2 Methods

foulchilianΤεχνίτη Νοημοσύνη και Ρομποτική

20 Οκτ 2013 (πριν από 4 χρόνια και 20 μέρες)

184 εμφανίσεις

1

N
EURAL
C
ONNECTIVITY IN
B
IPOLAR
D
ISORDER

Micah Chambers
1
, Andrei Irimia
1
, Carinna Torgerson
1
, John Darrell Van Horn
1
, and Lori Altshuler
2

12/13/2012

1

Laboratory of Neuro Imaging (LONI), Department of Neurology, David Geffen School of Medicine, University of
California Los Angeles, 635 Charles E. Young Drive SW, Suite 225 Los Angeles, CA 90095
-
7334; email:
micah.chambers@loni.ucla.edu

2

De
partment of Psychiatry and Biobehavioral Sciences, UCLA School of Medicine, 300 Medical Plaza, Suite 1544, Box
956968, Los Angeles, CA 90095
-
7057

For possible submission to:



Brain Imaging and Behavior



Frontiers in Psychiatry (http://www.frontiersin.org
/Psychiatry)

2

Abstract

Bipolar Disorder is an extremely complex and diffuicult to diagnose neuropsychiatric syndrome. Treatment efficacy
is unpredictable, making development of new drug therapies all the more uncertain. Despite such complexities, the
exist
ence of statistically significant differences between normal controls and bipolar patients indicate the possibility
that applied computation informatics could be used to identify bipolar symptoms. Though fMRI activation tasks
have shed lighton the problem,

variable results persist, and activation pattern differences depend on functional tasks.
The lack of a single point of problem has induced researchers to look for more complex


and more detailed


methods of analysis. While graph
-
theoretical approaches h
ave been applied to bipolar disorder, here we
simultaneously explore resting state functional and structural deficits in the brain network to elucidate how
connectivity differences within bipolar patients produces differences in the front
-
limbic
-
subcortica
l circuitry. Here,
we examine the Diffusion Tensor Image (DTI)
-
based structural and Functional Magnetic Resonance Imaging
(fMRI)
-
based functional connectivity differences exhibited by euthymic Bipolar Type I subjects. In do doing, we
seek to utilize these
methods to elucidate a more cogent explanation for the observed bipolar deficits. Our analysis
revealed that resting state functional connectivity is a highly discriminative analysis method for bipolar disorder.
Specifically, Euthymic Bipolar I subjects in

resting state had the significant isolation of the Posterior
-
Dorsal
Cingulate from the other network elements. The most immediate implication of this will be the construction of
connectivity
-
derived classifiers for bipolar disorder. Given the difficulty o
f appropriately diagnosis affective
disorders, being able to systematically identify dysfunctional a default mode network may be the key to
differentiating a complex set of illnesses.

1 Introduction

Bipolar Disorder (BD) is a debilitating psychiatric dis
order that causes swings between psychological states of
mania, depression and euthymia (otherwise beging asymptomatic). Sufferers of Bipolar Disorder have lower quality
of life [
50
], and drastically increased suicide

rate [
29
] contributing to a 10 year mean reduction in life expectancy.
Further, these individuals have triple the health care costs, double the work absence compared to average working
age individual [
27
], and with lifetime prevalence approaching 4%, the cost of BD to society is climbing. Yet despite
the urgent need for better diagnosis and understanding, the dynamic of nature of BD continues to elude simple
explanation. Just as th
e disease is difficult to treat due to the variance in drug effectiveness and symptomatology, the
exact imaging correlates are proving difficult to assess as well. While genetic heritability [
49
] indicates a
physiolog
ical problem, no single biokmarker has been found for BD. However, recent evidence points to a variety of
small functional and structural differences from Normal Controls (NCs) which somehow produces a time
-
varying
neurological imbalance.

3

Neuroanatomicall
y, bipolar disorder shares certain features of schizophrenia: namely irregular white matter
within the uncinate fasciculus [
66
,
67
], and greater amygdala volume (albeit greater
than schizophrenia) [
5
,
12
].
However, BD subjects also show white matter irregularities in the internal capsule, anterior thalamic radiation [
66
,
16
], between the Amygdala and Subgenual Cingulate, within frontal lobe white matter [
7
,
16
], [
36
,
67
],
periventricular white matter [
64
,
61
], corona radiata [
17
], be
tween hemispheres (reduced corpus callosal density)
[
44
,
67
] and within the inferior longitudinal fasciculi [
67
]. Hence, there is good eviden
ce that differences in neural
connectivity play a role in BD, although certain aberrations may be caused by the progression of the disorder itself.
Sassi et al. [
56
] also showed that taking lithium prevents global decrea
se in the grey matter normally found in aging
bipolar subjects. There is also evidence for decreased Grey Matter (GM) within specific regions: in the Subgenual
Prefrontal Cortex [
34
], the Dorsal Lateral Prefrontal Cor
tex [
21
], and superior parietal lobule [
3
]; and increased GM
in other regions: the amygdala [
5
,
12
],

anterior cingulate cortex, ventral prefrontal cortex, fusiform gyrus [
3
],
striatum [
7
], caudate [
22
], and putamen [
17
]. However Rossa et al. [
55
] found decreased amygdala volume after the
first manic episode, seemingly indicating that the amygdala could increase in volume over the course of the illness.
The most consistent stru
ctural abnormalities not correlated with disease progression are is an enlarged Striatum [
61
]
and reduction in white matter, although the locations differ [
55
,
16
,
17
]. The myriad of anatomical results, at times
contradictory, leaves much to be desired in identifying the underlying cause of BD.

Indeed, simple structural analysis of brain morphometry may not be s
ufficient; as a disorder that changes over
time, functional analysis is just as, if not more, important in explaining how BD works. Unfortunately, differences in
experimental setup across research groups and studies creates a large number of confounds. Mix
ed bipolar type I
and type II patients, different functional tasks, varied implementations of the same task (for instance happy vs. sad
facial affect), subjects’ perception of the task, medication level, varied or unknown mood state, and other minor
lab
-
sp
ecific patient protocols all have effects on outcomes, and are all found in real bipolar studies. Several reviews
have attempted to assess the heterogeneity of the results between bipolar studies, which is particularly important
when comparing and contrast
ing euthymia, depression and mania. Functional deficits vary across mood and brain
region and both hyper
-

and hypo
-
activity are exhibited across prefrontal and temporal brain regions; however a few
consistencies arise across all three states. There is a ge
neral agreement of a decrease in the Pre
-
Frontal Cortex (PFC)
activation (although precise localizations vary) and a generalized excitability of the amygdala


which euthymic
subjects seem to be able to normalize at times [
61
,
53
,
42
,
68
]. Increased subcortical activity is also consistent [
15
],
especially to
negative stimuli [
53
]; which has been widely attributed to decreased modulation by the frontal cortex
[
61
,
42
,
68
]. Despite a common focus on frontal brain regions, Kupferschmidt et al.’s meta
-
analysis found that lower
cuneus, and higher lingual gyrus activity have two of the largest effect sizes seen within bipolar subjects


indicating
the poss
ibility of a distributed brain
-
wide problem outside of prefrontal and anterior limbic network [
42
].

Behaviorally euthymic bipolar subjects still exhibit group differences from NCs during executive function,
verba
l learning and memory, and attention tasks [
25
,
54
,
57
]. These differences point to the possibility that, in fact,
4

there may be no truly euthy
mic state for Bipolar type I patients. Instead, Functional Magnetic Resonance Imaging
(fMRI) has revealed differences in cognitive function in remitted bipolars which indicate compensatory activity, for
instance through increased performance monitoring (hy
peractive Posterior Medial Cingulate Cortex activity during
errors) [
57
] or through different neural processing strategy [
62
], although adding task complexity (via stroop
interfe
rence) may overcome compensatory behavior or activity [
63
]. However, during working memory [
43
] and
emotional stroop [
47
] tasks bipolar
patients show worse performance and less activation indicating that euthymic
bipolars do not always make up for their deficits.

