The aim of this study was to test the quantitation accuracy of three-dimensional PET in brain scanning.
Methods: Three-dimensional data from 11 human subjects were tested using 11C-dihydrotetrabenazine, 11C-Schering 23390 and 18F-FDG as tracers. Two-dimensional scans were performed on the same subjects and the distribution volume, distribution volume ratio and local metabolic rate of glucose (LMRGlu) values obtained from these were used as reference. Three-dimensional data were processed as follows: iterative convolution subtraction scatter correction, detector normalization including radial and axial geometric factors, attenuation correction extracted from a two-dimensional transmission scan, Kinahan-Rogers reconstruction and region-of-interest-based sensitivity calibration.
Results: No major systematic differences between the two methods were found. The agreement between the two-dimensional and three-dimensional data was within 5%. Although statistical analysis generally did not show this difference to be significant, reliability analysis indicated that comparing two-dimensional and three-dimensional data might introduce some inaccuracies.
Conclusion: Three-dimensional PET yields quantitatively valid results for brain scanning.