Review
- Fatima T. Husaina, b, c,
,
,
- Sara A. Schmidtb, c
- a Department of Speech and Hearing Science, University of Illinois at Urbana-Champaign, Champaign, IL, USA
- b Neuroscience Program, University of Illinois at Urbana-Champaign, Champaign, IL, USA
- c Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana-Champaign, Champaign, IL, USA
Open Access
Highlights
- •
- We review recent studies of intrinsic networks that may subserve tinnitus.
- •
- These studies estimate resting state functional connectivity of fMRI data.
- •
- Converging evidence suggests alterations between diverse regions and limbic network.
- •
- Changes also occur in connections between attentional nodes and other brain regions.
Abstract
Resting
state functional connectivity (rs-fc) using fMRI has become an
important tool in examining differences in brain activity between
patient and healthy populations. Studies employing rs-fc have
successfully identified altered intrinsic neural networks in many
neurological and psychiatric disorders, including Alzheimer's disease,
schizophrenia, and more recently, tinnitus. The neural mechanisms of
subjective tinnitus, defined as the perception of sound without an
external source, are not well understood. Several inherent networks have
been implicated in tinnitus; these include default mode, auditory,
dorsal attention, and visual resting-state networks. Evidence from
several studies has begun to suggest that tinnitus causes consistent
modifications to these networks, including greater connectivity between
limbic areas and cortical networks not traditionally involved with
emotion processing, and increased connectivity between attention and
auditory processing brain regions. Such consistent changes to these
networks may allow for the identification of objective brain imaging
measures of tinnitus, leading to a better understanding of the neural
basis of the disorder. Further, examination of rs-fc allows us to
correlate behavioral measures, such as tinnitus severity and comorbid
factors including hearing loss, with specific intrinsic networks.
This article is part of a Special Issue entitled <Human Auditory Neuroimaging>.
Abbreviations
- rs-fc, Resting state functional connectivity;
- EEG, electroencephalography;
- MEG, magnetoencephalography;
- fMRI, functional magnetic resonance imaging;
- PET, positron emission tomography;
- ICA, independent component analysis;
- RSN, resting state network;
- DMN, default mode network;
- DAN, dorsal attention network;
- BOLD, blood oxygen level-dependent;
- DTI, diffuser tensor imaging
1. Introduction
Resting-state
functional connectivity (rs-fc) is a term used to describe
interregional correlation of brain activity measured using imaging
techniques. It has gained prominence in recent years not only for its
usefulness in highlighting several functional neural networks of the
brain, but also for identifying neuroimaging biomarkers of a condition
or a disorder (Horwitz and Rowe, 2011).
In this review, we focus on studies of rs-fc using functional magnetic
resonance imaging (fMRI) that have underscored the neural networks
subserving tinnitus and accompanying hearing loss and the use of such
studies in characterizing the pathophysiological markers of the
disorder. We also discuss a potential use of rs-fc as a means of
identifying subtypes based on pathology rather than on symptoms and its
use in assessing treatment efficacy. In this domain of identifying
objective biomarkers of a disorder, much can be learned from studies of
normal aging or neuropsychiatric disorders such as schizophrenia, which
have a longer history of using rs-fc. We highlight challenges of using
rs-fc in general and those that are unique to the study of tinnitus and
end the review with suggested directions for future rs-fc studies of
tinnitus.
1.1. Resting-state networks
Resting
state connectivity is, by definition, spontaneous fluctuations in brain
activity that can be reliably organized into coherent networks. The
term ‘resting state’ differentiates this type of activity from that
obtained as a result of some task or stimulus. Since at least the 1980s,
different brain imaging tools have noted such inherent networks,
including EEG, or electroencephalography (e.g., Giaquinto and Nolfe, 1988), MEG, or magnetoencephalography (e.g., Lu et al., 1992 and Salmelin and Hari, 1994), positron emission tomography (PET) (e.g., Horwitz et al., 1987) and fMRI (e.g., Biswal et al., 1995 and Lowe et al., 1998).
This review is concerned primarily with fMRI studies of resting state
networks (RSNs). The first fMRI study to examine RSNs discovered strong
correlations between motor regions when subjects were not performing a
motor task (Biswal et al., 1995).
Interestingly, the characteristics of this connectivity were similar to
how the network appears during a task. Other systems, including those
for auditory processing (Cordes et al., 2000), visual processing (Lowe et al., 1998), or even higher-order functions such as language processing (Hampson et al., 2002),
were also shown to have resting state counterparts. Exploration into
the potential use of RSNs as tools to better understand the connectivity
in the brain therefore began to grow in popularity. For a more detailed
description of the history of studying RSNs using fMRI, see Hampson et al. (2012) and Fox and Raichle (2007).
RSNs
are typically delineated via functional connectivity analyses. Here, we
briefly describe three popular methods of analysis: seeding, graph
connectivity analysis, and independent component analysis, or ICA. In a
seeding analysis, a seed region is selected based on the question being
asked by the researcher. The connectivity of the seed region can then be
examined by finding correlations between the time course of voxels (a
voxel is 3-D cubic element in a brain image, similar to a pixel in a 2-D
image) in the seed and the rest of the voxels in the brain.
Alternatively, the time course of the seed region could be correlated
with those of voxels in specific regions of interest rather than with
the whole brain. These correlations are then used to generate
connectivity maps that can be compared across groups via standard
statistical tests such as t-tests or tests of analysis of
variance. Seeding analysis benefits from the straight-forward nature of
interpretation and of the analysis itself. Results using this method
are, however, highly dependent upon the seed regions chosen, thereby
making it vulnerable to bias. Graph connectivity analysis is similarly
influenced by selecting regions of interest. Here, correlations between a
set of select nodes are calculated. These correlations are represented
by edges between the nodes, the strength of which is incorporated in the
resulting graph. Thus, group differences can be found by comparing how
nodes are connected via edges and the strength of those connections. ICA
differs from the other two approaches in that it is primarily data
driven and allows for the analysis of multiple whole-brain networks.
