64. State of Monte Carlo calculations in radiation treatment planning

Monte Carlo (MC) particle transport simulations are increasingly applied in treatment planning methods. This has become feasible through a number of adaptations of general MC codes, such as EGS4 or ETRAN, to the specific needs of treatment planning. The currently most advanced “conventional” planning methods, such as convolution or delta–volume algorithms still have serious limitations in terms of accuracy when tissue inhomogeneities, small and complex body shapes or high-density implants are involved. The Monte Carlo simulation mimics individual particle transport, in any applicable geometry, by applying first principles of radiation interaction with matter and random choice of collision parameters such as step length, type of interaction, energies and scattering angles. In principle, the accuracy of MC calculations is only limited the radiation beam quality definition and the interaction parameters and can be taken to below 12%. In practice, a very large number of particle “histories” have to be simulated to attain sufficient statistical accuracy, and various approximations (e. g. condensed history, variance reduction) have been introduced in the process of adaptation of MC codes to the special needs of treatment planning. Such codes have become known as V(oxel)MC and X(ray)VMC, M(acro)MC, S(uper)MC, MCPAT(ient…). These will be described in detail and performance characteristics as well as treatment planning examples given. While the general-purpose MC codes result in computing times per case of the order of several hours, the special treatment planning codes reduce this time to around an hour or even much less on modern workstations or Pentium-based PCs