AN INVESTIGATION OF PARAMETERS AFFECTING THE SPURIOUS VARIATION OF PHOTON FLUENCE IN INTENSITY MODULATED RADIATION THERAPY

Abstract

The parameters that influence the amount of spurious variation in a set of optimized fluence maps (FMs) for intensity modulated radiation therapy (IMRT) were investigated. For this purpose, several simple geometries were used for testing various approaches. The FMs were characterized in term of the mean, standard deviation and first and second derivatives in the two cardinal directions. The peripheral and interior regions of the FMs were analyzed. The parameters tested and the results of the investigations were as follows. (1) The effect of the field margin: it was found that a margin of less than or equal to 0.5cm from the planning target volume (PTV) to the field edge increases the variation of a FM. (2) Incident beam direction: changing to a different but equivalent beam direction caused a small change in the variation of the FM. (3) Minimum transmission: small increases in the minimum transmission values reduce the amount of spurious variation. (4) Sequence of application of objectives: the sequence in which the objectives were applied influenced the smoothness of the fluence map to a small degree. (5) The influence of organ at risk separation from the PTV: the proximity of the PTV to organ at risk (OAR) affects the variation in a fluence map, independent of the thickness of the phantom or the beam direction. (6) The effect of different smoothing strategies on the optimized fluence maps: intelligent smoothing did not appreciably raise the composite objective value whereas a smooth-filter algorithm did. (7) The effect of different dose calculation grids: the amount of variation was reduced when the superior-inferior grid size was a factor of the pencil beam size. The factors that influenced the spurious variation the most were field margin size, ROI separation, minimum transmission and dose calculation grid dimensions. Strategic smoothing algorithms may reduce spurious variation while maintaining acceptable composite objective values. These results were applied to a very limited set of phantom geometries. In order to generalize this work, experiments will need to be done in clinically relevant geometries.