There is no need for a priori hypotheses; instead, ICA uses the
time courses of voxels in the fMRI scans to produce a specified number
of components, which are optimally spatially independent (although this
optimal independence does not necessarily imply that there is no overlap
between components). Deciding on the number of components used is an
important part of the ICA technique and can strongly influence results.
The components produced by ICA should separate resting state networks
from each other and noise by placing them in separate components. Unlike
a seeding approach, the resulting data from group ICA may be more
difficult to interpret, but its data-driven nature makes it particularly
appropriate for exploratory analyses with no a priori hypotheses. Hampson et al. (2012) and Cole et al. (2010) both provide more detailed descriptions of these methods and the benefits and drawbacks of each.
Though
spontaneous activity can engage any brain region, the default mode
network, or DMN, has gained prominence as the canonical RSN. In this
formulation, the DMN typically comprises of nodes in the posterior
cingulate/precuneus, bilateral superior frontal gyrus, medial frontal
gyrus and angular gyrus (Mantini et al., 2007).
The DMN is the most active at rest and shows reduced activity when a
subject enters a task-based state involving attention or goal-directed
behavior (Shulman et al., 1997);
an opposite pattern is seen with other RSNs, which exhibit heightened,
correlated activity in the task-based state but retain connectivity
(although with reduced activity) during rest. The DMN exhibits a uniform
oxygen extraction fraction when examined using PET, indicating
equilibrium between the energy requirements of the neurons and the blood
supply to the brain (Raichle et al., 2001).
When the brain is involved in a task, neurons require an increased
amount of blood, and the oxygen extraction fraction reflects this.
Because the fraction is uniform in the DMN, the fluctuations in activity
seen are not related to a task and the brain does not need additional
physiological resources to maintain them. The DMN was therefore termed a
“baseline” state of the brain and may be involved in ongoing activity
over longer periods of time (Raichle et al., 2001). See Raichle and Snyder (2007)
for an overview of the DMN. It is also worth noting that rs-fc,
including connectivity of the DMN, may be at least in part independent
of ongoing cognition. The presence of the DMN has been noted in the
brains of anesthetized monkeys (Vincent et al., 2007), as well as in humans, where its coherence varies with the degree of consciousness (Guldenmund et al., 2012). It would be remiss of us not to note that the value of using DMN to study brain function is not without controversy (Morcom and Fletcher, 2007), but a discussion of its merits is outside the scope of this review.
Apart
from the DMN, several other RSNs are applicable in studying the neural
mechanisms of tinnitus or auditory processing in general (Fox et al., 2005, Langers and Melcher, 2011 and Mantini et al., 2007).
Studies of task-based and resting functional connectivity in normal
hearing healthy adults have shown that a diverse set of networks,
including the canonical RSNs defined previously, participate in auditory
processing (Langers and Melcher, 2011).
For the remainder of the review, we focus primarily on the DMN, the
attention networks; the visual RSN, the auditory RSN, and nodes of the
limbic network (see Fig. 1
for a representative figure of these networks). The visual RSN includes
the occipital cortex and temporal-occipital regions, whereas the
superior temporal cortex alone defines the auditory RSN (Mantini et al., 2007).
The dorsal attention network, or DAN, is comprised of the bilateral
intraparietal sulci, the ventral precentral gyrus, the middle frontal
gyrus, and the frontal eye fields (Mantini et al., 2007).
Other networks of attention, such as the ventral attention network
(including temporoparietal junction and superior temporal sulcus) and
that of the executive control of attention (including middle, inferior
and medial frontal gyri and anterior insula), may also be applicable to
tinnitus (Burton et al., 2012).
- Fig. 1.
Summary of main results of resting-state functional connectivity studies in tinnitus. The major networks highlighted are default-mode network (DMN, shown in blue), limbic network (green), auditory network (red), the visual network (in orange), several attention networks (specifically the dorsal attention network and the executive control of attention, shown in purple), and the visual network (in orange). Positive correlations between regions that are stronger in tinnitus patients than controls are shown in solid lines, while negative correlations are dashed lines. This figure shows modifications to the networks and does not represent the networks in their entirety. Connections are labeled with letters representing the studies in which they were reported, as follows: a) Schmidt et al., in press. b) Burton et al., 2012. c) Maudoux et al., 2012b. d) Kim et al., 2012. Abbreviations: PCC: posterior cingulate cortex; mpfc: medial prefrontal cortex; lifg: left inferior frontal gyrus; parahipp: parahippocampus; aud cortex: auditory cortex; fef: frontal eye fields.
1.2. Tinnitus
Subjective
tinnitus is the phantom perception of sound in the absence of an
external source. Tinnitus is a fairly common hearing disorder, with a
prevalence rate of 10–20% in the general population (Davis and Rafaie, 2000).
The great majority of individuals with tinnitus are well-adjusted to
it. However, 10–20% of those with tinnitus may seek medical care to
alleviate symptoms associated with tinnitus and in 2–5% of the tinnitus
population, the symptoms are severe and affect activities of daily
living (Davis and Rafaie, 2000).
About 90% of those with tinnitus have some degree of
clinically-diagnosed hearing loss, but the opposite is not true; only
about 40% of those with hearing loss may have tinnitus (Lockwood et al., 2002 and Vernon, 1997).
