Abstract
145
Objectives To implement a PET-only method for automated quantification of [18F]flutemetamol imaging data.
Methods To allow for registration of PET negative and positive scans an adaptive template registration method was implemented. Data from the [18F]flutemetamol Phase II study (n=72) was used to model uptake. Linear regression of voxel intensities on the SUVR value in a cortical composite region for all scans gave an intercept image and a slope image. An adaptive template image spanning the range from negative to positive was generated through a linear combination of these two images and the optimal template was selected as part of the registration process. The PET-based method was compared to MR-based spatial normalization using SPM8. Furthermore, spatially normalized [18F]flutemetamol Phase II scans were quantified using a VOI template. For comparison we used FreeSurfer to segment each subject’s MRI and parcellations were applied to the co-registered PET. We then correlated SUVR values for a composite region obtained with both methods. Furthermore, we applied the method to baseline AIBL [11C]PIB scans (n=287) and compared the PET-based composite score with the corresponding score obtained using a semiautomatic method that made use of the subject’s MR. All SUVR values were computed with pons as the reference region.
Results Spatial normalization was successful on all scans. The PET-only registration method and SPM had similar performance in the cortical regions but SPM had more variability in the cerebellum. For quantification of [18F]flutemetamol, there was a strong correlation between the PET-only method and FreeSurfer (r=0.98). We obtained a similar correlation for the AIBL data (r=0.98).
Conclusions The derived adaptive template registration method allows for robust, accurate and fully automated quantification of [18F]flutemetamol and [11C]PIB scans without the use of MRI