Abstract
1595
Objectives: To optimize clinical [11C]PIB PET imaging by a quantitative approach.
Methods: Distribution volume ratio (DVR) has been used for in vivo measurement of [11C]PIB retention of amyloid-β. Based on an equation derived from a simplified reference tissue model, DVR can be estimated by weighted linear regression (WLR) for the dynamic PET scanning from T0 to T0+ΔT post tracer injection. For a given ΔT, T0 was optimized (optT0) within [0 90-ΔT] by maximizing the correlation coefficient (R) between DVR([0 90]) and DVR([T0 T0+ΔT]), where the DVRs were estimated from ROI time activity curves (TACs) on [0 90] and [T0 T0+ΔT]. Sixty (50 for healthy controls, 10 for subjects with mild cognitive impairment (MCI)) dynamic PET scans (90 min, 37 frames) from the BLSA were collected (Zhou et al., 2007 Neuroimage). Regions of interest (ROIs) were defined on coregistered MRI's. ΔT was specified as 20, 30 and 40 min, relevant to optimizing clinical imaging. To reduce noise-induced bias and variance, a generalized ridge regression with spatial constraint algorithm (GRRSC) generated DVR images. Images generated by GRRSC for full 90 min were used as a reference.
Results: One optT0 (=9 min) was obtained for the 3 specified ΔTs, and the corresponding maximum Rs were 0.93, 0.95 and 0.95 for ΔT = 20, 30, and 40 min, resp. The ROIs (striatum, lateral temporal, frontal, parietal, cingulate) in the DVR([9 40]) images for the MCI group were (30%, 29%, 47%, 31%, 44%) higher (p<0.001) than controls. The mean ROI differences in DVR images between controls and the MCI group were similar among the optimized and conventional 90 min imaging protocols.
Conclusions: The DVR images generated by the GRRSC with an optimized imaging protocol are accurate for quantitative imaging [11C]PIB in brain for early diagnosis of Alzheimer’s disease.
- Society of Nuclear Medicine, Inc.