Impact of different strategies on spatial normalization of F-18-fallypride: Head-to-head comparison of MRI-based and PET-based methods

1545 Objectives F-18-fallypride (FP) is a dopamine D2/D3 receptor ligand with high affinity that allows also imaging of extra-striatal receptors. Parametric binding potential images (BPnd) of FP show in cortex only negligible binding. Stereotactic normalization of FP BPnd images is feasible by using integral images (add images) with more cortical information and by applying the transformation parameters to the BPnd images. The aim of the study was to test if SPM’s MRI-based DARTEL method (DM) may be more accurate for striatal normalization compared to the PET-to-PET method (PM) based on a ligand-specific D2 template. Methods 19 young healthy volunteers underwent dynamic FP PET and high resolution structural MRI. Grey matter template from 19 segmented MRI datasets was constructed using DARTEL toolbox. Parameters from MRI-based normalization were then applied to each of the 19 FP add images which were prior coregistered to MRI. In PM a previously constructed ligand-specific D2 template was used. Mean images of DM and PM normalizations were calculated. Volume-of-interests were defined to quantify accuracy of striatal normalization. Results Striatal normalization was more accurate using the PET-to-PET method. The maximum covariance in striatum was 13.9 % (DARTEL: 18.0 %). Moreover, the volume of the whole striatum in mean images of DM normalization was 22 % larger than in PM normalization indicating a higher anatomical variation of the striatum with the MRI-based DARTEL normalization. Conclusions Inaccurate anatomical normalization in striatum leads to an increase of the groups’ variances and thus, to a loss of sensitivity in voxel-wise statistical tests. Therefore, the normalization method with the smallest anatomical variation, the PET-to-PET method, should be used for striatal normalization before performing voxel-wise statistics of FP images.