Effect of data processing on the accuracy of biologically relevant measures in high resolution pet brain imaging using the HRRT

High resolution positron esmission tomotgraphy (PET) imaging using depth of interaction decoding (DOI) schemes, such as available on the high resolution research tomography (HRRT), is particularly susceptible to (i) potential instrumentation instability and (ii) statistical noise in the data and (iii) patient movement during scanning. These potential sources of uncertainty are exacerbated in dynamic imaging where a large range of count rate and number of acquired counts is observed and scan duration extends over 60 min or more. Here we evaluated the impact of scanner stability and patient motion on the feasibility of obtaining rather sophisticated biological measures such as intervention derived changes in synaptic dopamine levels Δ(DA). It was found that (i) observed variations in an empirical parameter accounting for the different energy response of the two scintillator layers (LSO/LYSO) and influencing the accuracy of the scatter correction may lead to up to 10% variations in binding potentials (BPND) estimates; frequent scanner calibrations are thus required; (ii) coregistration is essential when estimating Δ(DA) in small structures given typical patient motion. Up to ∼25% change in the BPND values was observed after accurate coregistration; this is larger than an 8–10% variation expected from the statistical quality of the data. With these corrections in place we found that data from the HRRT can robustly distinguish topological variation in Δ(DA) occurring as a consequence of different stimuli.