Abstract
2305
Objectives The cerebellar cortex, often used as the primary reference region to evaluate PET amyloid burden, profoundly impacts all standardized uptake value ratio (SUVR) measures. However, it is vulnerable to numerous sources of artifact that must be considered in obtaining and interpreting results. The goal of our study was to characterize sources of cerebellar reference region artifact in a multi-center data set (ADNI) and the impact upon SUVR results, with the goal of providing practical guidelines for analysis.
Methods We examined cerebellar attributes of 226 PIB scans from 103 ADNI subjects imaged at 13 imaging sites with 6 different PET scanner models. Scans were co-registered within subject and spatially normalized to a common tissue probability map. A full gray matter cerebellum VOI was applied. Scans were categorized as having truncation within reference region bounds, coincidence of reference region with edge of field-of-view, or location of the reference region at least 1 cm above the image edge. The signal intensity in gray matter cerebellum was also measured at each of 27 discrete z-axis slices to evaluate the impact of partial truncation and variability across scanner models.
Results Of the 226 scans, 4% had truncation causing sampling outside tissue, and 12% additional scans had reference region overlap with scan edge where artifact is known to occur. Sampling of truncated cerebellum without boundary correction altered SUVR values by ≥15%. Use of different degrees of reference region boundary reduction resulted in SUVR variability reaching >20%. Scanner models differed with regard to lower and upper slice variance and in the ratios of subcortical white matter to cerebellar cortex.
Conclusions Quality control of reference region sampling, coupled with sampling of a consistent region from scan to scan within subject, is of key importance particularly in threshold cases of amyloid positivity and in the measurement of longitudinal progression.
Research Support This study was supported by Janssen Alzheimer Immunotherapy Research & Development, LLC, and Wyeth Research, which was acquired by Pfizer Inc. in October 2009