Disrupted Axonal Fiber Connectivity As a Marker of Impaired Consciousness States

Abstract

Background: Persistent vegetative states (PVS) and locked-in syndrome (LIS) are well differentiated disorders of consciousness that can be reached after a localized brain injury in the brainstem. The relations of the lesion topography with the impairment in the whole-brain
architecture and functional disconnections are poorly understood. Methods: Two patients (PVS and LIS) and 20 age-matched healthy volunteers were evaluated using diffusion tensor imaging (DTI). Anatomical network was modeled as a graph whose nodes are represented by 71 brain regions. Inter-region connections were quantified through Anatomical Connection Strength (ACS) and Density (ACD). Complex networks properties such as local and global efficiency and vulnerability were studied. Mass univariate testing was performed at every connection using network based statistic approach. Results: LIS patients’ network showed significant differences from controls in the brainstem-thalamus-frontal cortex circuitry, while PVS patients showed a widespread disruption of anatomical connectivity in both hemispheres. Both patients showed a reorganization of network attributes, with decreased global and local efficiency, significantly more pronounced in PVS. Conclusions: Our results suggest that DTI-based network connectivity combined with graph theory is useful to study the long-range effect of confined injuries and the relationship to the degree of consciousness impairment, underlying PVS and LIS.

Full text

Funct Neurol Rehabil Ergon 2013;3(2-3):319-328 ISSN: 2156-941X
© Nova Science Publishers, Inc.
DISRUPTED AXONAL FIBER CONNECTIVITY
AS A MARKER OF IMPAIRED CONSCIOUSNESS STATES
Rafael Rodriguez-Rojas

1,2, Karla Batista1, Yasser Iturria3,
Calixto Machado4, Gerry Leisman5,6,7,8, Robert Melillo5,6,
Mauricio Chinchilla9, Philip DeFina8, Maylen Carballo1,
and Juan M. Morales1
1Brain Imaging Group, International Center for Neurological Restoration, Havana, Cuba
2The Abdus Salam International Centre for Theoretical Physics, Trieste, Italy
3Neuroimaging Department, Cuban Neuroscience Center, La Habana, Cuba
4Institute of Neurology and Neurosurgery, Department of Clinical Neurophysiology,
Havana, Cuba
5F.R. Carrick Institute for Clinical Ergonomics,
Rehabilitation and Applied Neuroscience, Garden City, NY USA
6The National Institute for Brain and Rehabilitation Sciences, Nazareth, Israel
7Biomedical Engineering, ORT-Braude College of Engineering, Karmiel, Israel
8Neuroscience, University of the Medical Sciences of Havana, Faculty Manuel Fajardo
9Neurology Department, Hermanos Ameijeiras Hospital, Havana, Cuba,
10International Brain Research Foundation, New York, USA
ABSTRACT
Background: Persistent vegetative states (PVS) and locked-in syndrome (LIS) are well-
differentiated disorders of consciousness that can be reached after a localized brain injury in the
brainstem. The relations of the lesion topography with the impairment in the whole-brain
architecture and functional disconnections are poorly understood.
Methods: Two patients (PVS and LIS) and 20 age-matched healthy volunteers were evaluated
using diffusion tensor imaging (DTI). Anatomical network was modeled as a graph whose nodes
are represented by 71 brain regions. Inter-region connections were quantified through Anatomical
Connection Strength (ACS) and Density (ACD). Complex networks properties such as local and
global efficiency and vulnerability were studied. Mass univariate testing was performed at every
connection using network based statistic approach.
Results: LIS patients’ network showed significant differences from controls in the brainstem-
thalamus-frontal cortex circuitry, while PVS patients showed a widespread disruption of
anatomical connectivity in both hemispheres. Both patients showed a reorganization of network
attributes, with decreased global and local efficiency, significantly more pronounced in PVS.
 Correspondence: Dr. Rafael Rodriguez-Rojas, International Center for Neurological Restoration, Ave 25 # 15805, CP 11300,
Havana, Cuba. E-mail: rafael.rguez@infomed.sld.cu

