Sampling techniques for the evaluation of treatment plans.

Sampling techniques using randomly distributed points and regular Cartesian grids were compared for the evaluation of volume, dose-volume histogram, tumor control, and normal tissue complication probabilities in radiation treatments. Particularly, the uncertainties associated with each sampling technique in estimating the dose-volume histograms for several dose distributions are analyzed in detail. It is found that the estimation of these parameters using sampling points on a regular Cartesian grid is, in general, significantly more efficient than using random points. This finding is different from other published results. The choice of grid size for sampling was analyzed according to the AAPM recommended uncertainty on the dose delivered to the patient. It was concluded that when grid sampling is used, a grid size of 0.5 cm is adequate for most plans to meet the guidelines.