Efficient Calculation of in vivo Efficiency Curves Using Variance Reduction Techniques

Abstract

To optimize the in vivo lung monitoring of nuclear workers, realistic calibration coefficients are assessed using 3D anthropomorphic models and Monte Carlo simulations. In this study, a Livermore voxel phantom and the torso of the ICRP adult female reference voxel phantom were used. Monte Carlo MCNPX simulations were achieved to compute the calibration curve for a typical germanium counting system with a photon source of energy ranging from 15 keV to 1.4 MeV. However, photons of low energy are highly attenuated by body compounds while high energy photons are too penetrative to interact in the detectors. Hence, statistically insufficient counting efficiency values are obtained for both energy ranges. Thus, MCNPX variance reduction techniques were considered here to accelerate the simulations. The variance reduction techniques used were: the source biasing to favor low energy emissions; the emission direction biasing to select emission towards detectors; the forced collision to improve high energy photons interaction in germanium active cells. These techniques reduced the computing time by a factor of 20 and improved the associated statistical error with no significant variation in counting efficiency. Cell importance, mesh tally, weight windows and exponential transform variance reduction techniques were also tested to further accelerate computation.

Personalised Body Counter Calibration Using Anthropometric Parameters

“…An overview of those methods is given by Shultis and Faw (2006). And a comparison with application to body counting is given by Farah, Broggio and Franck (2011a).…”