and Department of Neurology, Amsterdam Neuroscience, VU University Medical Center, Amsterdam, Netherlands; 3 University of Washington, Seattle, WA, USA; 4 Department of Radiology and Nuclear Medicine, Amsterdam Neuroscience, VU University Medical Center, Amsterdam, Netherlands; 5 Indiana Alzheimer Disease Center, Indianapolis, IN, USA; 6 Stanford University School of Medicine, Palo Alto, CA, USA. Contact e-mail: radear@iu.edu Background: The increasing number of neuroimaging software ap- plications and settings available has left researchers with the need to determine reproducibility between collaborating sites and methods that optimize study effect size. Our goal was to compare hippocampal values from multiple versions and settings within SPM and Freesurfer across three sites. Methods: Voxel based morphometry (VBM) was performed on T1-weighted MPRAGE images at the Indiana University School of Medicine (IUSM) from 20 cognitively normal older adults (CN) and 20 Alzheimer’s (AD) patients from the ADNI dataset with SPM versions 8 and 12 using DARTEL and prior alignment to a standardized template with or without modulation. We also ran Freesurfer v5.3 and v6 on the same individuals. Bilateral mean hippocampal gray matter values (from VBM) and volumes (Freesurfer) were extracted and effect size (Cohen’s d) of CN>AD were calculated using SPSS v24, covaried for age, sex, and ICV. Finally, VU University Medical Center processed images from 20 CN and 19 AD patients using SPM8 and SPM12 with DARTEL and modulation, and the Univer- sity of Washington processed these same scans using Freesurfer v5.3 and v6. Results were compared to those from IUSM. Results: The largest VBM effect size for CN>AD was observed using either SPM8 new segment or SPM12 segment (Cohen’s d ¼ 2.18; Table 1). Hippocampal volume from Freesurfer v5.3 had a higher effect size (Cohen’s d ¼ 2.01) than Freesurfer v6 (Cohen’s d ¼ 1.83). Hip- pocampal grey matter values were highly correlated between sites (Figure 1) for both versions of SPM (SPM8: r 2 ¼ 0.997; SPM12: r 2 ¼ 0.992) and both versions of Freesurfer (v5.3: r 2 ¼ 0.985; v6: r 2 ¼ 0.986). Conclusions: Hippocampal grey matter values from the new- est version of the SPM segmentation (SPM8 New Seg or SPM12 Seg) was found to have the greatest effect size between AD and CN, while Freesurfer v5.1 showed a higher effect size than v6 for hippocampal volume. Overall, we found strong agreement across sites, which is reassuring for the robustness of these analytic pipe- lines. As software upgrades replace previous versions, widely-used image processing methods need to be carefully reassessed. P3-378 EFFECT OF AGE, GENDER, ETHNICITY, COGNITION AND APOE GENOTYPE ON AMYLOID LOAD AMONG NORMAL, MCI AND MILD DEMENTIA SUBJECTS USING [F-18] FLORBETABEN Ranjan Duara 1,2,3 , David A. Loewenstein 4 , Gabriel Lizarraga 2 , Malek Adjouadi 2 , Warren W. Barker 1 , Maria T. Greig-Custo 1 , Ailyn Penate 1 , Kevin Hanson 3 , Michael Marsiske 3 , Shanna L. Burke 2 , Nilufer Ertekin-Taner 5 , David Vaillancourt 3 , Steve DeKosky 3 , Susan de Santi 6 , Todd E. Golde 3 , 1 Mount Sinai Medical Center, Miami Beach, FL, USA; 2 Florida International University, Miami, FL, USA; 3 University of Florida, Gainesville, FL, USA; 4 University of Miami Miller School of Medicine, Miami, FL, USA; 5 Mayo Clinic, Jacksonville, FL, USA; 6 Piramal Imaging Inc., Boston, MA, USA. Contact e-mail: Ranjan.Duara@msmc. com Background: We utilized visually rated amyloid PET scans to deter- mine the quantitative threshold of amyloid positivity for [F-18] Florbetaben PET scans and to determine the inter-relationships among amyloid load, age, ethnicity, gender, severity of cognitive impairment and ApolipoproteinE (APOE) genotype. Methods: Two experienced raters, blinded to clinical and genetic status of the subjects, conducted visual ratings on PET scans (inter-rater reli- ability was 100% for amyloid positive and 93% for amyloid nega- tive) for 152 participants (mean age 71.1 67.6 yrs) who were Cognitively Normal (n¼46), Mild Cognitive Impairment (n¼68) and Mild Dementia (n¼38). Quantitative amyloid load was measured with Standardized Uptake Value Ratios (SUVRs), refer- enced to the cerebellar gray matter and averaged for five cortical re- gions (frontal, temporal, parietal, anterior cingulate and posterior cingulate). Results: By visual rating, 10% of CN, 40% of MCI and 69% of Dementia subjects were amyloid positive. Mean SUVRs were 1.22 6 0.16 for CNs, 1.41 6 0.27 for MCI, and 1.59 for Dementia. Cognition assessed with the MoCA was strongly Table 1 Hippocampal Effect Sizes for CN>AD (Cohen’s d) SPM Version Modulated Alignment DARTEL Cohen’s D SPM8 Segment No Yes Yes 1.899 SPM8 Segment Yes Yes Yes 2.103 SPM8 New Segment No Yes Yes 1.984 SPM8 New Segment Yes Yes Yes 2.185 SPM12 Old Segment No Yes Yes 1.899 SPM12 Old Segment Yes Yes Yes 2.103 SPM12 Segment No Yes Yes 1.984 SPM12 Segment Yes Yes Yes 2.185 Fig 1. Association of Hippocampal Values from IUSM, VuMC (VBM), and UW (Freesurfer) Poster Presentations: Tuesday, July 24, 2018 P1237