Abstract
102
Objectives Partial volume correction (PVC) can be applied to brain PET scans in order to improve quantitative accuracy of measured radioactivity concentrations (AC). Dynamic PET scans, however, are analyzed using pharmacokinetic modeling. The aim of this study was to evaluate effects of various PVC methods on both measured AC and derived pharmacokinetic parameters.
Methods Dynamic [11C]PIB scans were acquired for healthy controls (n=14) and Alzheimer patients (n=17) using an EXACT HR+ PET scanner. T1-weighted MRI scans were used for coregistration and volume of interest (VOI) delineation. PVC was performed using Meltzer and Rousset methods, various implementations of the Müller-Gärtner (MG) method, and an iterative deconvolution (ITDEC) method. Time activity curves (TACs) obtained after PVC were analyzed using the simplified reference tissue model, yielding non-displaceable binding potential (BPND). In addition, dynamic PET scans were simulated based on gray and white matter segmentation, TACs and PSF (7mm) from the clinical scans to assess performance of PVC against ‘true’ simulated values.
Results All MR based PVC methods showed an increase in AC and BPND for both clinical and simulated data, whereas the ITDEC method primarily resulted in decreased BPND. Simulations showed that MR based PVC methods resulted in overestimations, which were moderate for AC (up to 116%, no PVC: 85-99%), but substantial for BPND (up to 161%, no PVC: 55-108%) and which depended on the VOI considered. The results point towards incorrect assumptions about gray and white matter spill-over. Both clinical and simulated data indicated that kinetic analyses without PVC provided more accurate results than those with PVC.
Conclusions None of the PVC methods listed were able to improve quantitative accuracy of BPND for [11C]PIB, despite the fact that measurement of AC may have been improved.
Research Support CTMM AIRFORC