Gregory Klein, Joël Schaerer, Florent Roche, Mehul Sampat, Gennan Chen, Joyce Suhy Bioclinica, Newark, CA, USA; Bioclinica, Lyon, France
Longitudinal analyses of amyloid PET data using a standard uptake value ratio (SUVR) and volume of interest (VOI) approach implicitely assume that the VOIs sample comparable regions at all imaging time points. It is known, however, that considerable volumetric changes can take place during a one-two year interval between PET imaging time points in a clinical trial. Annual cerebral atrophy rates of 2.37% have been reported for the Alzheimer's disease (AD) group, and 0.41% for a control group1. These volumetric changes can not only change the size and shape of each region, but also can change the relative partial volume effects of white matter and CSF areas adjacent to grey matter within a sampling region. A native space approach using VOIs separately defined from time-matched MRI data can help to minimize differences in VOI shape and size, however, a partial volume approach is needed to address differences in relative partial volume contributions. This work investigates the extent of change in partial volume between time points in subjects with AD, Mild Cognitive Impairment (MCI) and Normal Controls (NC).
Freesurfer (X-sectional / version 5.3) was used to obtain VOI segmentations on T1 MRI data from 270 Alzheimer's Disease Neuroimaging Initiative (ADNI) subjects (23 with probable AD, 145 with MCI and 102 NC) at two time points approximately 24 months apart. Following the partial volume correction (PVC) approach of Müeller-Gärtner2, Freesurfer segmentations at baseline (TP1) and follow-up (TP2) were used to produce two three-tissue segmentations of grey matter(GM), white matter(WM) and CSF for each subject. Segmentations were smoothed to the same resolution of the PET and the partial volume contributions of each tissue type were computed for each Freesurfer region. Partial volume components of the composite and regional cortical areas were computed, and also for the reference regions to separately analyse longitudinal behaviour of the numerator and denominator components of the SUVR. Florbetapir PET data at matching time points were registered to the MRI data in T1 native space, and composite SUVR's were computed using a grouping of four larger cortical regions, equally weighted as described by Landau3. Twelve potential SUVR reference regions were evaluated including whole cerebellum, cerebellar grey (CG), corpus callosum (CC) and subcortical white matter (SWM). SUVRs were computed with and without PVC.
Results indicate that changing partial volume effects in the cortical target regions and reference regions used in a VOI-based SUVR analysis can bias longitudinal results. Care must be taken in interpreting SUVR changes that may be influenced by extent of volumetric changes, particularly in the AD subject group.
- Fox et al. Arch Neurol 2000;57:339-344.
- Müeller-Gärtner et al. J Cereb Blood Flow 1992, 12:571-583.
- Landau et al. J Nucl Med 2013; 54:1–8.