Therefore, hearing loss remains a major trigger and contributor to the
neural changes concomitant with tinnitus. Tinnitus has also been
correlated with depression and anxiety, with increased rates of
co-occurrence of these conditions with tinnitus (Bartels et al., 2008). It is not surprising then that conceptual models of tinnitus have incorporated auditory processing (Bauer, 2004 and Kaltenbach et al., 2005) and emotional processing networks (Jastreboff, 1990 and Rauschecker et al., 2010)
in their explanation of the neural mechanisms of tinnitus. Other
reviews have pointed to contributions from the somatosensory system to
tinnitus (Levine, 1999 and Shore, 2011), and recent brain imaging studies have implicated the attention network as well (Gu et al., 2010, Husain et al., 2011b and Roberts et al., 2010). For overarching reviews of tinnitus mechanisms, see (Bauer, 2004, Eggermont and Roberts, 2004 and Roberts et al., 2010).
Because of its subjective nature,1
tinnitus may be uniquely suited to being studied using a resting-state
functional connectivity paradigm; there is no task-based modulation of
the tinnitus signal. Tinnitus is the perception of a phantom sound in
the absence of an external source. At the same time, perception of a
chronic internal noise may place the person in a task-based state and no
true resting-state may be achieved by individuals with tinnitus. A
better term than resting-state to denote this type of response would be
steady-state or inherent functional networks. For the sake of
maintaining compatibility with the broader resting-state literature and
with the published studies on tinnitus, we will use the term
resting-state functional connectivity in our review, but with the caveat
that no true resting-state may be achieved by those with chronic
tinnitus. In any case, the contrast in the spontaneous activity between
individuals with tinnitus and those without should provide insights into
neural bases of tinnitus.
2. Tinnitus and resting state functional connectivity
The
effects of tinnitus on resting state functional connectivity have
recently been explored using fMRI, although the results are variable,
partly due to differences in experimental and analytical methods and
partly due to the heterogeneity of the patient population. Nevertheless,
two main themes have emerged in data from tinnitus patients relative to
controls: an increased correlation between limbic areas and other brain
regions, as well as correlation differences between
attention-processing regions and other parts of the brain. A summary
figure of the main findings, which place a particular emphasis on the
examination of the auditory RSN, the DMN, and attention networks, is
shown in Fig. 1. Major findings along with experimental and analytical details are also reported in Table 1.
- Table 1. Summary table of the details of the various resting-state studies of tinnitus.
Number of subjects Age of subjects THI scores Hearing loss of TIN patients Method Networks examined Major findings (in TIN, relative to controls) K 6 NHC (2f), 4 TIN (1f) 45 ± 2.76 NHC, 45 ± 3.92 TIN Not given None to severe Group ICA, seed-to-voxel Aud r AC ↔ ↑ l AC; AC ↔ ↑amyg, dmpfc Ma 15 NHC (6f), 13 TIN (6f) 51 ± 13 NHC, 52 ± 11 TIN 16–84, mean 43.5 Mild to severe Connectivity graph Aud TIN & NHC different graphs; AC ↔ ↑ l phipp Mb 15 NHC (6f), 13 TIN (6f) 51 ± 13 NHC, 52 ± 11 TIN 16–84, mean 43.5 Mild to severe Between group ICA Aud AC ↔ ↓ l pfc, l fus, occip; AC ↔ ↑ stem, bg, cereb, phipp, r pfc, pari, sm B 17 NHC (10f), 17 TIN (6f) 50.6 ± 2.5 NHC, 53.5 ± 3.6 TIN 38–76, mean 53.5 None to severe Seed-to-seed, seed-to-voxel Aud, vis, Som, DAN, VAN, ECA AC ↔ ↓ VC; VC↔ ↓ tpj, ifg, ins; Occip ↔ ↓ ins, ifg W 23 NHC (11f), 18 TIN (6f) Median 46 (IQR 39–54) NHC, median 54 (IQR range 52–57) TIN 0–24, mean 9.67 None to severe Seed-to-seed, seed-to-voxel DAN, VAN, Cog, Aud, Vis, Som, DMN No differences S 15 NHC (6f), 13 HLC (8f), 12 TIN (3f) 52.93 ± 8.64 NHC, 57.62 ± 9.39 HLC, 55.00 ± 6.97 TIN 0–22, mean 8.33 Mild to moderate in TIN, HLC (matched) Seed-to-voxel DMN, DAN, Aud AC & fef ↔ ↑ phipp; rs-fc ↓ DMN; ips ↔ ↓ r smg - All
of the major findings were found in tinnitus patients relative to
controls. ↔ shows resting state functional connectivity (rs-fc) between
regions, with↑ indicating increased connectivity and ↓ decreased
connectivity.
Abbreviations: K: (Kim et al., 2012); Ma: (Maudoux et al., 2012a); Mb: (Maudoux et al., 2012b); B: (Burton et al., 2012); W: (Wineland et al., 2012); S: Schmidt et al., in press, NHC: normal hearing controls; HLC: hearing loss controls; TIN: tinnitus patients; THI: tinnitus handicap inventory; HL: hearing loss; Aud: auditory resting state network; Vis: visual resting state network; Som: somatosensory network; DAN: dorsal attention network; VAN: ventral attention network; ECA: executive control of attention network; Cog: cognitive network; DMN: default mode network; AC: primary auditory cortex; r: right; l: left; amyg: amygdala; dmpfc: dorsomedial prefrontal cortex; phipp: parahippocampus; pfc: prefrontal cortex; fus: fusiform gyrus; stem: brainstem; bg: basal ganglia; cereb: cerebellum; pari: partietal lobule; sm: sensorimotor; VC: visual cortex; tpj: temporoparietal junction; ifg: inferior frontal gyrus; ins: insula; Occip: occipital cortex; fef: frontal eye fields; smg: supramarginal gyrus; ips: intraparietal sulci.
