F. Verhaegen et Hh. Liu, Incorporating dynamic collimator motion in Monte Carlo simulations: an application in modelling a dynamic wedge, PHYS MED BI, 46(2), 2001, pp. 287-296
In radiation therapy, new treatment modalities employing dynamic collimatio
n and intensity modulation increase the complexity of dose calculation beca
use a new dimension, time, has to be incorporated into the traditional thre
e-dimensional problem. In this work, we investigated two classes of samplin
g technique to incorporate dynamic collimator motion in Monte Carlo simulat
ion. The methods were initially evaluated for modelling enhanced dynamic we
dges (EDWs) from Varian accelerators (Varian Medical Systems, Pale Alto, US
A). In the position-probability- sampling or PPS method, a cumulative proba
bility distribution function (CPDF) was computed for the collimator positio
n, which could then be sampled during simulations. In the static-component-
simulation or SCS method, a dynamic field is approximated by multiple stati
c fields in a step-shoot fashion. The weights of the particles or the numbe
r of particles simulated for each component field are computed from the pro
bability distribution function (PDF) of the collimator position. The CPDF a
nd PDF were computed from the segmented treatment tables (STTs) for the EDW
s. An output correction factor had to be applied in this calculation to acc
ount for the backscattered radiation affecting monitor chamber readings. Co
mparison of the phase-space data from the PPS method (with the step-shoot m
otion) with those from the SCS method showed excellent agreement. The accur
acy of the PPS method was further verified from the agreement between the m
easured and calculated dose distributions. Compared to the SCS method, the
PPS method is more automated and efficient from an operational point of vie
w. The principle of the PPS method can be extended to simulate other dynami
c motions, and in particular, intensity-modulated beams using multileaf col
limators.