RT Journal Article
SR Electronic
T1 Characterising the performance of the simplified and pseudo reference tissue models
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 2070
OP 2070
VO 52
IS supplement 1
A1 Cristian Salinas
A1 Graham Searle
A1 Roger Gunn
YR 2011
UL http://jnm.snmjournals.org/content/52/supplement_1/2070.abstract
AB 2070 Objectives The simplified reference tissue model (SRTM) is a kinetic analysis method that is commonly used to quantify brain PET studies and does not require the measurement of blood data. The model is based on a single tissue compartmental model configuration in both reference and target tissues, assumes that the reference region is devoid of specific binding and that blood volume (VB) contribution to the tissues is negligible. The Pseudo Reference Tissue Model (PRTM) is similar except that it corrects for the presence of specific binding in the reference region. We explore how SRTM and PRTM perform when their assumptions are not met. Methods Computer simulations were used to investigate the impact of specific binding in the reference region, a range of model topologies in the reference and target region (one or two tissue compartments), and VB contributions in terms of bias and precision of the binding potential (BPND) in the target region. Simulations were performed in the absence and presence of noise. Results As expected the presence of specific binding in the reference region produces underestimates of BPND. Similarly the VB contribution also produces underestimates of BPND. When the reference and target tissues are described by a one tissue compartment and the VB contribution is small then the BPND underestimation can be corrected by the PRTM. However when either (or both) reference and target tissues are described by a two compartment model then both SRTM and PRTM can produce biased estimates of BPND. Nevertheless, even when the full model assumptions are not met there are situations where BPND estimates remain minimally biased. The bias characteristics of the SRTM and PRTM were preserved when noise was added to the simulations. Conclusions It is important to characterize the performance of SRTM and PRTM for individual tracers to determine when they may be applied quantitatively