As the majority of published neuroimaging data suggests, BD is extremely complex, which makes diagnosis
difficult: often long te
rm observation is required to verify symptoms


symptoms which also overlap with other
mood disorders. Moreover, treatment efficacy is unpredictable, making development of new drug therapies all the
more uncertain. Despite such complexities, the existence
of statistically significant differences between normal
controls and bipolar patients indicate the possibility that applied computation informatics could be used to identify
bipolar symptoms. While fMRI activation tasks have shed light on the problem, vari
able results persist, and
activation pattern differences depend on functional tasks. The lack of a single point of problem has induced
researchers to look for more complex


and more detailed


methods of analysis. While graph
-
theoretical
approaches have b
een applied to bipolar disorder [
44
,
67
], here we simultaneously explore resting state functional
and structural deficits in the brain network to elucidate how connectivity differen
ces within bipolar patients
produces differences in the front
-
limbic
-
subcortical circuitry [
63
,
42
]

In the present article we examine the Diffusion Tensor Image (DTI)
-
base
d structural and fMRI
-
based functional
connectivity differences exhibited by euthymic Bipolar Type I subjects. In so doing, we seek to utilize these
methods to elucidate a more cogent explanation for the observed pattern of bipolar deficits.

2 Methods

2.
1 Data Acquisition

The University of California, Los Angeles Institutional Review Board approved the study protocol and prior written
consent was given by each participant. 35 euthymic (asymptomatic) bipolar type I subjects (Age 41.3±12.1 years, 18
male,
17 female, ) were recruited through the UCLA Mood Disorders Outpatient Clinic, local newspapers and
campus fliers. 32 age and gender matched normal controls (Age 39.8±11.6 years, 15 male, 17 female) were recruited
through local advertising. Diagnosis of ea
ch bipolar subject was confirms using Structured Clinical Interviews
consistent with DSM
-
IV (SCID) [
26
]. Potential bipolar subjects were excluded if they showed symptoms of any
other current Axis I disorders. Control sub
jects were excluded if they ever had been diagnosed with a psychiatric
illness, were taking medication or had a history of substance abuse. Additionally, potential bipolar and control
5

subjects were excluded for left left
-
handedness, hypertension, neurologi
cal illness, metal implants, and a history of
head trauma with loss of consciousness for more than 5 min.

A subset of the subjects also participated in the resting state portion of the study including 13 of the controls
(Aged 41.8±12.1 years, 8 male, 5 fem
ale) and 14 of the euthymic bipolars (Aged 40.9±12.03 years, 8 male, 6
female). Although 33 subjects were initially gathered 6 were excluded for motion artifacts or FOV problems. During
resting state, subjects were instructed to keep their eyes closed but
not to fall asleep or think of anything in
particular. All T1
-
weighted MPRAGE, fMRI and DTI volumes were acquired at the Ahmanson
-
Lovelace Brain
Mapping Center at the UCLA David Geffen School of Medicine, using a Siemens Magnetom Trio TIM 3 Tesla MRI
scann
er. The DTI scan acquired 64 non
-
colinear directions as well as B0 and B1000 image volumes having a TR of
8,400 ms and a TE of 93ms. Slices were 2mm thick with a FOV of 190mm giving a voxel size of 2×2×2mm.

2.2 Processing

All methods rely on labelmaps in

the subjects’ anatomical space. To ensure that tools work together properly, the
LONI Pipeline is being used for most processing [
1
]. Skull stripping and segmentation of the MPRAGE T1 image
will be calculated using Frees
urfer 5.1 [
2
]. Diffusion Weighted Images were eddy
-
current corrected with FSL, and
each subjects’ Fractional Anisotropy (FA) image was linearly registered to the skull
-
stripped T1 anatomical image
using the multimodal

mutual
-
information based registration tool BRAINSFit [
24
]. Fiber tracts were generated using
the Diffusion Toolkit (http://trackvis.org, which uses the Fiber Assignment by Continuous Tracking (FACT)
algorithm to perform d
eterministic tractography [
51
,
70
]. The FA

T1 linear transform was then applied to
white
-
matter tractography using the Visualization Toolkit (vtk.org). The connection weight betwee
n pairs of regions
were set to the number of tracts connecting them (at the endpoints), a method used extensively in literature [
58
,
33
,
35
].

This method provided a complete weighted connection matrix between every pair of Freesurfer’s a2009 brain
structures.

The fMRI data were motion corrected using the AFNI 3dvolreg tool [
19
]. To bring each subjects’ labelma
p into
fMRI space, the T1 anatomical images were registered to the middle fMRI volume using BRAINSFit [
24
]. We
manually inspected both the motion correlation results as well as the anatomical registration to insure no moti
on
artifacts remained. Prior to calculating Functional Connectivity (FC), the average ventricular signal, average cortical
white matter signal and the average outside
-
brain signal were be regressed out. Finally, a Butterworth bandpass
filter was applied be
tween a period of 120s (0.0083Hz) and 6.6s (.15Hz).

To calculate functional connectivity a mass
-
scale seed
-
based [
10
] analysis in the patients’ anatomic space was
performed using Shannon Mutual Information (MI) [
18
]. While MI does not provide inversely
-
related (negative)
correlations, it provides a better measure of overall connectivity, especially in non
-
linear/non
-
Gaussian signals. To
make computation of
15
,
000
2

signal

comparisons tractable, OpenCL [
41
] was used on an NVIDIA Tesla M2070
6GB GPGPU located at the Laboratory of Neuro Imaging (GPGPU). Following full
-
brain connectivity; the pairs of
6

voxels within every pair regions was avera
ged to compute average connectivity between each pair of graymatter
freesurfer a2009s labels. identify group differences.

2.3 Statistical Analysis

Differences were calculated for structural and functional connectivity, regional morphometric, and regional

graph
-
theoretic metrics. All between group comparisons have been controlled for age, sex, and brain width.
Similarly within group averages have been controlled for these confounds using a linear model:


S
control

0
+0β
bipolar
+0
.5β
sex

width
β
width

age
β
age

(1)


S
bipolar

0
+1β
bipolar
+0.5β
sex

width
β
width

age
β
age

(2)

where each β is the estimated parameter (effect),
μ
w
idth

is the group average width (69.07 mm), and
μ
age

is the
average age (40.58) Brain width was estimated using the distance between centroids of left and right middle frontal
gyri.

2.4 Multiple Comparison Correction

While it is
best to search broadly when analyzing a poorly
-
understood illness such as bipolar disorder, the
large
-
scale, exploratory nature of this research creates a multiple comparison problem. With 148 regions,
Bonferroni corrected Family
-
Wise Error Rate (FWER)
p
<0
.05 mandates a Per
-
Comparison Error Rate (PCER) of
p
<0.000298. For per
-
connection comparisons 979 structural connections were tested between groups (connections
not present in at least half the subjects were excluded) whereas 14,028 null
-
hypotheses were te
sted for functional
connectivity. With such large numbers of tests, only huge datasets or effect sizes could be found at
Bonferroni
-
corrected thresholds. Therefore False Discovery Rate (FDR) was applied to control the FWER while
still providing sufficient
power with the number of samples available.

False Discovery Rate is a method of controlling type 1 error (false positives) by placing an upper bound on the
percent of false discoveries (falsely rejected null hypotheses) out of the total number of discover
ies (rejected
null
-
hypothesis) [
9
]. Thus the FDR threshold is inversely proportional to the number of discoveries rather than the
number of tests (which is the case for FWER). Thus controlling the FDR <0.05 accepts t
hat up to 5% of the positives
may be false positives; if there are 1000 rejected null hypothesis then up to 50 could be false, if there is 1 then there
is a 5% chance that it is false. The gain in power makes it ideal for exploration of new theories on a l
arge scale;
which is why it has been used to widely in genetics [
4
,
13
], neuroscience [
48
] and neuroimaging [
6
]. While the two
step threshold estimation method outlined in Benjamin et al. controls the FDR, it does so conservatively. By
estimating the distribution of p
-
values, it is possible to estimate more accurate FDR levels; thus for very large
numbers of hy
pothesis tests (>500) this method provides more power while still maintaining the same FDR. Thus
7

for between group comparisons of connections, we use the R package
fdrtool

to compute the p
-
value threshold
consistent with a local FDR <0.05 [
65
]. Again this form of FDR control is popular in genetics where thousands of
comparisons are being performed, and conventional FWER would be destructively stringent [
20
,
45
].