2.1. Limbic system
A preliminary study (Kim et al., 2012),
which examined rs-fc using fMRI in tinnitus subjects revealed findings
concordant with both themes identified earlier. Increased connectivity
was estimated between the auditory cortices and the amygdala in the
tinnitus group when compared to age-matched normal hearing controls (see
Fig. 1).
This association between auditory and limbic regions in tinnitus has
been suggested by numerous other brain imaging studies and conceptual
models of tinnitus. The neurophysiological model of tinnitus proposed by
Jastreboff (1990)
describes the interaction between the limbic and auditory systems. The
model emphasizes the importance of habituation to the tinnitus percept,
which allows a patient to ignore the phantom sound. However, when
“negative reinforcement” is present, the limbic system can cause the
auditory activity to be perceived, which could then lead to a feedback
loop. The correlation between the auditory RSN and limbic areas fits the
framework of this hypothesis. A more recent update of the
limbic-auditory interactions was proposed by Rauschecker et al. (2010) and is based on structural MRI data (Leaver et al., 2012 and Muhlau et al., 2006). Task-based fMRI studies also provide evidence for the auditory-limbic link seen in the Kim et al. study. Golm et al. (2013)
examined the relationship between tinnitus and emotional processing in
an emotional sentences task, revealing changes in activation in limbic
and frontal areas in highly distressed tinnitus patients. They suggest
that the significant regions, including the anterior cingulate cortex,
the medical cingulate cortex, the insula and the precuneus, are part of a
general distress network and are not specific to tinnitus. The exact
mechanism of and the networks involved in tinnitus distress are still
being evaluated, and rs-fc studies are providing further information to
suggest the importance of limbic areas in this process. We have also
conducted a task-based fMRI study examining the effects of tinnitus and
hearing loss on emotional processing influenced by this hypothesis (Carpenter-Thompson et al., unpublished).
A separate rs-fc study (Maudoux et al., 2012b)
also found results that support the limbic-auditory link in tinnitus
patients. Using a combination of independent component and graph
connectivity analyses described in Soddu et al. (2011),
connectivity graphs of the auditory component network were built for
both tinnitus and control groups. The auditory network of the control
and tinnitus groups included bilateral primary and associative auditory
cortices, insula, prefrontal, sensorimotor, anterior cingulate and left
occipital cortices. In addition to these regions, the tinnitus group's
network comprised the brainstem, thalamus, nucleus accumbens, isthmus of
cingulate gyrus, and occipital, parietal and prefrontal cortices.
Increases and decreases in connectivity were seen in the tinnitus group
as compared to controls; specifically, the tinnitus group showed
increased connectivity in the brainstem, cerebellum, right basal
ganglia/nucleus accumbens, parahippocampal areas, right frontal and
parietal areas, left sensorimotor areas and left superior temporal
region and decreases in the right primary auditory cortex, left fusiform
gyrus, left frontal and bilateral occipital regions. In a companion
study using connectivity graphs (Maudoux et al., 2012a),
two connectivity patterns were found in the auditory RSN. The first
involved the bilateral auditory cortices and insula. This network was
positively correlated with the time course of the auditory RSN, and was
found in both tinnitus and control groups. A second network that was
anti-correlated with the auditory RSN time course was found only in
control subjects. This network included the frontoparietal lobe, the
anterior cingulate cortex, the amygdala, the brainstem, and the
parahippocampus. Increased functional connectivity that was found
between the auditory cortices and the left parahippocampus in tinnitus
can therefore be explained by the loss of coherence in this
anti-correlated network (Maudoux et al., 2012a). It also explains the inclusion of other brain regions in the auditory connectivity graphs created in Maudoux et al. (2012b). Of particular note in these study is the increase in connectivity in the parahippocampal areas (shown in Fig. 1),
which again demonstrates a relationship between tinnitus and limbic
areas in resting state analyses. There was also a trend for increased
correlation between the auditory cortices and the amygdala in tinnitus
patients, but it did not survive correction.
In our own study (Schmidt et al., in press)
we detected increased correlations in the activity of the limbic system
and inherent networks in tinnitus patients compared to controls. The
study employed continuous acquisition of data for 5 min while the
participants were at rest. Three groups of subjects were scanned – 12
middle-aged adults with hearing loss and tinnitus, 13 age-matched
controls with hearing loss without tinnitus, and 15 normal hearing
controls without tinnitus. We conducted a seed-to-voxel analysis to
examine the auditory RSN, DMN and DAN in the three groups. Comparable
with previous studies (Kim et al., 2012, Maudoux et al., 2012a and Maudoux et al., 2012b),
an increased correlation with the limbic network was found in tinnitus
patients in the auditory network. This correlation, found in the left
parahippocampus, was significant when the tinnitus group was compared to
normal hearing controls, but did not reach significance when the
patients were contrasted with hearing loss controls (though there was a
clear trend). We also found increased connectivity between the right
parahippocampus and the DAN, with seed regions located in the bilateral
frontal eye fields.
2.2. Attention system
Kim et al. (2012)
also found results that suggest alterations in resting state activity
in brain regions associated with attention. Specifically, increased
connectivity was found between the dorsal medial prefrontal cortex and
the auditory RSN. The authors suggest that this aberrant functional
connection may result in the tinnitus percept (Kim et al., 2012).
Hypotheses that tinnitus can cause changes in the organization of
sensory networks and interfere with networks of attention led (Burton et al., 2012)
to examine the visual, auditory, somatosensory, DAN, ventral attention
network and attention control resting state networks in 17 patients with
bothersome tinnitus (with scores ranging from 38 to 76 on the Tinnitus
Handicap Inventory (THI) (Newman et al., 1996)).
