Dose uncertainty due to aperture effects in dynamic fields

Dosimetry of intensity modulated radiation therapy requires accurate modeling of the beamlets that comprise each treatment segment. Planning systems such as Varian Eclipse and Philips Pinnacle recommend measuring dose distributions and output factors for fields as small as possible, generally down to at least 2x2 cm2. Conventionally, we perform these measurements for regular fields, defined by the secondary collimators. In practice, it is the multileaf collimation system (MLC) that defines the intensity map and provides dynamic dose modulation in either a moving window or segmented step-and-shoot mode. For this review we have only considered the latter delivery mode. Using this method, we have studied aperture motion effects on the dynamic collimator scatter (Sc), total scatter (Sc,p), and phantom scatter (Sp) factors for various combinations of collimator settings (4x4-14x40 cm2) and dynamically stepped leaf gaps (0.1 to 1.0 cm) in comparison with those for static field factors. For two different Varian linear accelerators, we found similar results in a systematic dependence of collimator scatter on gap width and collimator setting. As the gap increases from 0.1 to 1.0 cm the dynamic collimator scatter factors converge from a maximum difference of about 30% toward the static field values. At the same time, there is no measurable difference between dynamic field phantom scatter factors and those conventionally obtained for static fields. Second, we evaluated the two planning systems as to how well they account for collimator scatter by attempting to mimic the dynamic apertures used above by planning and measuring dose distributions to several small, cylindrical targets for a similar range of fixed collimator settings. We found that the ratio of measured-to-planned doses as a function of target size were similar to the measured, dynamic Sc data for the Varian Eclipse planning system, indicating underestimation of dose for targets smaller than 1 cm diameter, but were close to unity for the Philips Pinnacle system, suggestive of the underlying differences in the dose calculation algorithms. We discuss the measurements and results and potential impact on the dosimetry of small clinical targets