BNCT优化网格设计及相关算法研究

Optimized Voxel Model Construction and Simulation Research in BNCT

用MCNP蒙特卡罗程序模拟了硼中子俘获治疗(BNCT)3种国际基准网格模型,并与修正的 Snyder椭球模型进行了比较．在此基础上,给出了一种保质量守恒、内存量少、易于产生输入文件的4 种基本材料成分的BNCT网格模型．计算结果表明,在4mm网格下,新模型可以达到基准模型的精度; 根据解析模型剂量随深度的变化规律,研究构造了多网格组合模型,在重要区域计算精度不损失的条件下,计算时间大大缩短．最后研究给出了一个既保证精度、又在可接受的时间内完成剂量计算的模型、样本数和相应的算法,它基本上满足临床BNCT的要求．

The voxel model of Snyder head phantom is widely used in simulation of BNCT. This paper introduces a simplified method which is based upon only four kinds of materials. The new voxel model keeps the mass conservation when the mesh size is small such as 4mm. The 4-material voxel model and 286-material voxel model with different mesh sizes of ]6mm, 8mm, 4mm, are simulated respectively by MCNP Monte Carlo code. The results indicate that the 4- material voxel model can reach good accuracy. Using this method, multi-voxels are produced and simulated. Compared to the 4mm single voxel model, the multi-voxel model has good accuracy and less computational time. Finally, we recommend multi-voxel models and a reasonable number of meshes. Its computational time and accuracy have accessed the clinical BNCT requirement.