To do so, spherical seed regions were selected within each of these
networks (17 seeds in total). Temporal correlations were calculated
between pairs of regions, and connectivity maps were calculated for
those that had group differences with probabilities less than 0.05.
T-statistics were calculated to detect significant differences between
tinnitus and control groups (Burton et al., 2012).
Almost all seed pairings within the DAN were not found to be
significant between groups. Correlations between seeds in the auditory
and visual RSNs were found to be positive in controls but negative in
the tinnitus group, perhaps because the additional stimulation caused by
the tinnitus percept decreases activity in the visual cortex that is
irrelevant to processing the phantom sound (Burton et al., 2012). In the tinnitus group, functional connectivity in areas of attention control was greater than that in the control group (see Fig. 1).
This connectivity was positively correlated with activity in the
auditory cortex and negatively correlated with the occipital cortex.
Increased associations with limbic areas were also found in tinnitus
subjects when compared to controls, specifically between the primary
auditory cortex and the insula. However, this connection was not strong
enough to survive correction (Burton et al., 2012).
Alterations to attention networks were also seen in our work (Schmidt et al., in press).
The DAN, with seed regions in the bilateral intraparietal sulci, showed
decreased correlations with the right supramarginal gyrus in tinnitus
subjects compared to hearing loss controls. This is contrary to the lack
of significant results seen in the DAN seeds in Burton et al. (2012).
Such differences may be accounted for by differences in analysis
methods and heterogeneity of the participant groups, as discussed next.
In addition, in the DMN, our study revealed decreased correlations
between seed regions (located in the posterior cingulate cortex and
medial prefrontal cortex) and the precuneus in tinnitus patients when
compared to both normal hearing and hearing loss controls. The precuneus
is one of the main hubs of the DMN, so this decreased connectivity
indicates the network is disrupted and patients are not in a true
resting state. Tinnitus patients may therefore be attending to or
attempting to suppress the phantom sound.
2.3. Accounting for the variability
Although
the rs-fc studies to date share results with similar themes, the exact
brain regions involved in the RSNs and the strength of the connections
between them have been variable. This could be due to several reasons.
First, the methods of analyses varied across the studies. Kim et al. (2012) used a group ICA followed by a seed analysis, whereas Burton et al. (2012) used a seed-to-seed analysis followed by a seed-to-voxel analysis. Maudoux et al. used connectivity graphs (Maudoux et al., 2012a) and between group ICA (Maudoux et al., 2012b). Each of these approaches is driven by different a priori hypothesis. In the case of the Kim et al. (2012)
paper, the initial independent component analysis did not require any
prior hypothesis, but the results are very sensitive to the number of
subjects in each group and the number of components chosen during the
analysis. The analysis itself can vary between replications, because
there is no specific “optimal” solution to the computations. The Burton et al. (2012) seed analysis, in contrast, is highly dependent on the precise seed regions chosen. Our study (Schmidt et al., in press) is similarly influenced by seed selection, though we did not conduct a seed-to-seed analysis as Burton et al. (2012) performed as part of their analysis. It is not surprising that these different methods lead to different results.
Second,
the number of subjects examined in the different studies was variable.
Specifically, the Kim et al. (2011) study used a very small cohort (four
tinnitus subjects and 6 controls) in their pilot study. This is quite
different from the 13 patients and 15 controls used by Maudoux et al., 2012a and Maudoux et al., 2012b, the 17 patients and 17 controls used by Burton et al. (2012), and the 15 normal hearing controls, 13 hearing loss controls, and 12 patients used in our study (Schmidt et al., in press). Especially in the case of group ICA, subject number has a large impact on the results of a study.
A
third issue is variation in characteristics of the patient population.
For example, the extent of hearing loss in the tinnitus patients studied
is highly variable and varies greatly across subjects. In our study (Schmidt et al., in press)
we included a hearing loss control group to account for the effect of
this potential confound. The severity of the tinnitus experienced by the
patients in each study may also have a strong impact on the results. In
the Maudoux study (Maudoux et al., 2012a), severity was highly variable across patients, ranging from a THI score of 84 to a low of 16. In Burton et al. (2012),
all of the patients experienced bothersome tinnitus, but the THI scores
again varied quite a lot, from 38 to 76. All of the patients in our
study had nonbothersome tinnitus with THI scores ranging from 0 to 18 (Schmidt et al., in press). This variation plays a key role in the affect tinnitus has on resting state connectivity and is demonstrated by Wineland et al. (2012).
In the Wineland et al. study, an almost identical analysis to that used
in Burton et al. was performed on a group of patients with
nonbothersome tinnitus. In contrast to Burton et al. (2012),
no significant results were found. This finding strongly emphasizes
that alterations in the resting state are related to tinnitus severity,
and variations therein could be confounds in past research. A direct
comparison between bothersome and nonbothersome tinnitus groups would be
highly beneficial to confirm this result. Additionally, Maudoux et al. (2012a)
found that THI scores are significantly positively correlated with
regression measures of correlation in the posterior cingulate cortex.
Tinnitus questionnaire scores are also positively correlated with the
posterior cingulate response and also those of the left parietal region;
however, these correlations did not reach significance. These results
further emphasize the influence of tinnitus severity on results.
RSNs have been shown to alter with age and the great majority of individuals with tinnitus are middle-aged or older (Henry et al., 2005).
In the DMN, decreased connectivity in the posterior cingulate cortex,
frontal gyrus and parietal regions with age has been noted. In
task-based examinations, deactivations typically found in the DMN were
shown to be weaker in older adults, which indicates that this population
has more difficulty moving into a task-based scenario from rest (Hafkemeijer et al., 2012).