Rafael Rodriguez-Rojas, Karla Batista, Yasser Iturria et al.
320
Conclusions: Our results suggest that DTI-based network connectivity combined with graph
theory is useful to study the long-range effect of confined injuries and the relationship to the
degree of consciousness impairment, underlying PVS and LIS.
Keywords: Persistent vegetative state, locked-in syndrome, consciousness, functional disconnection,
diffusion tensor imaging, graph theory
INTRODUCTION
The brainstem plays a key role in cerebral cortex activation as the origin of the ascending reticular
activating system. Severe conditions can arise as a consequence of a traumatic or a non-traumatic
injury in this area. Characteristic clinical symptomatology includes well-differentiated disorders of
consciousness (DOC) such as “vegetative states” (VS) and “locked-in syndrome” (LIS). The VS
encompasses a spectrum of patients who have emerged from coma to exhibit basic orienting response
but show no indication of awareness [1,2]. By contrast, LIS is marked by tetraplegia and anarthria but
consciousness and somatosensory perception are preserved [3,4]. Because patients with LIS show
paralysis of all voluntary motor function except eye movements, they may be mistakenly thought to
be in VS. Behavioral deficits in these patients can be as severe as those produced by diffuse damage to
the cortical areas underlying high-level processing, indicative of awareness. The specific contribution
of individual regions to the clinical performance is determined by their topological integration into
brain networks [2,5]. This highlights the need to delineate the connectivity pattern in specific patients
in order to elucidate how each connection and sub-network is anatomically related to the injured area
and makes specific contribution to the level of consciousness.
Advancements in neuroimaging techniques, such as diffusion tensor imaging (DTI), offer the
potential to probe anatomical connectivity between remote neuronal populations. Recent studies have
suggested that structural networks of the human brain can be characterized by using graph theoretical
approaches [for review, see Bullmore and Sporns [6] and He and Evans [7]. Even though DTI has
provided unique insights on the underlying brain circuitry accounting for the presence of
consciousness and its alterations [5], the relationship of the network topology and efficiency metrics
with specific clinical states such as VS and LIS remains unclear.
On the other hand, recent studies in patients with DOC using functional MRI (fMRI) have shown
impaired connectivity in the thalamo-cortical and fronto-parietal networks, possibly reflecting
interruptions of higher-order processes [8,9]. However, these studies do not provide information about
neuroanatomic connectivity between different network nodes. Furthermore, the complexities of the
association between brain structure and function are not well understood.
To our knowledge, this work introduces for the first time, the use of graph theoretical approaches
and DTI to differentiate the topological organization of white matter network in patients in VS and
with LIS. In order to identify the characteristics properties of injured networks in comparison with
healthy architectures we used network-based statistics. In this study, we therefore sought to suggest
essential differences in reduced global network efficiency and altered nodal efficiency between these
DOC.
METHODS
The Ethics Committee of the Institute of Neurology and Neurosurgery, Havana approved this
study. Informed written consent was obtained from the patient’s legal representative and from all
healthy volunteers.

