GATE Monte Carlo Simulation in a Cloud Computing Environment
B Rowedder1*, Y Kuang2, (1) University of Nevada, Las Vegas, Las Vegas, Nevada, (2) University of Nevada, Las Vegas, Las Vegas, NVSU-E-T-18 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
Purpose: The GEANT4-based GATE is a unique and powerful Monte Carlo (MC) platform, which provides a single code library allowing the simulation of several specific applications, e.g. PET, SPECT, CT, internal and external radiotherapy, and hadron therapy. However, its lengthy computing time hinders its routine use in the clinic. Reducing its computing time is therefore of great importance. Thus, a commercial cloud compute service is well suited for GATE MC simulation, both in terms of cost and efficiency. This study achieves a reliable and efficient execution of GATE MC simulation and provides execution frameworks to end-users.
Methods: The GATE software was ported on a commercial compute cloud environment - Amazon Elastic Compute Cloud (EC2). Simulation data was initially loaded onto the master node, and then distributed among independent worker nodes. The filed output from EC2 was sent down to the Amazon Simple Storage Service (S3). The results were finally aggregated on the local computer for display and data analysis. The distributed implementation was executed using a photon beam interacting in a 40 cm by 40 cm by 40 cm water phantom and a four-head SPECT imaging as benchmarks.
Results: A cloud computing environment led to increased calculation speed for the cases implemented in this study. The speed increase scaled approximately linearly with the number of nodes used for computing. The output of the cloud-based GATE MC simulation was identical to that produced by the single-threaded implementation, and was resilient to hardware failure, indicating the reliability of the cloud computing platform. The user-friendliness offered by the workflow implementation does not introduce significant overhead.
Conclusion: A cloud computing infrastructure has been established for GATE MC simulation. It substantially improves the speed of simulation, and makes rapid MC simulation for imaging/radiotherapy application possible.