Though most studies on aging and rs-fc have thus far focused on the
DMN, other networks including those associated with attention have also
been examined. For instance, (Ferreira and Busatto, 2013) noted heightened functional interactions between frontal and parietal cortices (Ferreira and Busatto, 2013).
When using rs-fc to study tinnitus and hearing loss, both of which are
associated with an older population, it is important to keep in mind the
network alterations that come from aging alone. Subject groups should
be carefully age-matched in order to account for this confounding
variable.
3. EEG and MEG studies of resting state functional coupling
The first insights into inherent long-range cortical coupling in tinnitus were provided not by fMRI but by MEG (Lorenz et al., 2009, Schlee et al., 2009, Weisz et al., 2007a and Weisz et al., 2007b) and EEG (Vanneste et al., 2010a and Vanneste et al., 2010b) resting-state studies. Weisz and colleagues (Lorenz et al., 2009, Schlee et al., 2009, Weisz et al., 2007a and Weisz et al., 2007b)
in a series of studies have brought forth evidence that implicate alpha
(8–12 Hz), delta (<4 Hz) and gamma (30–60 Hz) wave oscillations
identified using MEG. We refer the reader to the article in this special
issue by Weisz et al. for a review of some of these studies and their
findings. EEG likewise has been used to determine long-range functional
coupling in the resting state, most prominently by De Ridder and
colleagues (Vanneste et al., 2011, Vanneste et al., 2010a and Vanneste et al., 2010b).
EEG and MEG do not offer the spatial resolution of fMRI, but they offer
the advantage of being quiet and not interfering or masking the
individual's hearing loss or tinnitus. The other advantage of these
techniques is their temporal resolution of the order of a few
milliseconds compared to the 1–3 s temporal resolution of most fMRI
studies.
The temporal and to
some extent spatial resolution differences of fMRI and EEG/MEG may mean
that these tools are measuring different aspects of spontaneous brain
activity (Tagliazucchi et al., 2012). For comparative studies of resting state cortical activity as measured by fMRI and EEG see (Britz et al., 2010, Laufs, 2010, Mantini et al., 2007, Musso et al., 2010 and Tagliazucchi et al., 2012). The studies validate to some extent the correlation between EEG microstates
occurring over a timescale of milliseconds with fMRI-BOLD (blood oxygen
level-dependent) oscillation patterns occurring over a timescale of
seconds for several RSNs. Britz et al. (2010)
computed 4 RSNs from the EEG data that were the equivalent of
stereotypical BOLD RSNs dedicated to auditory/phonological, visual,
attention and self-referential processing. However, no EEG equivalent of
the DMN was detected in the (Britz et al., 2010) study. Other studies have correlated the default-mode network with beta-2 (Laufs et al., 2003) or with delta (Mantini et al., 2007)
spectral bands of EEG. Therefore, a direct comparison of EEG and fMRI
studies of rs-fc is complicated by the fact that similar EEG power bands
may be correlated with varying fMRI-generated spatial maps and a single
RSN may be associated with different EEG spectral patterns ( Laufs et al., 2008 and Musso et al., 2010).
The
EEG/MEG studies also point to the manner in which rs-fc studies may be
used to determine efficacy of treatments for tinnitus. In one such
study, (Vanneste and De Ridder, 2011)
employed spontaneous electrical activity measured using EEG to
dissociate the networks of responders from nonresponders. Prior to the
intervention, patients who went on to become responders registered
heightened functional connectivity between the frontal cortex and (a)
the parahippocampus and (b) the subgenual anterior cingulate cortex,
compared to the future non-responders. The responders also differed from
the nonresponders with respect to connectivity of RSNs involving the
auditory cortex and the parahippocampal region. Adamchic et al. (2012)
verified the extent of changes in a pitch-processing network, which
correlated with degree of reduction in tinnitus-related symptoms, using
EEG; those with little or no change in their tinnitus pitch had the
fewest changes to their pitch processing network. The therapy used in
the study (Tass et al., 2012)
attempted to reduce tinnitus-related symptoms by having participants
listen to a series of brief tones of specific frequencies so as to
induce a ‘co-ordinated reset’ of the tonotopic organization near the
tinnitus pitch. Efficacy of repetitive transcranial magnetic stimulation
for those with tinnitus is also beginning to be evaluated using rs-fc
studies of MEG (Muller et al., 2013) and EEG (Fuggetta and Noh, 2012).
4. Comparisons with other disorders
Although
rs-fc has not been used for subtyping of various groups and
differential diagnosis and is only beginning to be used for
investigating treatment efficacy, it has a long history of such usage in
schizophrenia and disorders associated with aging. In this section, we
briefly review the findings from rs-fc studies related to Alzheimer's
disease and schizophrenia, which may provide insights into interpreting
results of tinnitus rs-fc studies and illustrate uses of this tool.
Rs-fc
studies have the potential to be used as diagnostic tools to predict
disease onset and for classifying patients into different prognostic
categories (Horwitz and Rowe, 2011).
This is illustrated via Alzheimer's disease, where patients with mild
cognitive impairment are differentially diagnosed as to whether they
will later develop Alzheimer's disease or will remain stable (Agosta et al., 2012, Binnewijzend et al., 2012, Chen et al., 2011, Greicius et al., 2004 and Koch et al., 2012). Binnewijzend et al. (2012)
specifically address this possibility with a longitudinal study using
43 controls, 39 patients with Alzheimer's disease, and 23 individuals
with mild cognitive impairment. The mild cognitive impairment group
further separated into a group of 7 people that developed Alzheimer's
disease and a larger group of 14 patients that remained stable. Changes
to RSNs were assessed by calculating a functional connectivity score.
