Comparison of dose calculation algorithms for Leksell Gamma knife perfexion using monte carlo voxel phantoms
Abstract
Dose calculation algorithms implemented in the treatment planning software for Leksell Gamma Knife differ in their results, especially in the areas of steep electron density gradients. The Tissue Maximum Ratio (TMR) algorithm (in two variants) does not employ any patient-specific data besides the idealized skull shape, while the Convolution algorithm takes full advantage of voxel data from computed tomography (CT) to create a more faithful description of the anatomy.
The presented Monte Carlo model of Leksell Gamma Knife Perfexion was created in order to investigate dose distribution characteristics of the machine and to verify the mentioned algorithms. It builds upon Geant4 (a simulation toolkit for particle passage in matter) and being designed in a modular fashion, it enables to put several geometry components together at runtime.
A precise description of the Perfexion collimation system is supplemented by voxel phantoms constructed from CT images in the DICOM format. As a preliminary study, three specific plans of varying complexity were exported from the treatment planning software - a phantom for gel dosimetry and two clinical cases - and simulated using the Monte Carlo model.
Scoring volumes were defined so that a direct dose distribution comparison could be made using three-dimensional gamma index analysis. All results show good agreement between the Monte Carlo method and calculated distributions with no significant difference between the algorithms (with over 99.9 % points satisfying the criteria M ΔD M = 0.03 and Δd M = 1 mm).
We plan to make further studies with more specific, artificially designed treatment plans in order to assess the subtle differences between the algorithms.