Page 66 - NNS 2014 Program & Abstract Book

A 48-year-old man was found unresponsive after a bicycle ac-
cident. A right subdural hemorrhage (SDH) prompted urgent
hemicraniectomy. The subject was consented and research MRI
was obtained on days 2, 7, and 101. The patient expired seven
months post-injury. Whole brain images were acquired (7T MRI)
after formalin fixation. The brain was sliced and selected for em-
bedding in paraffin based on MRI findings and large format tissue
sections were mounted on oversized glass slides. Sections were
stained using H&E, LFB H&E, and Perls iron. Using the 20x ob-
jective, histopathology sections were digitized with 32,000 image
tiles in the entire tissue section using a microscope scanner. The
resulting image files were compressed as hierarchical JPEG files
and viewed using proprietary software to confirm abnormalities
detected on MRI with lesions identified on the histological section
from the tissue specimen.
In addition to SDH and subarachnoid hemorrhages, MRI showed
temporal lobe contusions, linear-appearing hemorrhagic lesions in
bilateral frontal lobes, microbleeds, and meningeal injury. Based
on this evidence, the superior frontal gyrus was chosen as the
region of interest. The results demonstrate use of acute MRI
findings to target regions of interest for pathology. In-vivo imaging
revealed a complexity of injury: pre-existing, primary injury due to
acute event, and secondary progressive injury. From pathology
alone, it would be difficult to reconstruct the timing of injury.
Our approach integrates clinical data along with clinical and
postmortem MRI that improves interpretation of the pathological
progression.
Key words
acute injury, histopathology, imaging, MRI, pathology
A2-27
CROSS-SECTIONAL VOLUMETRIC COMPARISON OF MILD
AND MODERATE TRAUMATIC BRAIN INJURY
McEntee, J.E.1,2, Chou, Y.Y.1,2, McNally, S.M.1,3, Chan, L.1,3,
Butman, J.A.1,2, Pham, D.L.1,2
1Center for Neuroscience and Regenerative Medicine (CNRM), Be-
thesda, USA
2Radiology and Imaging Sciences, Clinical Center, NIH, Bethesda,
USA
3Rehabilitation Medicine Department, Clinical Center, NIH, Bethes-
da, NIH
Neuroanatomical differences across classifications of TBI severity
remain poorly understood. In this work, we examine the structural
brain differences between moderate and mild TBI using both voxel-
based morphometry (VBM) and regional volumes. Participants with
TBI ranging from the sub-acute to the chronic stage underwent MRI
during which high resolution (1 mm isotropic) T1-weighted images
were acquired. Mean ages of the mild group (N = 16, M = 6) and the
moderate group (N = 21, M = 16) were 43 (SD = 15.75) and 47
(SD = 17.66) respectively. Average time since injury in months was
8.56 for the mild group (SD = 7.81) and 10.43 months (SD = 5.22) for
the moderate group. TBI classification was determined by a physician
based on the DoD/VA Common Definition of Traumatic Brain Injury.
Image analysis was performed using 1) SPM8 with the VBM8 toolbox
and 2) FreeSurfer 5.3 for regional volumes. The VBM analysis con-
sisted of comparisons of gray matter (GM), white matter (WM) and
cerebrospinal fluid (CSF) using 2-sample t-tests with age, time since
injury and gender as nuisance variables. No significant differences
were found between the groups in WM. However, the VBM analysis
of GM showed reduced volumes in the moderate patients at the
cluster-level (p < 0.001 unc.) for the right postcentral gyrus ( pFWE =
0.002) and hippocampus ( pFWE = 0.002), and for the amygdala
( p FWE = 0.046). Lateral ventricles (p < 0.001 unc.) were also larger
in the moderate group, with cluster-level significance at pFWE =
0.000. The FreeSurfer results corroborated that the moderate group
had significantly smaller amygdala ( p < 0.01) and hippocampi
( p < 0.001), with larger lateral ventricles ( p < 0.001). These differ-
ences in structure, particularly in the amygdala and hippocampus,
suggest potential damage to pathways involved in memory and
emotion, two areas that are commonly implicated in TBI. Future
work will investigate these volumetric measures in relation to neu-
robehavioral measures.
Key words
cross-sectional, segmentation, TBI, TBI classification, voxel based
morphometry
A2-28
LATERAL VENTRICLE VOLUME ASYMMETRY PREDICTS
MIDLINE SHIFT IN SEVERE TRAUMATIC BRAIN INJURY
To´th, A.1, Schmalfuss, I.2, Heaton, S.C.2, Gabrielli, A.2, Hannay, H.J.3,
Papa, L.7, Brophy, G.M.4, Wang, K.K.2, Bu¨ki, A.1 , Schwarcs, A.1,
Hayes, R.L.5, Robertson, C.S.6 , Robicsek, S.A.2
1University of Pe´cs, Pe´cs, Hungary
2University of Florida, Gainesville, USA
3University of Houston, Houston, USA
4Virginia Commonwealth University, Richmond, USA
5Banyan Biomarkers, Inc, Alachua, USA
6Baylor College of Medicine, Houston, USA
7Orlando Regional Medical Center, Orlando, USA
Midline shift following traumatic brain injury (TBI) detected on acute
CT scans is an established predictor of poor outcome. We hypothe-
sized that lateral ventricular volume (LVV) asymmetry is an earlier
sign of developing asymmetric intracranial pathology than midline
shift. Since lateral ventricle size has individual variability, LVV ratio
(LVR) was used.
A retrospective analysis on 84 adults with blunt, severe (GCS < 8)
traumatic brain injury (sTBI), requiring a ventriculostomy was per-
formed. Seventy-six of these patients underwent serial CTs within
3 hrs and had an average of 3 scans within the first 10d of sTBI. LVVs
were quantified by computer assisted, manual volumetric measure-
ments. LVR and midline shift were determined on initial CT to
quantify asymmetry. The relationship between the initial LVR and the
development of midline shift was assessed.
Sixty percent (15/25) of the patients with high LVR (> 1.8) had
0-5mm midline shift on initial CT, while 40% (15/25) had significant
(> 5mm) midline shift. Eight of 15 patients with 0-5mm initial midline
shift developed significant midline shift on follow-up scans. For sig-
nificant midline shift development, an odds ratio of 6.2 (p < 0.01) was
yielded in the high LVR group compared to patients with low ( < 1.8)
LVR (7/45 patients developed midline shift).
We propose that LVR captures LVV asymmetry and is not only
related to, but predicts midline shift at initial CT examination.
Lateral ventricles may have a higher ‘‘compliance’’ than midline
structures to develope asymmetric brain pathology. This analysis
is simple, rapidly accomplished and may allow earlier interven-
tions to minimize midline shift and potentially improve ultimate
outcomes.
Key words
outcome, prediction
A-34