Because of the variety of tests in this study (connection
-
wise, region
-
wise and network
-
wise), no single solution
would suffice; therefore, in connection
-
wise statistics where there are enough p
-
values (>500) to estimate an
underlying distrib
ution the more powerful method of
fdrtool

is applied [
65
], but in brain
-
region statistics the more
traditional method of Benjamini et al. is used [
9
]. For statistics that apply
to whole brains only a single t
-
test is being
performed; thus, no multiple comparison issue exists.

Because of these difficulties, each section of the results will be reported in raw p
-
values (PCER) but every
reported discovery is below the local FDR thres
hold of 0.05. For connection level tests where
fdrtool

is used
histograms of p
-
values are shown as well to show the model (Fiber Count:
5
, Mean FA:
6
, Functional Connectivity:

9
).

3 Results

Normal and bipolar structural graphs are shown in
2

and
3
, respectively. The outermost ring color indicates the
region named and matches the corresponding region in
1
. The next 4 rings indicate the regions’ (GM) volume,
surface area, cortical thickness, and curvature; as in [
38
]. Extending the representation for connectivity developed
by Irimia et al., graph statistics are shown in the innermost 5 rings in the order: node strength, betweenness
centrality, eccentricity, nodal efficiency, and
eigenvector centrality [
38
]. Finally, as in Irimia et al., the color of each
line indicates the FA levels; where blues FA average through track less than 0.15, and red indicates greater 0.2, with
green in between. This
same layout is used to compare FA and functional connectivity in
4

and
10
, respectively. In
difference graphs, red indicates supra
-
normal statistics in bipolar subjects,
and blue indicates sub
-
normal.

3.1 Differences in Cortical Thickness

The only morphometric group
-
wise differences to pass an FDR threshold was average cortical thickness (
2
): bipolar
subjects showed decreased cortical thi
ckness bilaterally in the Middle
-

Anterior part of the Cingulate. Additionally
the precentral gyrus, central sulcus, subcentral gyrus and inferior precentral sulcus all on the right side showed
significant decreases in cortical thickness. The opercular par
t of the right inferior frontal gyrus and Left Lateral
occipto
-
temporal sulcus also showed decreased thickness.

3.2 Structural Connectivity

While the average strucural graphs show counts and FA, and are consistent with with previous normal brain graphs
p
ublished in Imiria et al. and Vah Horn et al., [
38
,
37
,
69
] no significant difference in
number

of fiber tracts
8

connecting regions showed signi
ficant differences after FDR correction
5
. However the Right Marginal Branch of
the Cingulate Sulcus did show significant decreases in FA
6
.

3.3 Functional Connectivity

As wi
th Structural Connectivity (SC), euthymic bipolar subjects do not show catastrophic differences in
connectivity from normal controls (
8

vs.
7
); however, there are significant
differences between the two groups (
10
).
The pattern of FC in both groups matches the expected pattern of the Default Mode Network, including overlapping
early
-
visual occipital lobe networks (Cuneus, Calcarine Sul
cus, Occipital Pole, Superior Occipital Gyrus), a largely
detached frontal lobe network (Transverse Frontopolar Gyri, Frontomarginal Gyri), and a highly connected ventral
Posterior Cingulate Cortex (PCC) [
31
,
30
,
59
]. Qualitatively, bipolar subjects show far more minimally attached
regions where in normal controls some connectivity was present; specifically within the temporal lobe. This follows

a trend of reduced connectivity in the bipolars in this resting state task.

In fact, the connectivity is heavily skewed toward significant differences in the FDR analysis
6
. Too many
regions were significantly differ
ent to list, thus there are massive differences in connectivity within the Default
Mode Network (DMN); particularly in the dorsal PCC. Thus, functional connectivity may be a more reliable method
of discriminating bipolar subjects from normal controls, when

compared to SC. Identical color scales are used in
both functional and structural difference graphs.
10

reveals a general decrease in functional connectivity during
eyes
-
closed resting state in Bipolar Subjectss
(BSs) when compared with NCs. The right posterior
-
dorsal cingulate
showed significantly reduced FC with the caudate nucleus, the Vertical Ramus of the Anterior Lateral Sulcus, the
Transverse Frontopolar Gyrus, the short Insular Gyri, Jensen’s Sulcus, the s
ubcallosal gyrus, the putamen,
fronto
-
marginal gyrus, the suborbital sulcus, and the Nucleus Accumbens (NAcc). Note that due to the smaller
sample size, the structural morphometry (volume, cortical thickness, curvature, surface area) differ between
10

and
4
.

4 Discussion

In the present study n=32 control and n=35 BD I euthymic subjects were analyzed for structural connectivity
differences, and of those subjects n=13 NC an
d n=14 BS underwent resting state functional connectivity analysis.
The results show a marked decreased in functional connectivity, but only one multiple
-
comparison corrected
differences in structural connectivity (in the Cingulate).

The GM morphometric re
sults agree with prior results that have shown decreased grey matter volume in right
inferior frontal regions [
46
]; here we found a decrease in the opercular part of the inferior frontal gyrus. Oddly, when
compare
d with typical analysis, not other regions in the frontal or prefrontal regions passed multiple
-
comparisons
testing. Although the mixture of results in the literature indicate that differences may depend on the patient.

9

Cortical thickness decreased within
the left occipital lobe


including the lingual gyrus


which builds on
previous evidence that the cuneus and lingual gyrus are also robust discriminators of bipolar disorder, per se [
42
].
This is further evidenc
ed by the decreased FA between the right marginal branch of the cingulate sulcus and the right
precuneus.

With a some variance, the general consensus of existing literature seems to indicate that bipolar disorder is
accompanied by targeted 1) shrunken pref
rontal volume with decreased activity 2) enlarged amygdala and increased
activity 3) abnormal limbic activity [
42
,
15
,
63
]. This has
led to the theory that the anterior limbic network is to
blame for bipolar disorders’ symptomatology. The present results seem to indicate that the cingulate (limbic) region
is not working with frontal or subcortical regions during resting state in the way

that it does in normal controls.
These results are consistent with normal posterior dorsoal cingulate activity accompanied by abnormal frontal and
striatal activity, or vice versa. Although effect size in task
-
free analysis is difficult to assess, only du
ring a go/no
-
go
task have there been reports of altered (increased) precuneus/posterior cingulate activity [
42
]; this implies that the
problem may not be in the Posterior
-
Dorsal Cingulate Gyrus (PosDCgG), but wit
h some DMN mechanism
connecting PosDCgG to the putamen, nucleus accumbens, caudate nucleus, insula, and prefrontal regions. Similar
dissociation of the prefrontal cortex and the posterior cingulate have been reported during self
-
reflection tasks [
40
],
suggesting that bipolar subjects may tend to fall into a more self
-
referential state than their normal peers. Given the
lack of a matching set of SC deficiency, it is unlikely that this difference is caused by the underlyi
ng white matter
connectivity but is rather an abnormal brain state. That said, both the precuneus and the marginal branch of the
cingulate are part of the default mode network; thus a deficit in FA could explain alteration of the default mode
network. If t
he drop in FA found in current study indeed are the cause of the resting state differences, then it shows
that the DMN is highly sensitive to focused network disconnections.

