Research on Two Kinds of Convolution Kernels in Intensity Modulated Radiotherapy

Two typical kinds of convolution kernels in IMRT are presented which are respectively based on measured photon beam data by deconvolution method and the EGS Monte Carlo Code. Both of the kernels are used to compute the dose distribution to optimize the same IMRT plans under otherwise same conditions. It is found that the different kernels have tiny influence on the contour of iso-dose distribution but notable one on dose-received of the normal tissue and organs at risk. The plan optimized by EGS Monte Carlo energy distribution kernel get obvious higher dose in normal tissue and organs at risk. The Monte Carlo kernel can show more detail information of the dose distribution and Monte Carlo technique provides an excellent tool for "theoretical benchmark experiments" to determine exact dose deposition. While the kernel extracted by the deconvolution method from experiment data is evidently influenced by the instruments used for measurement of the basic set and it is a rough kernel to some extent, but it is still a simple and clear way to get the kernel in much shorter time compared to the Monte Carlo Way, it treats the measured data as resulting from the scattering of photons, thus avoiding conceptual and implementational difficulties associated with scatter-and-primary dose methods.