Disrupted Axonal Fiber Connectivity As a Marker of Impaired Consciousness States
321
Participants
Patient in VS: We studied a 24-year-old female patient who suffered a meningoencephalitis when
she was 3 years old, and developed a hydrocephalus requiring several ventricular derivations to
peritoneum, pleura and gallbladder. Four years ago, she suffered an acute hydrocephalus due an
obstruction of the ventricular derivation, causing a central and uncal herniation with a compression of
the brainstem. After being in coma during 4 weeks she was diagnosed as been in a persistent
vegetative state (PVS). Magnetic resonance imaging showed destruction of the rostral part of the
pons, the mesencephalon, and both thalami.
Patient with LIS
We studied a 47-year-old male patient who developed an acute stroke of the basilar artery
territory with sudden loss of consciousness. The patient remained in coma for 3 weeks, but afterwards
opened his eyes and showed sleep-wakefulness cycles. He was diagnosed as having been in a PVS
state. Nonetheless, when our group clinically examined him we noted that we were able to establish
communication with him using through coded messages by blinking or vertical eye movements,
indicating answers of YES or NO. We diagnosed him as being in a Locked-in syndrome (LIS). The
patient demonstrated quadriplegia and the inability to speak, but he was undoubtedly conscious and
aware, with no loss of cognitive function. He only preserved vertical eye movements and the blink
responses. MRI showed ischemic lesions of the pons, corresponding to the basilar artery territory.
Controls
Twenty healthy subjects (13 male, 11 female, age: 35.0 ± 12.1) were selected from the Cuban
Human Brain Mapping Project database [10]. Volunteers reported no history of psychiatric or
neurological disorders. Those studies were selected so as to coincide with the average age of the
patients and the age of these were within the standard deviation of the controls database.
Image Acquisition and Preprocessing
All images were acquired using a MRI scanner Siemens Symphony 1.5 T (Erlangen, Germany).
Using a standard diffusion gradient direction scheme (twelve diffusion-weighted images and a b=0
image), DW-MRI data were acquired using a single shot EPI sequence. To each subject, two
interleaved sets of 25 slices of 6 mm thickness with a distance factor of 100% were acquired with the
following parameters: b=1200 s/mm2; FOV=256×256 mm2; acquisition matrix=128×128;
corresponding to an ‘in plane’ spatial resolution of 2×2 mm2; TE/TR=160 ms/7000 ms. Two
interleaved sets were necessary because it was impossible to cover the whole head with a good spatial
resolution using a single set due to a pulse sequence limitation (max: 35 slices). Both sets were joined
to form a volume of 50 contiguous slices of 3 mm thickness covering the whole brain for each subject.
The aforementioned acquisition was repeated 5 times to improve signal to noise ratio (SNR). In order
to improve EPI quality, magnitude and phase difference images of a T2 gradient echo field mapping
sequence were acquired with TE=7.71 ms and 12.47 ms. Also, a 3D high-resolution T1-weighted
MPRAGE pulse sequence covering the whole brain was acquired with the following parameters: 160
contiguous slices of 1-mm thickness in sagittal orientation; in plane FOV=256×256 mm2, and matrix
size 256×256 yielding an spatial resolution of 1×1×1 mm3. The echo time, repetition time, and
inversion time were set to TE/TR/TI=3.93 ms/3000 ms/1100 ms with a flip angle FA=15°. Lesion
areas were manually segmented on the MPRAGE images using MRIcron software
(http://www.sph.sc.edu/comd/rorden/mricron/) to create binary masks in native space.

Rafael Rodriguez-Rojas, Karla Batista, Yasser Iturria et al.
322
Network Definition
The graph framework used here has been widely described in Iturria-Medina et al. [11]. In brief,
the cerebral volume is represented as a non-directed weighted graph G in which nodes N correspond
to anatomically defined regions and arcs  to the connections joining them. In DTI-based graphs,
between-regions connectivity can be reasonably restricted to white matter tracts connecting voxels of
the surfaces of the corresponding anatomical areas. This approach significantly reduces the
computational cost of the procedure.
While definition of the network nodes is consubstantial with graph theory, standard image-
matching algorithms to map atlas to brains are challenged in patients with brain injury. To overcome
this inconvenience, MPRAGE volumes were spatially normalized to the T1-MNI template using the
‘unified segmentation’ approach, available in SPM8 (http://fil.ion.ulc.ac.uk/spm) [12]. This procedure
combines segmentation, bias correction and spatial normalization in the non-linear deformation
model. Lesioned area was excluded from the calculation using a cost function masking to improve
normalization results. Lesion masks for each patient were created using rough drawings of lesion
boundaries. Atlas was registered with the gray matter volume maps in native space and segmented
into 71 regions, using the anatomically labeled template corresponding to the Jacob Atlas developed
by the Montreal Neurological Institute (http://www.mni.mcgill.ca/) and the IBASPM toolbox
(available at http://www.fil.ion.ucl.ac.uk/spm/ext/#IBASPM) [13]. Thus, the topological properties of
the brain anatomical networks were defined on the basis of the 71x71 binary graph G.
Fiber Tracking and Node-Node Connectivity
Figure 1 shows an overview of the methodology used in this study to identify impaired
connections and compromised sub-networks in patients. An iterative streamline fiber tractography
algorithm is employed for finding the most probable trajectory between each pair of gray matter
regions [11]. Streamlines exceeding 20 mm and below 500 mm in length and a curvature threshold of
±90o were used to generate the connectivity matrix for each subject.
The connectivity between a pair of nodes was quantified by two different anatomical connectivity
measures: Anatomical Connection Strength (ACS) and Anatomical Connection Density (ACD) [11].
ACS is related to the amount of nervous fibers shared by regions Ai and Aj by counting the nodes on
the surfaces involved in the connection:

