Implementation of angular response function modeling in SPECT simulations with GATE

P. Descourt, T. Carlier, Y. Du, X. Song, I. Buvat, E. C. Frey, M. Bardies, B. M.W. Tsui, D. Visvikis

Research output: Contribution to journalArticlepeer-review

18 Scopus citations


Among Monte Carlo simulation codes in medical imaging, the GATE simulation platform is widely used today given its flexibility and accuracy, despite long run times, which in SPECT simulations are mostly spent in tracking photons through the collimators. In this work, a tabulated model of the collimator/detector response was implemented within the GATE framework to significantly reduce the simulation times in SPECT. This implementation uses the angular response function (ARF) model. The performance of the implemented ARF approach has been compared to standard SPECT GATE simulations in terms of the ARF tables' accuracy, overall SPECT system performance and run times. Considering the simulation of the Siemens Symbia T SPECT system using high-energy collimators, differences of less than 1% were measured between the ARF-based and the standard GATE-based simulations, while considering the same noise level in the projections, acceleration factors of up to 180 were obtained when simulating a planar 364 keV source seen with the same SPECT system. The ARF-based and the standard GATE simulation results also agreed very well when considering a four-head SPECT simulation of a realistic Jaszczak phantom filled with iodine-131, with a resulting acceleration factor of 100. In conclusion, the implementation of an ARF-based model of collimator/detector response for SPECT simulations within GATE significantly reduces the simulation run times without compromising accuracy.

Original languageEnglish (US)
Pages (from-to)N253-N266
JournalPhysics in medicine and biology
Issue number9
StatePublished - 2010

ASJC Scopus subject areas

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging


Dive into the research topics of 'Implementation of angular response function modeling in SPECT simulations with GATE'. Together they form a unique fingerprint.

Cite this