Hypo
-
connectivity across the hemispheres has been implicated as a potential reason

for euthymic bipolar’s
inability to smoothly transition between emotional states [
44
]. There is functional, thought not structural evidence of
inter
-
hemispheric dysfunction, consistent with that of Leow et al. [
44
] and Torgerson et al. [
67
] with a threshold of
p
<0.05; nearly three quarters of the significant differences in resting state connectivity are inter
-
hemispheric. These
previous studies wer
e hypothesis based and therefore implicitly more powerful than the current large scale
observations. The current structural results

Previous structural connectivity results also provides another potential explanation for increased amygdala
activity across

bipolar states. Recent work has shown that the left temporal pole is responsible for semantic memory
associated with high level sensory representations [
52
]. Significant increases in the temporal pole
-
amygdala
connectio
n indicates one of two possibilities: 1) more consistent paired activity between these two regions has
triggered neural plasticity, created a strengthening of white matter connectivity, or 2) that the increased connectivity
versus controls in fact causes h
igher than normal input of high
-
level sensory input to the amygdala, thereby inducing
irregular activity. It has been reported that BD subjects responds respond more emotionally to non
-
emotional
10

stimulus [
62
,
71
]; this increased connectivity could explain this hyper
-
sensitivity. Similarly, significantly increased
connectivity between high
-
level ventral visual stream (inferior temporal gyrus) and the right temporal pole


a
region a
ssociated with personal memories and interactions [
52
]


could either be the cause or effect of
over
-
emotionalization of stimuli. Decreased superior temporal sulcus (STS)/Supra
-
Marginal Gyrus (SMG)
connectivity was also
present in the bipolar subjects. While the STS is activated in relation to emotional attention,
gaze detection and auditory/voice integration [
8
,
23
,
28
], the SMG is related to phonological language perception
and reading [
14
,
60
]; although such a disconnect could explain deficits in verbal memory in euthymic BD subjects

[
54
].

The present study has several limitations. The lower effect size of bipolar on SC when compared to FC means
that more subjects are needed in order to verify hypotheses about how SC is induces the decreased sync
hrony within
resting bipolar subjects. Additionally, subjects had different numbers of episodes and different medication levels.
Decreased gray matter in the left inferior frontal cortex, and enlarged ventricles have both been associated with the
number of

manic episodes. Moreover, medication has been shown to normalize both hyperactivity in the motor
cortex, basal gangia and thalamus as well as phosphoinositol levels in bipolar subjects [
63
]. These confounds may
red
uce effect sizes in fMRI by hiding deficits and affect DTI results by altering perceived diffusion. Unfortunately,
the difficulties recruiting subjects and treating bipolar disorder makes elimination of these confounds difficult if not
impossible. However,

unlike early studies of bipolar disorder, all data collection took place during euthymia with
bipolar II subjects specifically not included [
42
].

Methodologically, structural connectivity is still a developing
field. Fiber reconstruction may be performed in
different ways, and while the FACT algorithm is the method of choice in structural connectivity papers, monte
-
carl
methods may be used to compute fuller connectivity matrices [
32
]. In this work we have corrected for eddy currents
using affine registration; although with extra data collection, much better results may be achieved [
39
,
11
]. Another
d
ifficulty is that counting fibers between pairs of regions may be performed in a variety of ways (nearest region to
endpoints, only count when a fiber passes through, or weighted based proximity), and may depend on the size and
shape of regions used. Here
we have used the Freesurfer a2009 atlas which makes the results easier validate in future
works, although regions are still non
-
uniformly sized.

Future work may build on the present functional differences to explore how other task
-
positive networks may be

affected by bipolar disorder; and how depression and mania further alter the functional network. Identification of the
locus of deficiency means finding the common difference across all situations and stages of the illness. While
additional subjects and p
erspectives on bipolar disorder will increase power and further elucidate the underlying
cause; this study is the first to simultaneously consider how full brain functional and structural connectivity give rise
to network differences in euthymic bipolars.

11

5 Conclusion

This work has revealed group differences between bipolar and normal subjects’ functional and structural networks.
These differences shed new light on how bipolar subjects experience the world and provide new direction for future
research. Sp
ecifically, Euthymic Bipolar I subjects in resting state had the significant isolation of the
Posterior
-
Dorsal Cingulate from the other network elements, whereas the Temporal Pole showed signifigantly
increased connectivity with the amygdala.

Additionally
, this work has shown that resting state functional connectivity is a highly discriminative analysis
method for bipolar disorder. The most immediate implication of the functional connectivity differences in bipolar
subjects will be the construction of conn
ectivity
-
derived classifiers for bipolar disorder. Given the difficulty of
appropriately diagnosis affective disorders, being able to systematically identify dysfunctional a default mode
network may be the key to differentiating a complex set of illnesses.

12

A Figures





Figure 1: Parcelation regions used in circos graphs

13





Figure 2: Structural connectivity of normal controls. Color indicates average FA (Blue, Green, Red
indicate increasing FA), o
pacity indicates number of connectivity fibers.

14





Figure 3
: Structural connectivity of bipolar subjects. Color indicates average FA (Blue, Green, Red
indicate increasing FA), opacity indicates number of connectivity fibers.

15





Figure 4: Difference in FA between Bipolar and Normal Subjects. Red indicates higher FA in bipolars,
blue lower.

16





Figure 5: Histogram and
fdrtool

generated model of null distribution for 979 t
-
tests of fiber count
structural connectivity metric.

17





Figure 6: Histogram and
fdrtool

generated model of null distribution for 979 t
-
tests of Fractional
Anisotropy structural connectivity metric.

18





Figure 7: Functional connectivity of normal controls.

19





Figure 8: Fu
nctional connectivity of bipolar subjects.

20





Figure 9: FDRTool model for functional connectivity scores.

21





Figure 10: Functiona
l differences at significance.

22

B Tables





Bipolar Subjects

Human Controls

2*Structural Connectivity

N =

35
(18M/17F)

N =

32
(15M/17F)


Age
=

41.3±12.1

Age
=

39.8±11.6

2*Functional Connectivity

N
=

15 (9M/6F)

N =

14 (8M/6F)


Age
=

41.1±11.6

Age
=

42.2±11.7



Table 1: Subject statistics

23





Average Cortical Thickness

Region

T
-
Score

p
-
stat

Right Middle
-
anterior part of the cingulate gyrus and sulcus

-
4.142375

0.0001058649

Left Lateral occipito
-
temporal sulcus

-
3.67796

0.0004935913

Right Precentral gyrus

-
3.414519

0.001131659

Right Central sulcus (Rolando’s fissure)
=
J
㌮㌴㐶㌶
=
〮〰ㄴ〲M㐶
=
oig桴⁓畢捥=tr慬⁧yr畳
捥湴ral=敲捵c畭F⁡湤=獵s捩
=
J
㌮㈵㈴㐶
=
〮〰ㄸ㔳M〵
=
oig桴⁏灥pc畬ar⁰=rt=潦=t桥hi湦eri潲=fr潮tal⁧yr畳
=
J
㌮ㄸ㈱㤸
=
〮〰㈲㠴M㤴
=
oig桴⁉湦敲i潲=灡ptf⁴h攠灲散敮tral⁳畬c畳
=
J
㌮ㄶ〵㜵
=
〮〰㈴㌵M㐹
=
䱥it=䵩摤de
J
慮a敲i潲⁰慲t=⁴桥⁣i湧畬at攠eyr畳⁡湤n獵s捵c
=
J
㌮㄰㔲㘸
=
〮〰㈸㘵M㘴
=
=
=
Table 2: Grey
matter cortical thickness differences between bipolar and normal subjects, (
-
) indicates
group decrease in BS.



24



Functional C
onnectivity

Region


Region

T
-
Score

p
-
stat

Right Marginal branch (or part) of the cingulate sulcus


=
oig桴⁐牥c畮敵s
=
J
㐮㜹㐹㐶
=
〮〰〰㄰M㠴㐸
=
=
=
Table 3: Difference in fractional anisostropy (structural connectivity) between bipolar and normal
subjects


(+) indicates increased bipolar connectivity versus controls, (
-
) indicates decreased.



25


Functional Connectivity



Functional Connectivity


Contiued on next page ...



