A Monte Carlo model for quality assurance of intensity-modulated radiotherapy

Intensity modulated radiotherapy provides improved target coverage and reduced dose to surrounding normal tissues compared with conformal radiotherapy. However, computational quality assurance is more challenging for the complex fluence maps used in IMRT treatments, and direct measurements can be labor-intensive. A Monte Carlo based phase space model has been developed based on the Novalis linear accelerator to simulate arbitrary static fields and IMRT sequences. The basis for the model is the MCNP4C code, which accounts for the lack of lateral electronic equilibrium present in the small fields used in IMRT. This work is based on a virtual phase space source model, which is a two-step process. In the first step, the open beam fluence is calculated by simulating the components of the linear accelerator treatment head above the field defining multileaf collimator. This is done one time for a machine, and the resulting fluence map is used in all subsequent dose calculations. In the second step, this fluence map is then adjusted to match the physical beam using an intensity grid, which incorporates a detailed model of the multileaf collimator. The intensity grid accounts for the shaped leaf tip geometry and the beam divergence that influence the dose at the edge of the open leaves. It includes the transmission through the leaves, the leakage between them, and the tongue-and-groove design, which affect the dose under the leaves. The variation in beam energy across the field is also incorporated with a look-up table of effective attenuation coefficients based on the position in the field. The model can simulate both segmented and dynamic sequences. Depth dose and profile calculations for three field sizes agree well with measurement. Irregular field calculations in homogeneous media are compared with film measurement, and IMRT plan simulations in heterogeneous media are compared with film measurement and an accepted treatment planning system. These results show the accuracy of the model as a dosimetric verification tool for clinical treatments. The ability of the model to simulate leaf sequencing effects is also demonstrated.