Evaluation of Dosimetric Differences Between the TMR 10 and Convolution Algorithm for Gamma Knife Radiosurgery Planning

Gamma Knife Radiosurgery (GKR) is a well established treatment modality for brain tumors and functional disorders of the brain. It relies on mathematical algorithms to predict dose distribution and to calculate the dose at arbitrary points in the head. For the last 25 years, doses applied using Gamma Knife Radiosurgery have been calculated using a simple algorithm, called the Tissue Maximum Ratio algorithm (TMR). Dose planning using this algorithm, relies on a number of approximations to enable fast isodose computation during treatment planning. One of the most significant of these is the approximation of the head to water-equivalent density. The increased electron density of brain and bone (relative to water) and the near-zero density of air cavities in the skull may make significant perturbations to isodose and beam-on time calculations.

With the advent of faster workstations, the effect of tissue in-homogeneities can finally be calculated in reasonable time during the treatment planning process; a newer, more modern algorithm known as convolution algorithm is now commercially available. It uses the values of density indicated in the CT scan to predict the dose distribution and is expected to more accurately calculate radiation dose, although it needs further investigation before clinical implementation. Inter- and intra-indication differences between the old and new algorithms need to be understood before this method can be confidently employed in a clinical setting. It is the aim of this study to understand the dosimetric differences between these dose calculation algorithms and to evaluate the implications of using the convolution algorithm for GKR. A large number of treatments will be re-planned using the convolution algorithm and compared to the TMR plans used to treat the patients. Beam-on-time, which is proportional to dose and a number of commonly used metrics for the targets such as coverage, selectivity, gradient index, and mean and maximum dose, will be estimated with both algorithms. Subgroup analysis will be done to assess whether any factor such as diagnosis, size of the head or location of the target could impact on the relative difference between the methods. The treatment plans will be compared and the potential implications on treatment planning will be elucidated.