Cros s
-
He mi s phere

Ri ght Pos t e r i or
-
dor s al pa r t of
t he ci ngul at e gyr us



Le f t Ver t i cal r amus of t he ant e r i or s e gme nt of

t he l at e r al s ul cus ( or f i s s ur e)

-
4.709301

0.0001067509

Ri ght Pos t e r i or
-
dor s al pa r t of
t he ci ngul at e gyr us



Le f t Ca udat e nucl e us

-
4.641349

0.0001259788

Ri ght Pos t e r i or
-
dor s al pa r t of
t he ci ngul at e gyr us



Le f t Hi ppoc ampus

-
4.534987

0.0001633075

Ri ght Pos t e r i or
-
dor s al pa r t of
t he ci ngul at e gyr us



Le f t Sul c us i nt er medi us pr i mus ( of J ens en)

-
4.484632

0.0001846743

Ri ght Pos t e r i or
-
dor s al pa r t of
t he ci ngul at e gyr us



Le f t Put a men

-
4.343687

0.0002605902

Ri ght Me di al or bi t al

s ul c us
( ol f act or y s ul cus )



Le f t Pos t e r i or
-
dor s al pa r t of t he ci ngul a t e
gyr us

-
4.089309

0.0004850855

Ri ght Pos t e r i or
-
dor s al pa r t of
t he ci ngul at e gyr us



Le f t Tr ans ver s e f r ont opol a r gyr i a nd s ul ci

-
4.014308

0.0005824701

Le f t Pos t e r i or
-
dor s al

par t of t he
ci ngul a t e gyr us



Ri ght Nucl e us a ccumbe ns

-
4.008663

0.0005905407

Ri ght Pos t e r i or
-
dor s al pa r t of
t he ci ngul at e gyr us



Le f t Nucl e us accumbens

-
3.93763

0.0007021137

Ri ght Pos t e r i or
-
dor s al pa r t of
t he ci ngul at e gyr us



Le f t
Subcal l os al a r ea, s ubcal l os al gyr us

-
3.816707

0.0009420977


Lef t Hemi s phere

Le f t Sul c us i nt er medi us pr i mus
( of J ens en)



Le f t Pos t e r i or
-
dor s al pa r t of t he ci ngul a t e
gyr us

-
4.659336

0.0001205735

Le f t Super i or pa r t of t he
pr ecent r al s ul cus



Le f t Sul c us i nt er medi us pr i mus ( of J ens en)

-
3.814938

0.0009461527


Ri ght Hemi s phere

Ri ght St r ai ght gyr us ( gyr us
r ect us )



Ri ght Pos t e r i or
-
dor s al pa r t of t he ci ngul a t e
gyr us

-
4.61513

0.0001342975

Ri ght Pos t e r i or
-
dor s al pa r t of
t he ci ngul at e gyr us



Ri ght Ant er i or par t of t he ci ngul at e gyr us and
s ul c us

-
4.127776

0.000441611

Ri ght Shor t i ns ul ar gyr i



Ri ght Pos t e r i or
-
dor s al pa r t of t he ci ngul a t e
gyr us

-
3.942076

0.0006945554

Ri ght Subcal l os al a r e a,
s ubcal l os al gyr us



Ri ght
Pos t e r i or
-
dor s al pa r t of t he ci ngul a t e
gyr us

-
3.865973

0.0008358312

Ri ght Me di al or bi t al s ul c us
( ol f act or y s ul cus )



Ri ght Pos t e r i or
-
dor s al pa r t of t he ci ngul a t e
gyr us

-
3.860217

0.0008476061

26

C Shortened Region Names in Connectogram

Connectogram Region Acronyms




Connectogram Region Acronyms


ACgG/S

Anterior part of the cingulate gyrus and sulcus

ACirInS

Anterior segment of the circular sulcus of the insula

ALSHorp

Horizontal ramus of the anterior segment of the lateral sulcus

(or fissure)

ALSVerp

Vertical ramus of the anterior segment of the lateral sulcus (or
fissure)

AngG

Angular gyrus

AOcS

Anterior occipital sulcus and preoccipital notch
(temporo
-
occipital incisure)

ATrCoS

Anterior transverse collateral sulcus

CcS

Calcarine sulcus

CgSMarp

Marginal branch (or part) of the cingulate sulcus

CoS/LinS

Medial occipito
-
temporal sulcus (collateral sulcus) and
lingual sulcus

CS

Central sulcus (Rolando’s fissure)

Cun

Cuneus

FMarG/S

Fronto
-
marginal gyrus (of Wernicke) and

sulcus

FuG

Lateral occipito
-
temporal gyrus (fusiform gyrus)

HG

Heschl’s gyrus (anterior transverse temporal gyrus)

InfCirInS

Inferior segment of the circular sulcus of the insula

InfFGOpp

Opercular part of the inferior frontal gyrus

InfFGOrp

Orbital
part of the inferior frontal gyrus

InfFGTrip

Triangular part of the inferior frontal gyrus

InfFS

Inferior frontal sulcus

InfOcG/S

Inferior occipital gyrus and sulcus

InfPrCS

Inferior part of the precentral sulcus

IntPS/TrPS

Intraparietal sulcus
(interparietal sulcus) and transverse
parietal sulci

InfTG

Inferior temporal gyrus

InfTS

Inferior temporal sulcus

JS

Sulcus intermedius primus (of Jensen)

LinG

Lingual gyrus, lingual part of the medial occipito
-
temporal
gyrus

LOcTS

Lateral
occipito
-
temporal sulcus

LoInG/CInS

Long insular gyrus and central insular sulcus

LOrS

Lateral orbital sulcus

MACgG/S

Middle
-
anterior part of the cingulate gyrus and sulcus

MedOrS

Medial orbital sulcus (olfactory sulcus)

MFG

Middle frontal gyrus

MFS

Middle frontal sulcus

MOcG

Middle occipital gyrus, lateral occipital gyrus

27

MOcS/LuS

Middle occipital sulcus and lunatus sulcus

MPosCgG/S

Middle
-
posterior part of the cingulate gyrus and sulcus

MTG

Middle temporal gyrus

OcPo

Occipital pole

OrG

Orbital

gyri

OrS

Orbital sulci (H
-
shaped sulci)

PaCL/S

Paracentral lobule and sulcus

PaHipG

Parahippocampal gyrus, parahippocampal part of the medial
occipito
-
temporal gyrus

PerCaS

Pericallosal sulcus (S of corpus callosum)

POcS

Parieto
-
occipital sulcus (or
fissure)

PoPl

Polar plane of the superior temporal gyrus

PosCG

Postcentral gyrus

PosCS

Postcentral sulcus

PosDCgG

Posterior
-
dorsal part of the cingulate gyrus

PosLS

Posterior ramus (or segment) of the lateral sulcus (or fissure)

PosTrCoS

Posterior
transverse collateral sulcus

PosVCgG

Posterior
-
ventral part of the cingulate gyrus (isthmus of the
cingulate gyrus)

PrCG

Precentral gyrus

PrCun

Precuneus

RG

Straight gyrus (gyrus rectus)

SbCaG

Subcallosal area, subcallosal gyrus

SbCG/S

Subcentral
gyrus (central operculum) and sulci

SbOrS

Suborbital sulcus (sulcus rostrales, supraorbital sulcus)

SbPS

Subparietal sulcus

ShoInG

Short insular gyri

SuMarG

Supramarginal gyrus

SupCirInS

Superior segment of the circular sulcus of the insula

SupFG

Superior frontal gyrus

SupFS

Superior frontal sulcus

SupOcG

Superior occipital gyrus

SupPrCS

Superior part of the precentral sulcus

SupOcS/TrOcS

Superior occipital sulcus and transverse occipital sulcus

SupPL

Superior parietal lobule

SupTGLp

Lateral
aspect of the superior temporal gyrus

SupTS

Superior temporal sulcus

TPl

Temporal plane of the superior temporal gyrus

TPo

Temporal pole

TrFPoG/S

Transverse frontopolar gyri and sulci

TrTS

Transverse temporal sulcus

Amg

Amygdala

CaN

Caudate nucleus

Hip

Hippocampus

NAcc

Nucleus accumbens

Pal

Pallidum

Pu

Putamen

Tha

Thalamus

28

CeB

Cerebellum

BStem

Brain stem



29

D Acronyms

BD

Bipolar Disorder

................................
................................
.........

