Verification Of Monte Carlo Simulations And Convolution Dose Algorithm Used For Heterogeneous Geometries In Leksell Gamma Knife? Dosimetry Jonas Johansson1, Anders Lundin1, Håkan Nordström1, Markus Eriksson1, Per Kjäll1 1Elekta Instrument AB, Stockholm, Sweden Keywords: physics, dose planning, dosimetry, gamma knife, Monte Carlo ## Interactive ManuscriptAsk Questions of this Manuscript: Abstract A new convolution dose algorithm has been developed for Leksell Gamma Knife® so that Leksell GammaPlan® can account for heterogeneities in the anatomy. Data used as input to the convolution algorithm is simulated by Monte Carlo methods. Therefore, first the implementation of the Pegasos Monte Carlo system, which is based on the Penelope code, need to be verified experimentally. In this study this is done by means of film dosimetry. When Pegasos is verified, it can be used to verify the convolution algorithm. The purpose is to verify dose distributions for both the Monte Carlo system and the convolution algorithm. This is done by comparing the dose distributions using a so called MADD evaluation (maximum allowed dose difference). Several inhomogeneous phantom geometries were considered, including three different isocenter positions, with non-water inserts in the geometry. As a basis for the study the geometries have been based on a modified Elekta Dosimetry Phantom. In the comparisons the relative dose distributions were compared for all Leksell Gamma Knife® collimator sizes. The criteria used for the MADD evaluation for a pass is given by an acceptance region defined by a 1-5% absolute dose deviation in a combination with distance to agreement, here typically set to 1 mm. MADD provide a pass/fail map, where 1% of the voxels/pixel may fail. Preliminary results show a good agreement between the different dose distributions for the two different investigations. This is a retrospective study. The dose distributions from both the Pegasos Monte Carlo system and the new convolution algorithm used in Leksell GammaPlan® are in agreement with measurements. Acknowledgements Project Roles:J. Johansson (), A. Lundin (), H. Nordström (), M. Eriksson (), P. Kjäll () |

a global knowledge network