 (,
)=∑() (1)
where () quantifies the conditional weight of the arc . On the other hand, ACD is a measure of
the fraction of the surface involved in the connection with respect to the total surface of both areas.
Thus, it can be estimated as the ratio between ACS and the number of nodes Ni and Nj belonging to
the surfaces of Ai and Aj, respectively:

 (,
)=
 (,)
|| (2)
Z transformation was used to provide probability statements about the deviation of the
connectivity measures from the normal values. Z scores of ACD and ACS were calculated by means
of Z = (x – µ)/

, where x represents the value of ACD or ACS in each patient, while µ and

are
respectively the mean value and the standard deviation of x in the population of normal subjects.

Disrupted Axonal Fiber Connectivity As a Marker of Impaired Consciousness States
323
Figure 1. I. Streamline fiber tractography; II. Nodes definition using non-linear transformations and IBASPM
toolbox; III. Anatomical network construction; IV. Calculation of network efficiency using graph theory; and V.
Identification of impaired connections using network based statistic.
Graph Analysis
To characterize the effect of the lesions over the efficiency in information exchange in brain
connectional architecture, we estimated the global and local efficiency of network as organizational
attributes of integration and segregation, respectively. For a graph G with N nodes, the global
efficiency can be computed as
 =
() ∑

()∈ (3)
where dij is the shortest geodesic lengths between pair of nodes i and j. On the other hand, the local
efficiency Eloc is measured as:
 =
∑()
∈ (4)
where Eglob(Gi) is the global efficiency of the local subgraphs, neighbors of node i.
In order to identify disconnected subnetworks in the patients, mass univariate testing was
performed at every connection using network based statistic (NBS) approach [14]. Primary t statistic
threshold for each link was set to 3.1 and 5000 permutations test were used to determine the statistical
significance of differences in network parameters.
RESULTS
Fiber tractography localized connectivity disruption at different levels in the brainstem in patients
in VS and with LIS, showing high consistency with lesion’s topology (Figure 2). In VS patients,
lesion in mesencephalon and rostral pons caused a disruption of cortico-spinal tract and cerebral
peduncle, causing a widespread alteration of thalamus-subcortical and cortico-cortical connectivity.

Rafael Rodriguez-Rojas, Karla Batista, Yasser Iturria et al.
324
On the other hand, brain injury in LIS is confined to the ventral pons disrupting the cortico-spinal and
spino-thalamic tracts, while cortical connectivity is preserved.
As shown in figure 3, for Cluster, E_glob and E_loc the values of ACD and ACS in VS subject
were lower than the corresponding mean values of control subjects, as expected. On the contrary, a
trend toward increased clustering and local efficiency accompanied by reduced global efficiency was
found for ACD in patients with LIS, but these differences were not significant.
Figure 2. From left to right: anatomical damage, streamline fiber tractography and spatial correlation of
impairments in patients. Upper: Patient in VS showing a widespread disruption in cortical and subcortical
connectivity. Lower: Patient with LIS showing an injury in pontine crossing tracts and preserved cortical
networking.
Figure 3. Differences in area under the global network properties curves: clustering index, global efficiency and
local efficiency. The clustering and efficiency index differences between normal controls and patients for
Anatomical Connection Density (ACD) and Anatomical Connection Strength (ACS) descriptors are depicted.
Bars represent the mean of each network property and black lines represent standard deviations in control
database.

