PT  - JOURNAL ARTICLE
AU  - Philipp T. Meyer
AU  - David Elmenhorst
AU  - Andreas Matusch
AU  - Oliver Winz
AU  - Karl Zilles
AU  - Andreas Bauer
TI  - <sup>18</sup>F-CPFPX PET: On the Generation of Parametric Images and the Effect of Scan Duration
DP  - 2006 Feb 01
TA  - Journal of Nuclear Medicine
PG  - 200--207
VI  - 47
IP  - 2
4099  - http://jnm.snmjournals.org/content/47/2/200.short
4100  - http://jnm.snmjournals.org/content/47/2/200.full
SO  - J Nucl Med2006 Feb 01; 47
AB  - 8-cyclopentyl-3-(3-18F-fluoropropyl)-1-propylxanthine (18F-CPFPX) is a novel PET ligand for in vivo quantification of cerebral A1 adenosine receptors. The present study investigated the applicability of voxelwise graphical analysis to the generation of parametric images of the total volume of distribution (DVt) of 18F-CPFPX as a prerequisite for voxel-by-voxel statistical analysis. The benefit of spatial smoothing for reduction of noise-dependent negative bias in graphical analysis was examined. Additionally, the effect of scan duration on the accuracy of analyses based on volumes of interest (VOIs) and individual voxels was explored. Methods: Ten healthy male volunteers underwent bolus-injection 18F-CPFPX PET (90 min). The data were analyzed using a 2-tissue-compartment model and graphical analysis. Voxelwise graphical analysis was performed with and without preceding spatial gaussian smoothing. Results: Voxelwise graphical analysis yielded high-quality parametric images. However, voxelwise graphical analysis suffered from a negative bias (mean bias in cortical regions, −9.6% to −7.5%), which could be attenuated considerably by spatial smoothing (using a kernel of 5 mm in full width at half maximum, bias of −4.0% to −1.9%). Shortening the total scan duration to 60 min had minor effects on the accuracy of VOI-based analysis (15 VOIs × 10 subjects); the resulting error only occasionally exceeded ±5% in individual regions (n = 6 for 2-tissue-compartment model, n = 10 for VOI-based graphical analysis, always within ±7.5%). Quantification accuracy was acceptable with scan durations of 60 and 75 min in voxelwise graphical analysis with spatial smoothing (mean bias in cortical regions, −8.4% to −5.9%) and without spatial smoothing (bias, −9.2% to −11.6%), respectively. Conclusion: High-quality DVt parametric images of 18F-CPFPX can be generated by voxelwise graphical analysis. The noise-dependent negative bias of voxelwise graphical analysis is greatly reduced by spatial smoothing. A shortened scan of 60 min will enhance the clinical applicability of 18F-CPFPX PET.