S1964
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
treatment planning. References:
1. Lucas A: Ising Fomulations of many NP problems. Front. Phys. 2014; 2:5. doi: 10.3389/fphy.2014.000052. Quinton FA., et al.: Quantum annealing applications, challenges and limitations for optimisation problems compared to classical solvers, Sci Rep 2025, 15:12733. https://doi.org/10.1038/s41598-025-96220-23. D-wave Systems. Simmulated Annealing Sampler, dwave-neal 0.5.7 documentation. https://dwave-neal- docs.readthedocs.io/en/latest/reference/sampler.html 4. D-wave Systems. Dwave.samplers.TabuSampler documentation. https://docs.dwavequantum.com/en/latest/ocean/api_ ref_samplers/generated/dwave.samplers. TabuSampler.sample.html 5. Nishimura K., et al.: OpenJij: Framework for the Ising model and QUBO 2025. https://www.openjij.org/ Keywords: Brachytherapy, Quantum Computing, BQM Characterizing and Correcting Material-Dependent Differences Between Acuros XB and Monte Carlo Dose Calculations Mohammed Ghazal 1,2 , Åsa Carlsson Tedgren 1,2 , Hamza Benmakhlouf 3,2 1 Nuclear Medicine and Medical Physics, Karolinska University Hospital, Stockholm, Sweden. 2 Oncology- Pathology, Karolinska Institutet, Stockholm, Sweden. 3 Radiation Oncology, Karolinska University Hospital, Stockholm, Sweden Purpose/Objective: This study investigates the residual differences reported in previous research between the deterministic Acuros dose calculation algorithm and benchmarked Monte Carlo (MC) simulations1,2. The objective is to identify the underlying sources of these Digital Poster 3935 discrepancies and explore potential corrective strategies to improve agreement between the two methods. Material/Methods: Acuros dose calculations were evaluated against MC simulations using PRIMO. Calculations were performed in a water phantom geometry containing an insert volume filled with the material of interest. The investigated materials included lung, adipose, muscle, cartilage and bone. Mean doses were reported as dose-to-medium in medium Dm,m (Acuros and PRIMO), dose-to-water in water Dw,w (PRIMO) and dose-to-water in medium Dw,m (Acuros). Observed discrepancies were further analyzed, explained and corrected where possible.
Results: All optimizers produced similar solutions in terms of dwell-times and solution quality, with target-coverage and OOI-sparing respecting prescribed dose guidance (fig.1, bottom). Increasing bit-depth improved solution quality, with exponential decrease in residual optimization cost, plateauing at 6 bits. All optimization methods remained stable starting 6 bits (fig.2). Wall- clock time scaled linearly for SA, Tabu, and SQA, but exponentially for GD. Tabu was consistently fastest with 0.04 s at 8 bits compared to GD 56.9 s, SA 8.1 s and SQA 18.7 s. Importantly, SQA reproduced the dynamics of real quantum annealing by introducing a Trotter dimension and quantum-tunneling effects; thus suggesting that real quantum annealers would yield similar solutions [5].
Conclusion: We successfully demonstrated that HDR brachytherapy planning can be formulated as a BQM, representing the first direct mapping of this optimization problem to a QUBO model. Classical and quantum-inspired solvers provided similar plan quality for the considered setup, suggesting the feasibility of this approach. With increasing availability of quantum hardware, this work highlights a pathway towards exploiting fast quantum computing for radiotherapy
Made with FlippingBook - Share PDF online