Quantum-Inspired Optimization for Radiotherapy Dose Planning: A Comparative Study with Classical Simulated Annealing

This comprehensive study presents an in-depth comparative analysis betweenquantum-inspired optimization algorithms and classical simulated annealing for ra-diotherapy dose planning optimization. Building on recent advancements in quan-tum computing applications for medical physics [11, 2], this research addresses acritical challenge in radiation oncology: optimizing dose distributions to achievetherapeutic tumor coverage while minimizing radiation exposure to healthy tis-sues. Using identical initial conditions (PTV D95: 28.2 Gy, target: 66.5 Gy, PTVsize: 14,147 voxels), both methods underwent 120 optimization iterations. Thequantum-inspired approach demonstrated remarkable efficacy, improving D95 to63.0 Gy (123.4% improvement), while classical simulated annealing achieved onlymarginal improvement to 28.3 Gy (0.3% improvement). Beyond the primary D95metric, the quantum-inspired method significantly outperformed in homogeneity(0.313 vs 1.931 HI), dose uniformity (6.70 vs 15.40 standard deviation), and overalldose distribution quality. Statistical analysis confirmed these differences as highlysignificant (p ¡ 0.000001, Cohen’s d = 0.431). This research not only establishesthe superiority of quantum-inspired methods for this application but also providesmechanistic insights into why these approaches succeed where classical methods fal-ter, offering valuable guidance for future algorithm development in medical physics.