Disrupted Axonal Fiber Connectivity As a Marker of Impaired Consciousness States
325
Table 1. Z scores of anatomical connectivity measures for both patients. Note that the patient
with LIS shows a (non-significant) tendency to higher local properties than healthy controls
ACD ACS
Cluster E_glob E_loc Cluster E_glob E_loc
% Z % Z % Z % Z % Z % Z
PVS -19.2 2.79* -55.2 8.84* -15.1 2.13* -55.5 1.44 -75.4 2.06* -53.8 1.20
LIS 3.5 -0.51 -5.3 0.85 4.4 -0.62 -3.7 0.09 -2.5 0.07 11.4 -0.26
*statistically significant (Z > 2)
The Z score test showed significantly lower (Z > 2) global and local efficiency, as well as
clustering index in ACD network in patient in VS, while ACS network showed significant reduction
only for global efficiency. On the other hand, anatomical connectivity measures showed no
differences in patients with LIS compared to controls (Table 1). Seeing together, these results suggest
that balance between local specialization, provided by Cluster and E_loc, and global integration
indicated by E-glob, has been spoiled in VS while remains essentially intact in LIS.
Figure 4. Nodes and streamline representation of impaired connections in VS (A), and LIS (B). Each node is
depicted as a black circle positioned at its node’s center of gravity. Gray circle is positioned in brainstem where
structural injury is located. Significance of differences was calculated using network-based statistics.
A network based statistical analysis was used to identify particular node pairs that were
abnormally connected in patients. NBS reveals that the architectural network presents anomalous
organization in both patients compared with the corresponding healthy values, but is suggestively
different between them. Streamlines interconnecting each of these node pairs are visualized in Figure
4. A widespread network was found to be significantly impaired in VS patient (p<0.05, corrected)
interconnecting several nodes comprising all cerebral lobes. Only two sub-networks were found to be
significantly impaired in the patient with LIS (p<0.05, corrected), corresponding to the basal ganglia-
thalamus-frontal cortex circuitry.

Rafael Rodriguez-Rojas, Karla Batista, Yasser Iturria et al.
326
DISCUSSION
An accurate and reliable assessment of the level and content of consciousness in DOC is critical
for the subsequent management and rehabilitation, as well as legal and ethical decision-making. To
date, the gold standard for diagnosis of the level of consciousness is behavioral. However, the scope
and reliability of available scales, examiner experience and overlapping of impairments in
consciousness limit clinical diagnosis. These concerns may lead to misdiagnosis in patients with LIS,
where cortical functions are preserved but the inability of production of voluntary motor behavior may
resemble patients in VS.
This study provides supports that DTI is a potent tool for dissecting the complex neuroanatomic
substrate of different DOC. Although the result of brain fiber tracking was not necessarily parallel to
the clinical symptoms, essential differences between patients in VS and LIS are evidenced in Figure 4.
Viewed as a network disorder, vegetative state has been found widespread impairment in cortico-
cortical connectivity in addition to the brainstem connections. This result suggests that the loss of
awareness in patients with lesions in the brainstem might be associated with aberrant neuronal
connectivity among widely distributed brain regions, and provide structural evidence for the notion of
VS as a disconnection syndrome. Significant reduction of all structural network attributes in this
pathological condition might be interpreted as a considerable decline in the amount of possible
nervous information that can be exchanged over the brain, and how deficiently it can be managed at
local and global levels.
On the contrary, patient with LIS showed a cortical connectivity pattern that was comparable with
that observed in healthy controls. This is in agreement with previous studies showing a near-to-normal
functional connectivity in LIS [15,16]. Disruption of connections between brainstem, basal ganglia
and frontal motor regions is consistent with quadriplegia, aphonia and quadriplegia and paralysis of
the cranial nerves, while cortical networks supporting consciousness remain intact. Those results are
highly consistent with remarkable functional and cognitive differences between these pathological
states, extensively reported in previous studies [1,3,16-20].
Curiously, ACD shows a higher sensitivity to differentiate network properties of pathological
states, especially VS, from healthy connectivity patterns. This is an unexpected result, considering that
differences in ACS exceed 50% for all the network parameters. More likely, this apparent
contradiction could be a result of the comparison of single subject against a normalized database. The
relatively large range of age in a small population of healthy subjects induces an overstated standard
deviation, which decreases the predictive ability of the Z score. A seminal study made by Iturria et al.
[11] found significant correlations among these connectivity matrixes. However this high correlation
was obtained between gray matter structures of healthy subjects, which may support our conclusion.
By definition, ACD is searched as a measure of the fraction of the surface involved in the connection
with respect to the total surface of both areas, while ACS give an estimate of the amount of nervous
fibers shared by these areas [11]. Thus, ACD normalize for the differences in surface area and volume
of brain regions (network nodes) related with normal inter-subject variability and/or aging [21].
Our MRI configuration allows for acquisition in 12 isotropically distributed diffusion-encoding
directions. This limitation determines the capacity of a tensor to model properly multiple fiber tracts
in one voxel. For that reason we applied the fiber assignment by continuous tracking (FACT)
streamline tracking algorithm [22]. This algorithm is computationally inexpensive and has been
demonstrated to be able to robustly reproduce brain circuitry. Figure 2 illustrate the performance of
the tractography method and the high topological association between damaged tissue and fiber
disruption in both patients. Application of advanced tractographic algorithms with high angular
resolution can avoid limitations inherent to streamline tracking. However, they might not be
computationally feasible as diagnostic tools in a clinical environment.

