A Comparison of GEANT4 and DETECT2000 for the simulation of light transport in scintillation detectors

Monte Carlo simulation techniques are powerful tools for investigating the performance of imaging detectors. DETECT2000, a Monte-Carlo package for light transport simulation, has been commonly used to model the optical properties of scintillation detectors and is able to realistically estimate the response of photodetectors. However, DETECT2000 generally runs slowly when complex detector geometries are specified and the lack of visualization tools makes it difficult to accurately define complex geometries. GEANT4 is a simulation toolkit that can also realistically model optical photon transport for scintillation detectors. This paper describes a case study in which GEANT4 was found to be significantly faster both in computing time and, aided by visualization tools, in the user time required to develop the geometry of a scintillation detector. In addition, because the detector geometry can be easily parameterized using GEANT4, it was possible to perform automated searches of large amounts of the solution space for an optimal design. In this work, we compared the results from simulations of a custom-designed scintillation detector obtained using both packages. Both yielded similar flood images and were able to resolve the same number of pixel elements from a segmented light guide with slight differences in the peak-to-valley ratios. A simulation-speed comparison on a common computer using a simplified geometry showed that GEANT4 is about 40% faster than DETECT2000. In this paper we compare simulation results for the initial design with experimental data and describe subsequent simulations that were used to arrive at the optimal design.