H.R. Khosravi - National Radiation Protection Department of Iranian Nuclear Regulatory Authority. Medical Physics Department, Tehran University of Medical Sciences, Tehran, Iran
S. Sarkar - Medical Physics Department, Tehran University of Medical Sciences, Tehran, Iran. Nuclear Medicine Department, Shariati Hospital, Tehran University of Medical Sciences
A. Takavar - Research Center of Science and Technology in Medicine, Tehran University of Medical Sci. Nuclear Medicine Department, Shariati Hospital, Tehran University of Medical Sciences
H. Khosravi - Medical Physics Department, Tehran University of Medical Sciences, Tehran, Iran

Objectives: Monte Carlo (MC) simulations for SPECT imaging are typically computationally inefficient as much of the work involved in photon tracking is wasted when the photon is not ultimately detected by the camera. Various variance reduction techniques such as forced detection (FD) have been implemented in MC in an effort to decrease the simulation time while maintain accuracy, which will be used for dosimetry application in Radionuclide Therapy (RNT). However most of these techniques still result in Monte Carlo simulation times that are too long for incorporation into routine use. Methods: We have recently incorporated a method of variance reduction technique (CFD) into the SIMIND MC photon simulation software. This model uses the conventional SIMIND program to calculate the emitted photon path through an object and ultimately to the camera. However, rather than this photon being detected at a single point on the camera, the photon is spread over a region whose size and shape is function of the specific collimator and the collimator-source distance. Results: We have performed a variety of imulations between FD and CFD versions of SIMIND modeling I-131 radionuclide and camera configurations. Geometric response functions using the CFD version agrees very well with experimental measurements for gamma camera. Simulations of larger extended sources also agree well with FD versions of SIMIND, but simulation time is reduced for similar SNR values by a factor of 10. Conclusions: Convolution-based forced detection (CFD) has been modeled into the SIMIND MC program to improve the overall simulation time needed for accurate photon distribution estimates. With the current implementation of CFD, simulation times were approximately 10 times faster with equivalent signal to noise ratio compared with FD Monte Carlo.

Monte Carlo, SPECT, Dosimetry application, I-131