Disrupted Axonal Fiber Connectivity As a Marker of Impaired Consciousness States
327
The main limitation to our study is that our findings are limited to two patients. However, large
scale trials from patients with severe brain damage present challenging complexities related with
acquisition, analysis and interpretation of neuroimaging data and inhomogeneities in lesion topology,
which are unique for each patient. Our results suggest that DTI-based network connectivity combined
with graph theory is useful to study the long range effect of confined injuries and the relationship with
the degree of consciousness impairment, underlying PVS and LIS. The feasibility and reliability in the
application of these techniques to characterize single cases have encouraging implications for use
them as diagnostic and prognostic techniques in DOC.
Our results suggest that DTI-based network connectivity combined with graph theory is useful to
study the long range effect of confined injuries and the relationship with the degree of consciousness
impairment, underlying PVS and LIS. This is the first approach to characterize the dissimilarities in
branching configuration underlying the clinical and behavioral differences between vegetative states
and locked-in syndrome. More detailed comparisons between normal and pathological ACS and ACD
maps combined with graphic theoretical tools could become a potential procedure to diagnose and
distinguish different disorders of consciousness related to white matter injury in the pathways from
brainstem to cerebral cortex.
ACKNOWLEDGMENTS
RRR was partially supported by the “Abdus Salam” International Center for Theoretical Physics,
Trieste, Italy. Authors would like to thank Andy Stanert for the revision of the manuscript.
REFERENCES
[1] Monti MM, Laureys S, Owen AM. The vegetative state. BMJ. 2010;341:c3765.
[2] Querol-Pascual MR. Clinical Approach to Brainstem Lesions. Semin Ultrasound CT MRI.
2010;31:220-9.
[3] Cardwell MS. Locked-in syndrome. Tex Med. 2013;109(2):e1.
[4] Patterson J, Grabois M. Locked-in syndrome: a review of 139 cases. Stroke. 1986;17(4):758-64.
[5] Edlow BL, Takahashi E, Wu O, Benner T, Dai G, Bu L, et al. Neuroanatomic connectivity of
the human ascending arousal system critical to consciousness and its disorders. J. Neuropathol
Exp Neurol. 2012 Jun;71(6):531-546.
[6] Bullmore E, Sporns O. Complex brain networks: graph theoretical analysis of structural and
functional systems. Nat. Rev. Neurosci. 2009 Mar;10(3):186-198.
[7] He Y, Evans A. Graph theoretical modeling of brain connectivity. Curr. Opin Neurol. 2010
Aug;23(4):341-350.
[8] Crone JS, Ladurner G, Holler Y, Golaszewski S, Trinka E, Kronbichler M. Deactivation of the
default mode network as a marker of impaired consciousness: an fMRI study. PLoS One.
2011;6(10):e26373.
[9] Noirhomme Q, Soddu A, Lehembre R, Vanhaudenhuyse A, Boveroux P, Boly M, et al. Brain
connectivity in pathological and pharmacological coma. Front Syst Neurosci. 2010;4:160.
[10] Hernandez-Gonzalez G, Bringas-Vega ML, Galan-Garcia L, Bosch-Bayard J, Lorenzo-Ceballos
Y, Melie-Garcia L, et al. Multimodal quantitative neuroimaging databases and methods: the
Cuban Human Brain Mapping Project. Clin. EEG Neurosci. 2011 Jul;42(3):149-159.
[11] Iturria-Medina Y, Canales-Rodriguez EJ, Melie-Garcia L, Valdes HP, Martinez E, Aleman-
Gomez Y, et al. Characterizing brain anatomical connections using diffusion weighted MRI and
graph theory. Neuroimage. 2007;36:645-660.

