The scans were acquired on a 1.5 T scanner using a three-dimensional (3D) sagittal magnetization-prepared rapid gradient-echo imaging sequence, which was specially adjusted for the US-ADNI protocols (http://adni.loni.ucla.edu/research/protocols/mri-protocols/). Repetition time (TR), echo time (TE), inversion time (TI), and flip angle were 9.2 msec, 40 msec, 225 msec, and 8°, respectively. The in-plane resolution was 256 × 256 (1.25 × 1.25

mm) with a slice thickness of 1.2 mm. Image analysis of 11C-PIB PET Data analyses of 11C-PIB PET were performed using the PMOD software package (version 3.0; PMOD Technologies, Ltd., Zürich, Switzerland). Distribution volume ratio images referenced to the cerebellum were generated Inhibitors,research,lifescience,medical using noninvasive Logan graphical analysis (Price et al. 2005). Two experts in neuro-nuclear medicine, both with over 10 years of experience, interpreted the regional β amyloid load, focusing Inhibitors,research,lifescience,medical on whether it was consistent with a diagnosis of AD. Gray AZD8055 ic50 matter extraction from brain MRI In statistical parametric mapping 8 (SPM8) (http://www.fil.ion.ucl.ac.uk/spm), we use the default segmentation parameters for MR images because this program is originally developed for MRI images; with very light regula-rization, warp frequency cut-off of 25 Hz,

a shorter sampling distance of 3, and a customized number of Gaussians per tissue class for each patient: 2 for gray and white matter, 2 for cerebrospinal fluid, and 4 for other tissues. The Inhibitors,research,lifescience,medical MR images were then segmented to gray matter, white matter, cerebrospinal fluid, and other compartments using an unmodified version of the clustering algorithm (Ashburner and Friston 2000). Gray matter extraction from brain CT We changed many default setting to the segmentation program Inhibitors,research,lifescience,medical in SPM8 taking the difference of CT and MR into account. Before using the segmentation function in SPM8, MRIcro (http://www.cabiatl.com/mricro)

and Image J (http://rsb.info.nih.gov/ij) were used to preprocess the CT images. The Brain Inhibitors,research,lifescience,medical Extraction Tool (Smith 2000) in MRIcro was used to remove the head holder segment. Image J was used to make the bounding box and voxel sizes equivalent to the tissue probability maps in SPM8. In SPM8, we set the segmentation parameters with extremely heavy regularization for unbiased CT images, a larger warp frequency cut-off of 35 Hz, a shorter sampling distance of 2, and a customized number of Gaussians per tissue class for each patients: very 1 or 2 for gray and white matter and 6–8 for cerebrospinal fluid and other tissues. The number of Gaussians per tissue class was adjusted for each patient until successful segmentation was achieved. The CT images were then segmented to gray matter, white matter, cerebrospinal fluid, and other compartments using an unmodified version of the clustering algorithm (Ashburner and Friston 2000) (Fig. ​(Fig.11). Figure 1 CT-based VBM procedure. (A) A slice from an original CT image. (B) Gray matter extracted from (A) using the segmentation module in SPM8.