Monte Carlo validation of a portal imager scatter dose model

A common and clinically significant drawback of current portal scatter dose estimation methods is the workload required to measure the scatter dose data for use in the scatter model. To address this problem, a scatter model based on first order Compton scatter was developed and validated for a photon beam energy of 6 MV. This approach was motivated by the observations that at large air gaps (i) the scatter dose is uniform across the portal image (as previously shown by others) and (ii) the scatter dose is principally from first order Compton scatter (shown here). Monte Carlo simulation was chosen for the development and validation of the model since with its use the scatter dose can be separated according to particle type and interaction history, which cannot be done experimentally. The model uses Monte Carlo derived scatter kernels that describe the imager dose from first order scattered photons generated in a 1 cm³ voxel located 50 cm or more above the imager. The model includes the divergence and attenuation of the primary and once scattered photons. Dose from multiple scatter was dealt with effectively by slightly overestimating the dose from first scatter. For 36 phantoms (homogeneous, slab, and anthropomorphic), the root mean square deviation between the SPRs calculated using the scatter model and the SPRs from Monte Carlo data was 0.6% or less. For the anthropomorphic phantoms, the authors´ model is shown to be comparable in accuracy to a current scatter dose estimation method used for in vivo dosimetry