Amyloid-β Imaging with Pittsburgh Compound B and Florbetapir: Comparing Radiotracers and Quantification Methods

Cortical retention ratios are shown for consecutive PiB and florbetapir scans obtained for same participants and processed using pons (A), whole cerebellum (B), and cerebellar gray matter (C) for intensity normalization. Initial diagnosis at enrollment and any subsequent diagnostic change are represented by shape markers. Stable normal cognition or MCI diagnosis is represented with solid shapes, individuals who progressed are represented with unfilled shapes, and single individual who regressed from MCI to normal is represented with gray-filled square. Raw data were analyzed with PET-template method. Regression equations and Spearman rank correlation coefficients (ρ) are shown for each scatterplot.

Cortical retention ratios for 2 consecutive PiB scans and PiB scan followed by florbetapir scan obtained for same participants are compared at different levels of preprocessing and data analysis methods. All cortical retention ratios were normalized using whole cerebellum. Top 2 rows show raw data (not at uniform voxel size or smoothing) (A) and unsmoothed data (at uniform voxel size but not uniform smoothing) (B), both processed using PET-template method. Bottom 2 rows show unsmoothed (C) and smoothed data (uniform voxel size and smoothing) (D) processed with Freesurfer method. Thus, middle 2 rows both show unsmoothed data that differs only on basis of which processing method was used (PET template [B], Freesurfer [C]). Regression equations and Spearman rank correlation coefficients (ρ) are shown for each scatterplot.