基于多核的并行模拟退火放疗计划优化加速方法

Speedup Inverse Treatment Planning Based on Multi-core Parallel Simulated Annealing

为了满足图像导引(imaging guided radiotherapy,IGRT)的实时四维适形调强逆向放疗计划在临床实施的需要,用一台HP桌面型工作站和OpenMP多线程库,试验了用多核并行计算技术实现的模拟退火逆向治疗计划快速全局优化算法,并将其应用在一个虚拟体模和4个临床病例的优化中。实验结果表明:基于多核的并行计算技术能大幅度提高逆向放疗技术的优化计算的效率;进一步选用更多核的并行计算机图像工作站,该技术可以满足目前大多数4D医学成像的临床需要。

In order to realize the real time image-guided 4D prompt conformal intensity modulated radiotherapy in clinical practice using inverse optimization algorithm, the authors tested the global optimization search of the simulated annealing algorithm with a HP desk workstation and OpenMP multi thread library, which was further used for the optimization process for a virtual phantom and 4 real treatment plans of cancer patients. The results show that the method used in the test can speed up the search efficiency notably, if more CPU or core are chosen, the multi-core based parallel calculating technology can satisfy most requirements for 4D medical imaging in clinical at current time.