Lower scores in the DMN were found in the Alzheimer's disease group when
compared to normal groups. Connectivity scores for the mild cognitive
impairment group were between those of the Alzheimer's disease and
control groups, though not in a statistically significant manner. When
the mild cognitive impairment subgroups were examined, Alzheimer's
disease patients had lower scores than stable mild cognitive impairment
patients, but the scores between the mild cognitive impairment patients
who later developed Alzheimer's disease and the Alzheimer's disease
group itself were not dissimilar. The experimenters point out that this
similarity could be due to the small sample size, particularly in the
converted group (Binnewijzend et al., 2012).
The diagnostic capabilities of rs-fc also have potential applications
in tinnitus, as there is currently no reproducible objective measure of
the disorder. Further, previous studies of subtyping tinnitus has relied
on the symptoms, rather than on pathophysiology (Tyler et al., 2008).
Reproducibility
of rs-fc results has varied depending on the disorder being studied.
For example, alterations to the DMN and attention networks in
Alzheimer's disease have been relatively consistent (Agosta et al., 2012, Binnewijzend et al., 2012, Koch et al., 2012, Li et al., 2012, Zhang et al., 2010 and Zhao et al., 2012). In contrast, results of schizophrenia have been variable (Greicius, 2008).
Although schizophrenia is a cluster of profound neuro-psychiatric
symptoms, a subtype of patients experience phantom perception of sounds,
although there are some fundamental differences with tinnitus, notably
in the interpretation of the sound. In addition, the schizophrenic
patient population is extremely variable and for both of these reasons,
it may be beneficial to examine the work that has been done concerning
rs-fc in schizophrenic patients (comprehensive reviews of schizophrenia
rs-fc studies may be found in Greicius, 2008 and Karbasforoushan and Woodward, 2012. In schizophrenia, connectivity within the DMN has been shown to be both increased (Zhou et al., 2007) and decreased (Bluhm et al., 2007)
relative to controls. Research concerning networks anti-correlated with
the DMN has also produced mixed results; Zhou et al. find increased
inverse correlations between the DMN and other networks, whereas Bluhm
and colleagues find no such effect. This variation could be attributed
to differences in medications taken by subjects, age of participants and
severity of the disease. With regards to the function of the auditory
RSN in auditory/verbal hallucinations, Northoff and Qin (2011)
have proposed a theory in three parts. First, there is increased
activity in the auditory RSN, specifically in the secondary auditory
cortex, when a patient experiences an auditory hallucination. Second,
there are alterations in the activation in the DMN and an irregular
relationship between the DMN and the auditory RSN, though the exact
nature of this interaction is not clear. Lastly, a change in the
relationship between rest and task states in the primary auditory cortex
occurs. Because the auditory RSN exhibits elevated activity at rest,
when a stimulus is presented there is reduced increase in activity when
transitioning to a task state. This hypothesis, though it refers
specifically to auditory hallucinations as a consequence of
schizophrenia, may also be applicable to tinnitus patients. Of
particular note is the third component of the hypothesis, which predicts
that elevated level of auditory response would reduce the rest-to-task
activity difference. This hypothesis has also been proposed for tinnitus
by Melcher et al., 2000 and Melcher et al., 2009
with regards to elevated response in the inferior colliculus to noise
stimuli in the tinnitus group compared to a control group. However, an
interleaved task- and rest-based functional connectivity study that
would explicitly test this hypothesis of reduced rest-to-task activity
difference in tinnitus has not yet been published.
5. Challenges of rs-fc
As
we write this, several new techniques of fc and rs-fc are being
developed and implemented. The different tools used to study rs-fc are
likely not measuring the same thing and as of yet do not index temporal
interactions (Horwitz, 2003 and Horwitz et al., 2005).
A continuing problem of rs-fc studies is with their interpretation.
Another concern when studying rs-fc using fMRI is the amount of noise
produced by the MRI scanner. Though studies have attempted to minimize
the amount of noise perceived by subjects during the scan with
headphones and ear plugs, we cannot completely prevent participants from
hearing some sound. Indeed, scanner noise has been shown to cause some
suppression of the DMN (Perrachione and Ghosh, 2013).
Noise is of particular concern when studying tinnitus. It is important
to question subjects to verify that scanner noise does not mask their
tinnitus sound. Further, the presence of the extraneous scanner noise
may make the sound perceived by tinnitus subjects less salient and
therefore reduce the differences in rs-fc found between tinnitus and
control groups. Tinnitus subjects would not be unique in attempting to
ignore an auditory stimulus at rest; all subjects are dealing with
noise. An alternative is to use sparse-sampling or clustered acquisition
which greatly reduces scanner noise at the expense of fewer scans (Gaab et al., 2007 and Hall et al., 1999).
However, switching the scanner noise on and off may remove subjects
from a resting state as well. Whereas participants may habituate to the
scanner noise during a continuous scan and come closer to a true resting
state, such habituation would be challenging in a clustered-acquisition
paradigm. In addition, sparse sampling collects fewer volumes than
continuous scanning (for instance, 25 in sparse vs. 150 in continuous
scan for a 5 min session) (Perrachione and Ghosh, 2013).
Thus, to achieve the same statistical power, scan time would need to be
significantly increased. Reconstruction of the signal from the sparse
data also complicates the analysis, but likely does create an accurate
portrayal of RSNs. In a functional connectivity study involving normal
hearing healthy participants (Langers and van Dijk, 2011)
found inherent networks to be fairly consistent whether determined
through conventional continuous scanning or via sparse sampling.
