103 / 198
B6-24
SHARED NEUROIMAGING DATA OF MILITARY AND
BLAST TRAUMATIC BRAIN INJURY USING A DATA
SHARING REPOSITORY
Justin Senseney
, Rich Hammett, Terrence Oakes, Gerard Riedy
Walter Reed National Military Medical Center, National Intrepid
Center of Excellence, Bethesda, USA
Objective:
We are sharing neuroimaging data (N
=
300) through the
Federal Interagency Traumatic Brain Injury Repository (FITBIR) so
that academic researchers have access to our unique military popu-
lation for secondary and meta-analysis.
Methods:
Subjects receive two 45-minute scans on a GE Signa 750
3T MRI scanner with a 32-channel phased array head coil. This in-
cludes conventional structural volumetric brain imaging of T1, T2,
flair, and post-contrast imaging. Subjects also receive perfusion, MR
spectroscopy, susceptibility-weighted, diffusion tensor, and both
resting and task-based functional imaging. Subjects also elect to un-
dergo a brain CT scan and PET scan using F-18 fluorodeoxyglucose.
In addition to demographic and prior injury information, the SF-36
Health Survey, Neurobehavioral Symptom Inventory (NSI), Combat
Exposure Scale (CES), and PTSD Checklist (PCL-C) are shared. We
use a custom tool shared with FITBIR to submit these data. These data
provide military-specific information that we hope spurs increased
research on TBI in the military, and provide many of the same
measures that are collected in civilian studies.
Results:
We chose subsets of our neuroimaging subjects to share at
an accelerated schedule (N
=
300) to spur academic research in mili-
tary TBI. These are curated into subsets ideal for resting-state and
other modalities. These data are the first in FITBIR with CES, PET,
spectroscopy, and annotated task-based fMRI data; we are working
with FITBIR to develop methods to query these data. These data will
be the first large military TBI dataset in FITBIR.
Conclusion:
Researchers throughout the TBI community now have
access to a large and diverse military clinical TBI population for
secondary and meta-analysis. We anticipate that this will spur in-
creased study of military and blast TBI at civilian research centers,
and augment smaller research studies. While data currently being
submitted to FITBIR have a long schedule for data sharing, we are
making these datasets widely available within FITBIR for increased
research in military TBI.
Keywords: database, FITBIR, neuroimaging, military, blast
B6-25
CONNECTOME-SCALE ASSESSMENT OF BRAIN NETWORK
CONNECTIVITY IN MILD TRAUMATIC BRAIN INJURY
Zhifeng Kou
1
, Armin Iraji
1
, Hanbo Chen
2
, Natalie Wiseman
1
, E Mark
Haacke
1
, Robert Welch
1
, Tianming Liiu
2
1
Wayne State University, Biomedical Engineering and Radiology,
Detroit, USA
2
University of Georgia, Computer Science, Athens, GA
Most mild traumatic brain injury (mTBI) patients have normal
findings in clinical neuroimaging despite their constellation of
clinical and neurocognitive symptoms after injury that significantly
impact their quality of life. Mounting evidence suggests alterations
in brain functional connectivity (FC). Identifying FC alterations on a
large scale or connectome-scale can provide us a better under-
standing of the network substrates of brain injury, which can po-
tentially assist physicians to select appropriate treatments. In this
longitudinal study, FC of the brain has been evaluated on a large
scale. Diffusion tensor and resting state functional magnetic reso-
nance imaging (fMRI) data were acquired for 24 healthy subjects at
two time points with a 6-weeks interval and 16 mTBI patients at
acute and sub-acute stages at Detroit Receiving Hospital. A novel
prediction framework was utilized to identify 358 network nodes,
known as dense individualized common connectivity based cortical
landmarks (DICCCOLs), distributed all across the whole brain. The
location of each DICCCOL was identified based on the white matter
fiber connection profile optimized from fiber tractography. Each
DICCCOL preserves structural and functional properties across in-
dividuals with maximum group consistency. The longitudinal sta-
tistical analysis was performed using a mixed design analysis of
variance (ANOVA) and Network-Based Statistic (NBS) to identify
disrupted FCs, known as connectomic signatures. The group effect
identified 258 FCs significantly affected in mTBI patients. All
connectomic signatures showed increased FC in the patient group.
These 258 connectomic signatures were further categorized using
meta-analysis and a data-driven approach, called multiview spectral
cluster analysis. Meta-analysis reveals that the interaction between
‘‘Action’’ and ‘‘Cognition’’ functional domains, specifically the in-
teraction between ‘‘Execution’’ (from ‘‘Action’’) and ‘‘Attention’’
(from ‘‘Cognition’’) are affected the most. Categorizing con-
nectomic signatures using a clustering approach identified that the
general pattern of FC changes could be related to a Posterior-
Anterior compensatory mechanism of the brain.
Keywords: Connectome, Functional connectivity, Large-scale brain
networks, mild traumatic brain injury
B6-26
VALIDATION OF DUAL-INJECTION PERFUSION IMA-
GING: A PILOT STUDY
Natalie Wiseman
1
, Mahmoud Zeydabadinezhad
2
, Meng Li
3
, Jessy
Mouannes-Srour
3
, Yongquan Ye
3
, E. Mark Haacke
2,3
, Zhifeng Kou
1–3
1
Wayne State University, Department of Psychiatry and Behavioral
Neurosciences, Detroit, USA
2
Wayne State University, Department of Biomedical Engineering,
Detroit, USA
3
Wayne State University, Department of Radiology, Detroit, USA
Traumatic brain injury (TBI) could render deficits in cerebral blood
flow and metabolism, measurable by advanced imaging techniques.
Dynamic susceptibility contrast perfusion weighted imaging (DSC-
PWI) suffers from blooming, clipping, and saturation effects in large
vessels, which make of arterial input function (AIF) determination
unreliable. To combat this, we have used a dual-injection method, in
which 1/6 of the contrast dose is used to determine the AIF and the
remaining 5/6 is used to visualize the perfusion in brain tissue. We
compared cerebral blood flow (CBF) measurements to those from
pseudo-continuous arterial spin labeling (pCASL), an MR method
with good reliability, validated against positron emission tomography
(PET). Two subjects underwent imaging with pCASL and DSC-PWI.
DSC-PWI was performed twice (with 1/6 dose, then with 5/6). AIF
was generated from the low-dose PWI scan, scaled up, and applied to
the high dose data. The pcASL CBF maps were coregistered to these
and regions of interest (ROIs) drawn, each in a single tissue type. The
correlation coefficients of 0.85 and 0.63 show good correlation in
large ROIs and slightly poorer correlation in the small ROIs, re-
spectively. PWI analysis performed with only the high dose data and
no AIF prediction showed a worse R
2
of 0.76 in the large ROIs. The
improvement of R
2
between DSC-PWI and pCASL with the low-dose
A-67