An audit of repositioning accuracy of Extend in the first year after installation with hints on how to minimize repositioning problems. Michael Torrens, PA1, Chryssa Paraskevopoulou2, Evangelia Kelesidou2 1Department of Neurosurgery, Hygeia Hospital 2Hygeia Hospital, Athens, Greece Keywords: Extend, repositioning, Fractionated radiosurgery, gamma knife, physics ## Interactive ManuscriptAsk Questions of this Manuscript: Abstract Repositioning is possible using the Extend system with the Gamma knife. Our purpose was to analyze the repositioning accuracy of Extend and identify factors that assist in the avoidance of problems and error during repositioning. Information was gathered prospectively on 14 treatments by Gamma Knife Perfexion using Extend from 2010-2011. A total number of 398 repositioning check measurements (RCM) were analyzed to determine the average error per patient treatment, the average total error, the difference between set up and treatment measurements and the possible influence of a learning curve. The repositioning checks were analyzed in groups according to the direction of the probe – superior (z axis), right or left lateral (x axis) or anterior (y axis). Notes were made at each treatment and problems recorded. Statistical analyses included repeated measures Anova and the Friedman test. All measurements are in mm. The distribution of measurements was not Gaussian. There was a difference in repositioning error between the setup/CT RCM (mean 0.94±1.20SD, median 0.50) and the CT/treatment RCM (0.67±1.01SD, 0.30) which was significant (P=0.004). Following this observation the setup/CT RCM were excluded from the total analysis. The average reposition error in each treatment was 0.76, range 0.1-2.8, but excluding the one outlier where there was evidently a mistake reduced this to an average of 0.60±0.45SD with a highest value of 1.5. In the assessment of the variations between the direction of check measurements it was found that anterior (y axis) was most accurate (mean 0.519), right/left (x axis) was intermediate (mean 0.675) and superior (z axis) least accurate (mean 0.891). This trend was not significant on a Friedman test (P=0.748). There was strong correlation between the repositioning accuracy and the time elapsed after installation (P<0.0001). In addition to manufacturer’s recommendations, it is essential to use the same (Perfexion) mattress at the same height on a similar flat surface for the CT scanner. Other qualitative error avoidance factors will be elaborated in a presentation. Submillimeter repositioning accuracy with Extend (mean 0.67, median 0.30) is achieved but requires meticulous attention to detail and considerable patience. This is accurate enough to perform single fraction radiosurgery and cases of this will be demonstrated. Acknowledgements Project Roles:M. Torrens (), C. Paraskevopoulou (), E. Kelesidou () |

a global knowledge network