Rafael Rodriguez-Rojas, Karla Batista, Yasser Iturria et al.
328
[12] Ashburner J, Friston KJ. Unified segmentation. Neuroimage. 2005;26(3):839-851.
[13] Alemán-Gómez Y, Melie-García L, Valdes-Hernández P. IBASPM: toolbox for automatic
parcellation of brain structures. Proceeding of 12th Annual Meeting of the Organization for
Human Brain Mapping. Available on CD-Rom in NeuroImage. 2006;27(1).
[14] Zalesky A, Fornito A, Bullmore E. Network-based statistic: Identifying differences in brain
networks. Neuroimage. 2010;53:1197-1207.
[15] Soddu A, Vanhaudenhuyse A, Bahri M, Bruno M, Boly M, Demertzi A, et al. Identifying the
Default-Mode Component in Spatial IC Analyses of Patients with Disorders of Consciousness.
Hum. Brain Mapp. 2012;33:778-796.
[16] Vanhaudenhuyse A, Noirhomme Q, Tshibanda LJ, Bruno MA, Boveroux P, Schnakers C, et al.
Default network connectivity reflects the level of consciousness in non-communicative brain-
damaged patients. Brain. 2010 Jan;133(Pt 1):161-171.
[17] Guldenmund P, Vanhaudenhuyse A, Boly M, Laureys S, Soddu A. A default mode of brain
function in altered states of consciousness. Arch Ital Biol. 2012 Jun;150(2-3):107-121.
[18] de Jong BM. "Complete motor locked-in" and consequences for the concept of minimally
conscious state. J. Head Trauma Rehabil. 2013 Mar;28(2):141-143.
[19] Bardin JC, Schiff ND, Voss HU. Pattern classification of volitional functional magnetic
resonance imaging responses in patients with severe brain injury. Arch Neurol. 2012
Feb;69(2):176-181.
[20] Bruno MA, Vanhaudenhuyse A, Thibaut A, Moonen G, Laureys S. From unresponsive
wakefulness to minimally conscious PLUS and functional locked-in syndromes: recent
advances in our understanding of disorders of consciousness. J. Neurol. 2011 Jul;258(7):1373-
1384.
[21] Hedman AM, van Haren N, Schnack H, Kahn R, Hulshoff Pol HE. Human brain changes across
the life span: a review of 56 longitudinal magnetic resonance imaging studies. Hum. Brain
Mapp. 2012;33(8):1987-2002.
[22] Mori S, Crain BJ, Chacko V, Van Zijl PC. Three-dimensional tracking of axonal projections in
the brain by magnetic resonance imaging. Ann. Neurol. 1999;45:265-269.
Received: June 13 2013 Revised: June 18 2013 Accepted: June 23 2013.