1

BS

Bipolar Subjects

................................
................................
.........

6

DMN

Default Mode Network

................................
..............................

6

DTI

Diffusion Tensor Image

................................
.............................

3

FA

Fractional Anisotropy

................................
................................

3

FACT

Fiber Assignment by Continuous Tracking

...............................

4

FC

Functional Connectivity

................................
.............................

4

FDR

False Discovery Rate

................................
................................
.

5

fMRI

Functional Magnetic Resonance Imaging

................................
..

2

FWER

Family
-
Wise Error Rate

................................
.............................

4

GM

Grey Matter

................................
................................
................

2

NC

Normal Control

................................
................................
..........

1

NAcc

Nucl
eus Accumbens

................................
................................
...

6

PCC

Posterior Cingulate Cortex

................................
.........................

6

PosDCgG

Posterior
-
Dorsal Cingulate Gyrus

................................
..............

7

PCER

Per
-
Comparison Error Rate

................................
........................

4

PFC

Pre
-
Frontal Cortex

................................
................................
.....

2

SC

Structural Connectivity

................................
..............................

6

SMG

Supra
-
Marginal Gyrus

................................
...............................

8

STS

superior temporal sulcus

................................
............................

8

30

E Bibliog
raphy

References

[1]

LONI Pipeline.

[2]

Freesurfer, 2012.

[3]

Caleb

M Adler, Ari

D Levine, Melissa

P DelBello, and Stephen

M Strakowski. Changes in gray matter volume in
patients with bipolar disorder.
Biological psychiatry
, 58(2):151

7, July 2005.

[4]

David

B Allison, Xiangqin Cui, Grier

P Page, and Mahyar Sabripour. Microarray data analysis: from disarray to
consolidation and consensus.
Nature reviews. Genetics
, 7(1):55

65, January 2006.

[5]

L

L Altshuler, G

Bartzokis, T

Grieder, J

Curran, T

Jimenez, K

Leight, J

Wilkins, R

Gerner, and J

Mintz. An MRI
study of temporal lobe structures in men with bipolar disorder or schizophrenia.
Biological psychiatry
, 48(2):147

62, July
2000.

[6]

Manzar Ashtari, Kelly

L Cervellione, Khader

M Hasan, Jinghui Wu, Carolyn
McIlree, Hana Kester, Babak

a
Ardekani, David Roofeh, Philip

R Szeszko, and Sanjiv Kumra. White matter development during late adolescence in
healthy males: a cross
-
sectional diffusion tensor imaging study.
NeuroImage
, 35(2):501

10, April 2007.

[7]

E

H Ayl
ward, J

V Roberts
-
Twillie, Patrick

E Barta, Ashok

J Kumar, G

J Harris, Michael Geer, C

E Peyser, and G

D
Pearlson. Basal ganglia volumes and white matter hyperintensities in patients with bipolar disorder.
The American journal
of psychiatry
, 151(5):687

93,

May 1994.

[8]

G.C. Baylis, E.T. Rolls, and C.M. Leonard. Selectivity between faces in the responses of a population of neurons in the
cortex in the superior temporal sulcus of the monkey.
Brain Research
, 342(1):91

102, September 1985.

[9]

Yoav Benjamini a
nd Yosef Hochberg. Controlling the false discovery rate: a practical and powerful approach to
multiple testing.
Journal of the Royal Statistical Society. Series B
, 57(1):289

300, 1995.

[10]

B

B Biswal, J

Van Kylen, and J

S Hyde. Simultaneous assessment of
flow and BOLD signals in resting
-
state
functional connectivity maps.
NMR in biomedicine
, 10(4
-
5):165

70, 1997.

[11]

Nils Bodammer, Jörn Kaufmann, Martin Kanowski, and Claus Tempelmann. Eddy current correction in
diffusion
-
weighted imaging using pairs of im
ages acquired with opposite diffusion gradient polarity.
Magnetic resonance
in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in
Medicine
, 51(1):188

93, January 2004.

[12]

Paolo Brambilla, Keith

Harenski, Mark Nicoletti, Roberto

B Sassi, Alan

G Mallinger, Ellen Frank, David

J
Kupfer, Matcheri

S Keshavan, and Jair

C Soares. MRI investigation of temporal lobe structures in bipolar patients.
Journal of psychiatric research
, 37(4):287

95, July 2003.

[13]

Maria

Stella Carro, Wei

Keat Lim, Mariano

Javier Alvarez, Robert

J Bollo, Xudong Zhao, Evan

Y Snyder,
Erik

P Sulman, Sandrine

L Anne, Fiona Doetsch, Howard Colman, Anna Lasorella, Ken Aldape, Andrea Califano, and
Antonio Iavarone. The transcriptional
network for mesenchymal transformation of brain tumours.
Nature
,
463(7279):318

25, January 2010.

31

[14]

P

Celsis, K

Boulanouar, B

Doyon, J

P Ranjeva, I

Berry, J

L Nespoulous, and F

Chollet. Differential fMRI
responses in the left posterior superior temporal
gyrus and left supramarginal gyrus to habituation and change detection in
syllables and tones.
NeuroImage
, 9(1):135

44, January 1999.

[15]

Michael

a Cerullo, Caleb

M Adler, Melissa

P Delbello, and Stephen

M Strakowski. The functional
neuroanatomy of bipola
r disorder.
International review of psychiatry (Abingdon, England)
, 21(4):314

22, January 2009.

[16]

Wai
-
Yen Chan, Guo
-
Liang Yang, Ming
-
Ying Chia, Puay
-
San Woon, Jimmy Lee, Richard Keefe, Yih
-
Yian
Sitoh, Wieslaw

Lucjan Nowinski, and Kang Sim. Cortical and
subcortical white matter abnormalities in adults with
remitted first
-
episode mania revealed by Tract
-
Based Spatial Statistics.
Bipolar disorders
, 12(4):383

9, June 2010.

[17]

Zhuangfei Chen, Liqian Cui, Mingli Li, Lijun Jiang, Wei Deng, Xiaohong Ma, Qiang
Wang, Chaohua Huang,
Yingcheng Wang, David

a Collier, Qiyong Gong, and Tao Li. Voxel based morphometric and diffusion tensor imaging
analysis in male bipolar patients with first
-
episode mania.
Progress in neuro
-
psychopharmacology & biological
psychiatry
, 3
6(2):231

8, March 2012.

[18]

T.M. Cover and J.A. Thomas.
Elements of information theory
, volume

6. John and Wiley Sons, Inc., 1991.

[19]

R

W Cox and A

Jesmanowicz. Real
-
time 3D image registration for functional MRI.
Magnetic resonance in
medicine : officia
l journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in
Medicine
, 42(6):1014

8, December 1999.

[20]

F

De Vita, M

Orditura, E

Martinelli, L

Vecchione, R

Innocenti, V

C Sileni, C

Pinto, M

Di Maio, a

Farella,
T

Troiani, F

Morgillo, V

Napolitano, E

Ancona, N

Di Martino, a

Ruol, G

Galizia, a

Del Genio, and F

Ciardiello. A
multicenter phase II study of induction chemotherapy with FOLFOX
-
4 and cetuximab followed by radiation and
cetuximab in locally advanced oesophageal cancer
.
British journal of cancer
, 104(3):427

32, February 2011.

[21]

Daniel

P. Dickstein, Michael

P. Milham, Allison

C. Nugent, Wayne

C. Drevets, Dennis

S. Charney,
Daniel

S. Pine, and Ellen Leibenluft. Frontotemporal alterations in pediatric bipolar disorder:
results of a voxel
-
based
morphometry study.
Archives of general psychiatry
, 62(7):734

41, July 2005.

