Simultaneous emission and transmission scanning in PET oncology: effect on parameter estimation

We have previously reported the use of simultaneous emission and transmission (SET) scanning in whole body PET and the technique may also be applicable in kinetic studies. However, there are potential sources of bias, particularly at low emission count rates, which may affect the accuracy of parameter estimates. We investigated these potential sources of bias and their effect on parameter estimation in oncological PET studies. The sources of bias considered include: (i) variation in transmission spillover (into the emission window) throughout the field of view, (ii) increased scatter arising from the rod sources, and (iii) inaccurate dead time correction. Net bias was calculated as a function of emission count rate and used to predict distortion in FDG tissue curves simulating normal and metastatic liver. The distortion was characterised by spectral analysis and the effect on parameter estimates was assessed by compartmental modelling. Variation in spillover was found to be a negligible source of bias in SET measurements. Scatter and dead time errors work in opposing directions and approximately cancel during the early part of the study when count rate is maximal. The resulting distortion causes apparently decreased tracer clearance, particularly in tumours, which mainly affects estimates of k/sub 4/ and volume of distribution. Estimates of K/sub 1/-k/sub 3/ were relatively unaffected, while bias in glucose metabolic rate was +8% and +2% for normal liver and tumour respectively.