Cognitive Assessment and Neuroimaging (CAN) Core E

SUMMARY/ABSTRACT: Cognitive Assessment and Neuroimaging Core E
The goal of Cognitive Assessment and Neuroimaging (CAN) Core E, is to provide neurological data with
cognitive assessment, MRI, and carotid ultrasound that will directly support Projects 1, 2 and 4, leading to the
identification of key neurological signatures of individual differences in cognitive aging. Cognitive assessment
protocols will include performance measures of memory, executive functions, processing speed, premorbid
function, and overall cognitive function. Cognitive tests derive from the experimental cognitive aging literature
that have been demonstrated to be sensitive to individual differences across and within age groups, as well as
standardized clinical neuropsychological tests typically used to identify age-related cognitive impairment and
potential early dementia. We will standardize and coordinate online cognitive assessments in both English and
Spanish, directly supporting Projects 1 and 2, and in-person assessments at four clinical for Project 2
(Tucson, Baltimore, Atlanta, Miami) to ensure quality and consistency throughout the project duration.
In addition, we plan to acquire brain MRI and ultrasound carotid images from 1620 participants. Our MRI
protocols will be built upon the advanced MRI protocols of ADNI 3, which have been optimized and tested for
our MRI platforms. From the acquired data we will produce quantitative measures of brain morphology, white
matter hyperintensities, structural and functional connectivity, perfusion, microbleeds, carotid intima-media
thickness, plaque presence, size and morphology, and blood flow velocities. The acquired imaging data will be
annotated with meta-data that support near real-time quality control, robust mining, query and analyses of
imaging data, meeting the meta-data requirements for various MRI software packages, machine learning
procedures, and large scale analyses proposed in Project 4. The meta-data management and annotation
system will also support incorporation of decentralized data (from the public domain), with which the training of
our machine learning modules can be further enhanced. Furthermore, we plan to implement a series of data
harmonization and pre-processing pipelines in our XNAT server (integrated with the COINS platform; HPC
nodes; and CyVerse), to streamline imaging data QC and quantitative analysis. Our proposed pipelines
address limitations of existing software packages (e.g., existing challenge in robustly segmenting hippocampus
and other critical brain structures across age groups) through machine learning, and can inherently harmonize
multi-modal MRI data (e.g., fast spin-echo MRI in minimally distorted coordinates; and echo-planar imaging in
distorted coordinates), enabling streamlined analysis (e.g., from hippocampal segmentation to memory
network connectivity analysis) without manual intervention.

Grant Number: 5U19AG065169-05
NIH Institute/Center: NIH
Principal Investigator: NAN-KUEI CHEN