Development of a 3-D convolution/superposition algorithm for precise dose calculation in the skull

In this paper an algorithm for calculating 3-D dose distributions within the brain is introduced and adapted to the demands of modem radiosurgery. The dose calculation with this model is based on a 3-D distribution of the primary photon intensity which is calculated with a ray casting algorithm. A prelocated matrix takes into account field sizes as well as modifying elements as collimator positions (MLC), blocks, wedges and compensators. Monte Carlo precalculated monoenergetic kernels from 0.1 MeV to 50 MeV were at our disposal. The components of the spectrum were either determined by deconvoluting depth dose curves measured in water or analyzed with a Ge-Li detector system in the case of 60Co. The calculated fluence distribution has to be superposed to the complete kernel containing the spatial energy deposition. Inhomogeneities and tissue interface phenomena (rhoe, Z) have been investigated. The divergence of the rays and the curved surface of the patient are taken into account. Assuming homogenous media, it is possible to shorten the computation time by using the Fast Fourier Transformation (FFT) delivering a first overview within seconds. The algorithm was evaluated and verified under specific conditions of small fields as used in radiosurgery and compared to dose measurements and Monte Carlo calculations. In using both the fast algorithm (FFT) for mainly homogenous conditions on one hand and the very precise superposition for inhomogeneous cases on the other, this algorithm can be a very helpful instrument especially for critical locations in the skull.