[22]

Renee

M Dupont, Nelson Butters, Kimberly Schafer, Thomine Wilson, John Hesselink, and J

Christian
Gillin. Diagnostic specificity of focal white matter

abnormalities in bipolar and unipolar mood disorder.
Biological
psychiatry
, 38(7):482

6, October 1995.

[23]

Andrew

D Engell and James

V Haxby. Facial expression and gaze
-
direction in human superior temporal
sulcus.
Neuropsychologia
, 45(14):3234

41, Novemb
er 2007.

[24]

Andriy Fedorov, Steven Dunn, and Steve Pieper. BRAINSFit.

[25]

I.

N. Ferrier, B.

R. Stanton, T.

P. Kelly, and J.

Scott. Neuropsychological function in euthymic patients with
bipolar disorder.
The British journal of psychiatry : the journal of

mental science
, 175(3):246

51, September 1999.

[26]

Michael

B. First, Robert

L Spitzer, Gibbon Miriam, and Janet

B.W. Williams.
Structured Clinical Interview
for DSM
-
IV
-
TR Axis I Disorders, Research Version, Patient Edition. (SCID
-
I/P, version 2.0)
. PhD t
hesis, New York State
Psychiatric Institute, New York, 1995.

[27]

Harold

H Gardner, Nathan

L Kleinman, Richard

a Brook, Krithika Rajagopalan, Truman

J Brizee, and
James

E Smeeding. The economic impact of bipolar disorder in an employed population from an e
mployer perspective.
The Journal of clinical psychiatry
, 67(8):1209

18, August 2006.

32

[28]

Asif

a Ghazanfar, Chandramouli Chandrasekaran, and Nikos

K Logothetis. Interactions between the superior
temporal sulcus and auditory cortex mediate dynamic face/voic
e integration in rhesus monkeys.
The Journal of
neuroscience : the official journal of the Society for Neuroscience
, 28(17):4457

69, April 2008.

[29]

Frederick

K Goodwin, Bruce Fireman, Gregory

E Simon, Enid

M Hunkeler, Janelle Lee, and Dennis
Revicki. Suicide risk in bipolar disorder during treatment with lithium and divalproex.
JAMA : the journal of the American
Medical Association
, 290(11):1467

73, S
eptember 2003.

[30]

Michael

D Greicius, Vesa Kiviniemi, Osmo Tervonen, Vilho Vainionpää, Seppo Alahuhta, Allan

L Reiss,
and Vinod Menon. Persistent default
-
mode network connectivity during light sedation.
Human brain mapping
,
29(7):839

47, July 2008.

[31]

Michael

D Greicius, Ben Krasnow, Allan

L Reiss, and Vinod Menon. Functional connectivity in the resting
brain: a network analysis of the default mode hypothesis.
Proceedings of the National Academy of Sciences of the United
States of America
, 100(1):253

8,

January 2003.

[32]

P.

Hagmann, J.
-
P. Thiran, L.

Jonasson, P.

Vandergheynst, S.

Clarke, P.

Maeder, and R.

Meuli. DTI mapping
of human brain connectivity: statistical fibre tracking and virtual dissection.
NeuroImage
, 19(3):545

54, July 2003.

[33]

Patric Ha
gmann, Leila Cammoun, Xavier Gigandet, Reto Meuli, Christopher

J Honey, Van

J Wedeen, and
Olaf Sporns. Mapping the structural core of human cerebral cortex.
PLoS biology
, 6(7):e159, July 2008.

[34]

Y

Hirayasu, M

E Shenton, D

F Salisbury, C

C Dickey, I

a Fi
scher, P

Mazzoni, T

Kisler, H

Arakaki, J

S
Kwon, J

E Anderson, D

Yurgelun
-
Todd, M

Tohen, and R

W McCarley. Lower left temporal lobe MRI volumes in
patients with first
-
episode schizophrenia compared with psychotic patients with first
-
episode affective disor
der and
normal subjects.
The American journal of psychiatry
, 155(10):1384

91, October 1998.

[35]

C

J Honey, O

Sporns, L

Cammoun, X

Gigandet, J

P Thiran, R

Meuli, and P

Hagmann. Predicting human
resting
-
state functional connectivity from structural connecti
vity.
Proceedings of the National Academy of Sciences of the
United States of America
, 106(6):2035

40, February 2009.

[36]

J

Houenou, M

Wessa, G

Douaud, M

Leboyer, S

Chanraud, M

Perrin, C

Poupon, J
-
L Martinot, and M
-
L
Paillere
-
Martinot. Increased white mat
ter connectivity in euthymic bipolar patients: diffusion tensor tractography between
the subgenual cingulate and the amygdalo
-
hippocampal complex.
Molecular psychiatry
, 12(11):1001

10, November
2007.

[37]

Andrei Irimia, Micah

C Chambers, Carinna

M Torgerso
n, Maria Filippou, David

A Hovda, Jeffry

R Alger,
Guido Gerig, Arthur

W Toga, Paul

M Vespa, Ron Kikinis, and John

D Van Horn. Patient
-
tailored connectomics
visualization for the assessment of white matter atrophy in traumatic brain injury.
Frontiers in neu
rology
, 3(February):10,
January 2012.

[38]

Andrei Irimia, Micah

C Chambers, Carinna

M Torgerson, and John D

Van Horn. Circular representation of
human cortical networks for subject and population
-
level connectomic visualization.
NeuroImage
, 60(2):1340

51,
April
2012.

[39]

P

Jezzard, a

S Barnett, and C

Pierpaoli. Characterization of and correction for eddy current artifacts in echo
planar diffusion imaging.
Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in
Medicine / S
ociety of Magnetic Resonance in Medicine
, 39(5):801

12, May 1998.

33

[40]

Marcia

K Johnson, Carol

L Raye, Karen

J Mitchell, Sharon

R Touryan, Erich

J Greene, and Susan
Nolen
-
Hoeksema. Dissociating medial frontal and posterior cingulate activity during self
-
re
flection.
Social cognitive and
affective neuroscience
, 1(1):56

64, June 2006.

[41]

Khronos. OpenCL.

[42]

David

a Kupferschmidt and Konstantine

K Zakzanis. Toward a functional neuroanatomical signature of
bipolar disorder: quantitative evidence from the neu
roimaging literature.
Psychiatry research
, 193(2):71

9, August 2011.

[43]

Jim Lagopoulos, Belinda Ivanovski, and Gin

S Malhi. An event
-
related functional MRI study of working
memory in euthymic bipolar disorder.
Journal of psychiatry & neuroscience : JPN
,
32(3):174

84, May 2007.

[44]

Alex Leow, Olusola Ajilore, Liang Zhan, Donatello Arienzo, Johnson Gadelkarim, Aifeng Zhang, Teena
Moody, John Van Horn, Jamie Feusner, Anand Kumar, Paul Thompson, and Lori Altshuler. Impaired Inter
-
Hemispheric
Integration in B
ipolar Disorder Revealed with Brain Network Analyses.
Biological psychiatry
, pages 1

11, October 2012.

[45]

Jin Li, Limei Wang, Liangde Xu, Ruijie Zhang, Meilin Huang, Ke

Wang, Jiankai Xu, Hongchao Lv, Zhenwei
Shang, Mingming Zhang, Yongshuai Jiang, Maozu
Guo, and Xia Li. DBGSA: a novel method of distance
-
based gene set
analysis.
Journal of human genetics
, 57(10):642

53, October 2012.

[46]

Melissa

P López
-
Larson, Melissa

P DelBello, Molly

E Zimmerman, Michael

L Schwiers, and Stephen

M
Strakowski. Regional p
refrontal gray and white matter abnormalities in bipolar disorder.
Biological psychiatry
, 52(2):93

100, July 2002.

