Department of Radiation Oncology, University of California San Diego, La Jolla, CA 92037, USA.
Phys Med Biol. 2010 Aug 7;55(15):4309-19. doi: 10.1088/0031-9155/55/15/008. Epub 2010 Jul 20.
Online adaptive radiation therapy (ART) has great promise to significantly reduce normal tissue toxicity and/or improve tumor control through real-time treatment adaptations based on the current patient anatomy. However, the major technical obstacle for clinical realization of online ART, namely the inability to achieve real-time efficiency in treatment re-planning, has yet to be solved. To overcome this challenge, this paper presents our work on the implementation of an intensity-modulated radiation therapy (IMRT) direct aperture optimization (DAO) algorithm on the graphics processing unit (GPU) based on our previous work on the CPU. We formulate the DAO problem as a large-scale convex programming problem, and use an exact method called the column generation approach to deal with its extremely large dimensionality on the GPU. Five 9-field prostate and five 5-field head-and-neck IMRT clinical cases with 5 x 5 mm(2) beamlet size and 2.5 x 2.5 x 2.5 mm(3) voxel size were tested to evaluate our algorithm on the GPU. It takes only 0.7-3.8 s for our implementation to generate high-quality treatment plans on an NVIDIA Tesla C1060 GPU card. Our work has therefore solved a major problem in developing ultra-fast (re-)planning technologies for online ART.
在线自适应放疗(ART)具有很大的潜力,可以通过基于当前患者解剖结构的实时治疗自适应来显著降低正常组织毒性和/或提高肿瘤控制效果。然而,在线 ART 临床实现的主要技术障碍,即无法在治疗重新规划中实现实时效率,尚未得到解决。为了克服这一挑战,本文介绍了我们在基于之前在 CPU 上的工作在图形处理单元(GPU)上实现调强放疗(IMRT)直接孔径优化(DAO)算法的工作。我们将 DAO 问题表述为一个大规模凸规划问题,并使用一种称为列生成方法的精确方法来处理其在 GPU 上的极高维度。我们对 5 个 9 野前列腺和 5 个 5 野头颈部 IMRT 临床病例进行了测试,每个病例的射束大小为 5 x 5 mm(2),体素大小为 2.5 x 2.5 x 2.5 mm(3),以评估我们在 GPU 上的算法。在 NVIDIA Tesla C1060 GPU 卡上,我们的实现只需 0.7-3.8 秒即可生成高质量的治疗计划。因此,我们的工作解决了开发在线 ART 的超快速(重新)规划技术的一个主要问题。