基于GPU加速的质子调强放疗鲁棒优化器

开发一种基于图形处理器（GPU）加速的质子调强放疗鲁棒优化器,用于减小质子束射程不确定性和靶区定位偏差对质子放疗的影响。建立的鲁棒优化模型使用的目标函数包括9种边界剂量目标,分别是:无偏差情况、2种射程偏差（偏长与偏短）、6种摆位不确定性（前后、侧向、上下入射方向各2种正负偏差）。首先靶区和危及器官的剂量贡献矩阵使用笔形束算法计算得到,然后使用共轭梯度法优化目标函数让其满足约束条件,这两部分均采用GPU加速。头颈部、肺部和前列腺三个临床病例被用来检测本优化器的性能表现。与传统基于计划靶区（PTV）的质子调强放疗计划相比,鲁棒优化器能够优化出对射程不确定性和摆位误差更加不敏感的治疗计划,让靶区实现了高剂量均匀性的同时危及器官（OARs）也得到了更好的保护。经过100次迭代,三个病例的优化时间均在10 s左右。该结果证明了基于GPU加速的质子调强放疗鲁棒优化器能够在短时间内设计出高鲁棒性的质子治疗计划,从而提高质子放射治疗的可靠性。This paper describes the development of a fast robust optimization tool that takes advantage of the GPU technologies. The objective function of the robust optimization model considered nine boundary dose distributions--two for ±range uncertainties, six for ±set-up uncertainties along anteroposterior (A-P), lateral (R-L) and superior{inferior (S-I) directions, and one for nominal situation. The nine boundary influence matrices were calculated using an in-house dose engine for proton pencil beams of a finite size, while the conjugate gradient method was applied to minimize the objective function. The GPU platform was adopted to accelerate both the proton dose calculation algorithm and the conjugate gradient method. Three clinical cases-one head and neck cancer case, one lung cancer case and one prostate cancer case-were investigated to demonstrate the clinical significance of the proposed robust optimizer. Compared with conventional planning target volume (PTV) based IMPT plans, the proposed method was found to be conducive in designing robust treatment plans that were less sensitive to range and setup uncertainties. The three cases showed that targets could achieve high dose uniformity while organs at risks (OARs) were under better protection against setup and range errors. The run times for the three cases were around 10 s for 100 iterations. The GPU-based fast robust optimizer developed in this study can serve to improve the reliability of traditional proton treatment planning by achieving a high level of robustness in a much shorter time.