Normalization and randoms corrections for MADPET-II small animal PET scanner using Monte Carlo techniques

PET offers the possibility of quantitative measurements of tracer concentration. However, there are several corrections which must be addressed in order to fully exploit this potential. Using Monte-Carlo simulation techniques through GATE software, we are investigating normalization and randoms corrections applied to MADPET-II. In general, corrections are applied before image reconstruction. We want to explore the possibilities of including these corrections as part of the system matrix in an EM algorithm. In order to obtain higher sensitivity, the low energy threshold (LET) should be reduced. In this situation, inter-crystal (IC) scatter takes place (25% at 150 keV LET). We have observed that randoms obtained using standard methods are higher than randoms found in simulation for lower LET (37% increase at 150 keV LET). We are interested to see how this overestimation affects quantification of the images. Related to normalization, we plan to study the feasibility of using singles for estimating the intrinsic efficiencies of the crystals. This would reduce the number of unknowns that must be obtained and the statistics would be increased (measured singles are always higher than coincidences). The time alignment of the channels would be considered as part of the normalization factors