基于多算法的电子剂量场反演方法

为了研究精确放射治疗中多算法的电子剂量场反演方法及反演效果,建立非线性规划电子剂量场反演模型,利用体外透射剂量重建全空间电子剂量场,并通过多种算法实现。采用7个例题测试结果表明:除NMinimize算法不能实现该模型外,其他算法都能实现,其中Levenberg-Marquardt算法最精确快速。进一步测试表明,利用接近电子射程的透射剂量会带来较大的反演误差(尾差),原因可能是能谱平均带来的,因此改进方法需考虑改进对平均能量的近似。但由于反演时间较短,精度高,可实现实时剂量场反演;并且每一种算法都有特点,为实现提供多方法的选择性,满足临床上对实时剂量验证的需要。

Electron dose map inversion based on several algorithms

The reconstruction to the electron dose map in radiation therapy was investigated by constructing the inversion model of electron dose map with different algorithms. The inversion model of electron dose map based on nonlinear programming was used, and this model was applied the penetration dose map to invert the total space one. The realization of this inversion model was by several inversion algorithms, The test results with seven samples show that except the NMinimize algorithm, which worked for just one sample, with great error, though, all the inversion algorithms could be realized to our inversion model rapidly and accurately. The Levenberg-Marquardt algorithm, having the greatest accuracy and speed, could be considered as the first choice in electron dose map inversion. Further tests show that more error would be created when the data close to the electron range was used (tail error). The tail error might be caused by the approximation of mean energy spectra, and this should be considered to improve the method. The time-saving and accurate algorithms could be used to achieve real-time dose map inversion. By selecting the best inversion algorithm, the clinical need in real-time dose verification can be satisfied.