), were drawn from maximal tumor extension (MTE) distances from the GTV border, Mi, determined by histopathological studies4 (Figure 1). Using the tumor extent probability curve, one can discretize probability at the voxel level according to [1]. Applying the related TCP formulation3, we prove that:The expected tumor load decreases with reduced voxel size for the same total volume, even with recomputed probabilities.Probabilistic TCP will always be biased by voxel size and is incompatible with the independent voxels hypothesis (Figure 2).No formulation of the independent voxels hypothesis can maintain expected tumor load over the same total volume when changing voxel size.
and interestingly their average clinical acceptability rate was only 50%, highlighting considerable inter- observer variability in plan evaluation. Plan preference between the two automated methods was evenly split (50%vs50%), with full expert agreement in 3 of 8 cases (Table2). Patient-specific QA results demonstrated excellent deliverability for all automated plans.
Conclusion: Both automated VMAT planning methodologies successfully generated high-quality, clinically acceptable plans with minimal human intervention. These findings demonstrate the feasibility and reliability of automated end-to-end VMAT planning for oropharyngeal cancer, paving the way for integration into routine clinical workflow to enhance consistency, efficiency, and plan quality. References: [1] M. Hussein et al., “Automation in IMRT planning – recent innovations,” Br. J. Radiol., vol. 91, p. 20180270, 2018.[2] P. Meyer et al., “Automation in RT planning: clinical use and trends,” Cancer/Radiother., vol. 25, pp. 617–622, 2021.[3] A. Munshi et al., “Dose fall-off with VMAT and 3D-CRT,” Cancer Radiother., vol. 23, pp. 138–146, 2019.[4] A. W. Lee et al., “Dose prioritization in RT for NPC,” Int. J. Radiat. Oncol. Biol. Phys., vol. 105, pp. 567–580, 2019. Keywords: auto-planning, head-and-neck cancer Digital Poster 1575 Overcoming voxel bias in probabilistic CTV definition and mapping Eliot Peeters, Ana Maria Barragan Montero, Edmond Sterpin, John Aldo Lee MIRO, UCLouvain, Brussel, Belgium Purpose/Objective: Shifting from binary to probabilistic target volumes has the potential to enhance healthy tissue sparing while maintaining tumor control probability (TCP)1-3. Probabilistic cartesian mapping of the CTV (clinical
Results: As cells (i.e., voxels) neighboring the tumor are neither totally independent nor dependent5, we emphasize that only Monte Carlo simulations of tumor expansion and cell kill by irradiation will reflect the true expected
Made with FlippingBook - Share PDF online