Voxel based Monte Carlo calculations of nuclear medicine images and applied variance reduction techniques

摘要

Images acquired on modern digital volumetric clinical imaging instruments can be used to create 3-dimensional surfaces of the human anatomy for use in computer databases. Such a volume array based software phantom delineates internal organs with millimeter resolution, and lends itself to fully 3-dimensional Monte Carlo simulations. Our simulation models 45 internal human organs (each with an associated radioisotope concentration and attenuation coefficient), calculates gamma radiation histories through these structures, and accepts gamma events onto a collimated planar camera. Variance reduction techniques are applied to decrease the time required to compute a given number of events at the detector. Stratification and two implementations of forced detection variance reduction techniques are compared to ‘brute force’ calculations for their efficiency speed-ups in this heterogeneous geometry. Simulated clinical images of the liver are shown.