RAPT - Monte Carlo Stereotactic Radiosurgery Simulation Software

Gamma Knife: many beams, one focus

Gamma Knife® Radiosurgery is a non-invasive medical procedure that uses beams of ionising photons from 201 ⁶⁰Co sources to treat intra-cranial lesions (Elekta/Leksell Gamma Knife). The Gamma Knife® unit comes with a treatment planning system - GammaPlan® - that uses an approximate description of the photon interactions within the head of the patient to calculate the energy dose deposited by these photons in the region of the tumour. There is significant benefit to be obtained from improving the fidelity of these calculations, particularly in cases where photons traverse regions of widely differing electron densities (e.g. soft tissue and bone). Monte Carlo modelling can accommodate these complexities and can play a useful role in complementing the GammaPlan solution (with the potential to eventually supersede it in the event of short enough calculation times).

The work of GEMSS has produced a Monte Carlo simulation tool for Gamma Knife radiosurgery - RAPT - based around the open source EGS4 (Electron Gamma Shower 4) Monte Carlo code. Grid computing has been explored as a means of keeping compute times to an acceptable minimum. The GEMSS Grid enables calculation of the energy dose delivered to the brain from a Gamma Knife® treatment unit to be obtained in clinically useful timescales (less than one hour).

The Leksell Gamma Knife

Running a RAPT job begins with assigning a job name to the simulation followed by entry of numerous parameters that define the simulation scenario. The software has been designed to be compatible with the Gamma Knife radiosurgery planning software and therefore accepts all parameters that are relevant to GammaPlan. This includes information such as:

• Number of shots
• Isocentre position for each shot
• Collimator size for each shot
• Plugging pattern
• Weight for each shot
• Gamma Rotation
• Bubble head measurements

Note also that additional parameters particular to Monte Carlo modelling are required, in particular the number of photons to be modelled in the simulation. For a single shot simulation (or one involving closely co-located isocentres), the use of approximately half-a-billion (0.5x10⁹) photons provides a solution that is accurate to approximately 2-3%. Grid support provides the necessary performance for the dose distribution to be computed in under an hour (running on 30 processors). Click here to see a demo of the application (5MB video)...

Part of the RAPT interface

The computed dose distribution is visualized with a viewer that enables the user to explore the distribution of radiation dose within the treatment volume. The dose data can be viewed as contours in principal orthogonal planes (transaxial, sagittal, coronal) or visualised as isosurfaces in 3D (3MB video). An additional utility - MetriX - has been developed to permit quantitative comparison of plans. This is useful when comparing the merits of different proposed treatments or comparing the dose distribution obtained by GammaPlan with the solution obtained by RAPT (1.5MB video).

A dose isosurface in RAPT

The GEMSS project has now finished, but RAPT remains functional as a GEMSS application demonstrator on the NEC Grid server (St Augustin, Germany). As a result, numerous projects are being undertaken during 2005 to more fully quantify the differences between RAPT and GammaPlan, particularly in cases of tissue inhomogeneity. Already this work is raising questions that may influence the way that such treatments are undertaken in the future. It is a direct result of the work of GEMSS that Sheffield now possesses a tool that can simulate clinical plans using Monte Carlo techniques. Few centres in the world have this capability, and Sheffield is actively pursuing publication in prestigious journals to highlight the value of this work. Simultaneously, several of the GEMSS partners are investigating additional funding to further promote the use of RAPT beyond the end of GEMSS.

RAPT can accommodate tissue inhomogeneity

The success of this project is a result of the commitment of many people. Thanks are extended to:

• Dept Medical Physics, University of Sheffield, UK
• Stereotactic Radiosurgery, Royal Hallamshire Hospital, Sheffield, UK
• IT Innovation Centre, University of Southampton, UK
• The GEMSS project partners - Funded by the European Union FP5 programme