[47]

Gin

S Malhi, Jim Lagopoulos, Perminder

S Sachdev, Belinda Ivanovski, and Ron Shnier. An emotional
Stroop functional MRI study of euthymi
c bipolar disorder.
Bipolar disorders
, 7 Suppl 5:58

69, January 2005.

[48]

Bryce

a Mander, Vikram Rao, Brandon Lu, Jared

M Saletin, John

R Lindquist, Sonia Ancoli
-
Israel, William
Jagust, and Matthew

P Walker. Prefrontal atrophy, disrupted NREM slow waves a
nd impaired hippocampal
-
dependent
memory in aging.
Nature neuroscience
, 16(3):357

64, March 2013.

[49]

Peter McGuffin, Fruhling Rijsdijk, Martin Andrew, Pak Sham, Randy Katz, and Alastair Cardno. The
heritability of bipolar affective disorder and the genet
ic relationship to unipolar depression.
Archives of general
psychiatry
, 60(5):497

502, May 2003.

[50]

Erin

E Michalak, Lakshmi

N Yatham, and Raymond

W Lam. Quality of life in bipolar disorder: a review of
the literature.
Health and quality of life outcomes
, 3:72, January 2005.

[51]

S

Mori, B

J Crain, V

P Chacko, and P

C van Zijl. Three
-
dimensional tracking of axonal projections in the
brain by magnetic resonance imaging.
Annals of neurology
, 45(2):265

9, February 1999.

[52]

Ingrid

R Olson, Alan Plotzker, an
d Youssef Ezzyat. The Enigmatic temporal pole: a review of findings on
social and emotional processing.
Brain : a journal of neurology
, 130(Pt 7):1718

31, July 2007.

[53]

Mary

L Phillips and Eduard Vieta. Identifying functional neuroimaging biomarkers of b
ipolar disorder:
toward DSM
-
V.
Schizophrenia bulletin
, 33(4):893

904, July 2007.

[54]

Lucy

J Robinson, Jill

M Thompson, Peter Gallagher, Utpal Goswami, Allan

H Young, I

Nicol Ferrier, and
P

Brian Moore. A meta
-
analysis of cognitive deficits in euthymic pat
ients with bipolar disorder.
Journal of affective
disorders
, 93(1
-
3):105

15, July 2006.

34

[55]

Isabelle

M Rosso, William D

S Killgore, Christina

M Cintron, Staci

a Gruber, Mauricio Tohen, and
Deborah

a Yurgelun
-
Todd. Reduced amygdala volumes in first
-
episode

bipolar disorder and correlation with cerebral
white matter.
Biological psychiatry
, 61(6):743

9, March 2007.

[56]

Roberto

B Sassi, Mark Nicoletti, Paolo Brambilla, Ala
n

G Mallinger, Ellen Frank, David

J Kupfer,
Matcheri

S Keshavan, and Jair

C Soares. Increased gray matter volume in lithium
-
treated bipolar disorder patients.
Neuroscience letters
, 329(2):243

5, August 2002.

[57]

Gianna Sepede, Domenico De Berardis, Daniel
a Campanella, Mauro

Gianni Perrucci, Antonio Ferretti,
Nicola Serroni, Francesco

Saverio Moschetta, Cosimo Del Gratta, Rosa

Maria Salerno, Filippo

Maria Ferro, Massimo Di
Giannantonio, Marco Onofrj, Gian

Luca Romani, and Francesco Gambi. Impaired sustained

attention in euthymic bipolar
disorder patients and non
-
affected relatives: an fMRI study.
Bipolar disorders
, 14(7):764

79, November 2012.

[58]

Pawel Skudlarski, Kanchana Jagannathan, Vince

D Calhoun, Michelle Hampson, Beata

a Skudlarska, and
Godfrey Pear
lson. Measuring brain connectivity: diffusion tensor imaging validates resting state temporal correlations.
NeuroImage
, 43(3):554

61, November 2008.

[59]

Stephen

M Smith, Karla

L Miller, Steen Moeller, Junqian Xu, Edward

J Auerbach, Mark

W Woolrich,
Christ
ian

F Beckmann, Mark Jenkinson, Jesper Andersson, Matthew

F Glasser, David

C Van Essen, David

a Feinberg,
Essa

S Yacoub, and Kamil Ugurbil. Temporally
-
independent functional modes of spontaneous brain activity.
Proceedings of the National Academy of Scienc
es of the United States of America
, 109(8):3131

6, February 2012.

[60]

Cornelia Stoeckel, Patricia

M Gough, Kate

E Watkins, and Joseph

T Devlin. Supramarginal gyrus
involvement in visual word recognition.
Cortex
, 45(9):1091

6, October 2009.

[61]

S

M Strako
wski, M

P Delbello, and C

M Adler. The functional neuroanatomy of bipolar disorder: a review of
neuroimaging findings.
Molecular psychiatry
, 10(1):105

16, January 2005.

[62]

Stephen

M Strakowski, Caleb

M Adler, Scott

K Holland, Neil Mills, and Melissa

P De
lBello. A preliminary
FMRI study of sustained attention in euthymic, unmedicated bipolar disorder.
Neuropsychopharmacology : official
publication of the American College of Neuropsychopharmacology
, 29(9):1734

40, September 2004.

[63]

Stephen

M Strakowski,
Caleb

M Adler, Scott

K Holland, Neil

P Mills, Melissa

P DelBello, and James

C
Eliassen. Abnormal FMRI brain activation in euthymic bipolar disorder patients during a counting Stroop interference
task.
The American journal of psychiatry
, 162(9):1697

705, Se
ptember 2005.

[64]

Stephen

M Strakowski, Melissa

P DelBello, Molly

E Zimmerman, Glen

E Getz, Neil

P Mills, Jennifer Ret,
Paula Shear, and Caleb

M Adler. Ventricular and periventricular structural volumes in first
-

versus multiple
-
episode
bipolar disorder.
The American journal of psychiatry
, 159(11):1841

7, November 2002.

[65]

Korbinian Strimmer. A unified approach to false discovery rate estimation.
Bmc Bioinformatics
, 9:303,
January 2008.

[66]

Jessika

E Sussmann, G

Katherine

S Lymer, James McKirdy, T

Willi
am

J Moorhead, Susana Muñoz
Maniega, Dominic Job, Jeremy Hall, Mark

E Bastin, Eve

C Johnstone, Stephen

M Lawrie, and Andrew

M McIntosh.
White matter abnormalities in bipolar disorder and schizophrenia detected using diffusion tensor magnetic resonance
imag
ing.
Bipolar disorders
, 11(1):11

8, March 2009.

35

[67]

Carinna

M Torgerson, Andrei Irimia, Alex

D Leow, George Bartzokis, Teena

D Moody, Robin

G Jennings,
Jeffry

R Alger, John

Darrell Van Horn, and Lori

L Altshuler. DTI tractography and white matter fiber tr
act characteristics
in euthymic bipolar I patients and healthy control subjects.
Brain imaging and behavior
, October 2012.

[68]

Jennifer Townsend and Lori

L Altshuler. Emotion processing and regulation in bipolar disorder: a review.
Bipolar disorders
, 14(4
):326

39, June 2012.

[69]

J

D Van Horn, A

Irimia, C

M Torgerson, M

C Chambers, R

Kikinis, and A

W Toga. Mapping Connectivity
Damage in the Case of Phineas Gage.
PLoS ONE
, 13, 2012.

[70]

R

Wang, T

Benner, AG

Sorensen, and VJ

Wedeen. Diffusion Toolkit: A Sof
tware Package for Diffusion
Imaging Data Processing and Tractography. In
International Society for Magnetic Resonance in Medicine
, volume

15,
pages 3720

3720, 2007.

[71]

Michèle Wessa, Josselin Houenou, Marie
-
Laure Paillère
-
Martinot, Sylvie Berthoz, Eric A
rtiges, Marion
Leboyer, and Jean
-
Luc Martinot. Fronto
-
striatal overactivation in euthymic bipolar patients during an emotional go/nogo
task.
The American journal of psychiatry
, 164(4):638

46, April 2007.