Nevertheless, in this same study, the signal spectra were shown to be
better in continuous acquisition. It may therefore be unfavorable to
employ sparse scanning given the reduction in acquired volumes. A study
examining the differences in rs-fc in continuous and sparse scanning
methods should be employed before any definitive conclusions can be
drawn.
Other non-noise
related concerns relate to variation in analysis methods and
experimental design. The optimal method for analyzing resting state data
is yet to be determined. Even the inclusion of different pre-processing
steps is still being debated. Further, resting state data is often
collected in a larger experimental paradigm that includes several tasks.
How long a task can influence the resting state has not been
determined. It is therefore possible that by performing a task
beforehand, the resting state scan is confounded by residual activations
brought about by the task. Different methods of analysis, including
group independent component analysis, graph connectivity analysis, and
seed-based analysis, also have their unique benefits and drawbacks that
need to be kept in mind when assessing rs-fc studies, and there is
currently no standardized method for obtaining and analyzing resting
state data (Cole et al., 2010).
A
third, but probably the most important challenge, to interpreting rs-fc
studies of tinnitus is the heterogeneity of the tinnitus patient
population. There are several ways rs-fc studies can minimize the
heterogeneity of the subject sample. One is to restrict the sample to
individuals with a particular sub-type of tinnitus (e.g., those with
mild or non-bothersome tinnitus), to a specific hearing loss profile
(e.g., normal hearing up to 8 kHz), and to minimize variation in age and
gender. Typically, individuals are classified into sub-types based on
their overall scores on standardized questionnaire, such as the Tinnitus
Questionnaire (Kuk et al., 1990), the Tinnitus Handicap Inventory (Newman et al., 1996) and the Tinnitus Functional Index (Meikle et al., 2011).
However, restricting to an overall score without paying attention to
scores on the different subdomains may not increase homogeneity.
Variability on different subdomains of these questionnaires may reflect
variability of the different cortical networks, which will affect
interpretation of the rs-fc data. Although it is possible to restrict
heterogeneity, it is impossible to have a completely homogenous patient
group. A worthwhile longer-term goal of rs-fc and task-based imaging
studies is to identify subtypes via imaging paradigms, which may allow
us to better interpret imaging data. This has broader implications for
treatment strategies as well.
Understanding
the neural bases of tinnitus using rs-fc is in an exploratory stage.
Therefore it is not surprising that multiple tools have been used with
different assumptions and that no coherent, integrated explanation of
tinnitus has been put forward. One goal of this review was to advance a
qualitative understanding of tinnitus, as determined from the rs-fc
studies published so far. Another goal was to identify the challenges
and gains of this tool and how it may be used in future, as described
next.
6. Future directions
6.1. Combined DTI-rs-fc study
One
direction that remains to be explored in studying neural correlates of
tinnitus is that of combining anatomical and functional connectivity
within the same framework. An anatomical link may not be functionally
engaged in a task or network and a functional connection may encompass
indirect anatomical links. See for instance, the (Simonyan et al., 2009)
study that combined information about white matter tracts obtained
using diffusion tensor imaging (DTI) with BOLD functional connectivity.
DTI studies of tinnitus find altered white-matter tracts connecting
inferior colliculus to the auditory cortex (Crippa et al., 2010), the amygdala and the auditory cortex (Crippa et al., 2010), and the frontal and parietal cortices with the auditory cortex (Lee et al., 2007). However, other studies do not find similar changes in the auditory processing pathways (Husain et al., 2011a). A confounding factor is hearing loss, which is accounted for in the (Husain et al., 2011a),
but not in the other studies. Further, all studies suffer from low
subject numbers and their results are not generalizable. A combined
DTI-rs-fc study may illuminate the reason behind these differences and
better inform the changes occurring along connections between regions.
6.2. Effective connectivity and modeling studies
Effective
connectivity differs from functional connectivity in that it provides a
framework to test both directionality and strength of functional
connections within a specified anatomical model (Horwitz, 2003).
To date, no effective connectivity studies of neural bases of tinnitus
have been published. It is likely that some of the confusion arising
from different functional connectivity studies may be allayed by
employing effective connectivity models. As has been noted before,
interpretation of results obtained from resting-state or task-based
functional connectivity studies is not straightforward (Horwitz, 2003), especially in the context of patient groups, but neural network modeling may provide one way of interpreting the data (Kim and Horwitz, 2009). As with effective connectivity studies, they provide one more tool to supplement rs-fc investigations.
6.3. Temporal dynamics
The
tinnitus and hearing loss resting-state studies reviewed here do not
take into account temporal dynamics, but are rather static portraits of
spontaneous activity acquired over a long period of time, ranging from 5
to 15 min. However, faster changes in the rs-fc as measured by the
recently introduced dynamic BOLD functional connectivity in humans (Chang and Glover, 2010 and Smith et al., 2012) and in the rodent model (Keilholz et al., 2013 and Pan et al., 2010)
may provide further insights into the neural bases of tinnitus and will
provide greater compatibility with the published EEG and MEG results.
Major findings of the dynamic functional connectivity include the fact
that changes in BOLD appear to occur on a scale of a few minutes and
could be correlated with changes in EEG spectra (Tagliazucchi et al., 2012).
Increased alpha and beta power were correlated with decreased
functional connectivity, whereas increased gamma activity was associated
with increased connectivity (Tagliazucchi et al., 2012).
The results of combined EEG/fMRI studies provide a means to integrate
the EEG and fMRI rs-fc data obtained from separate experiments.
6.4. Future rs-fc studies of tinnitus
In
the near future, there may be no standardized method of obtaining and
analyzing rs-fc data. However, within the realm of tinnitus studies, a
more standardized approach may be voluntarily adopted by the
researchers. More than any task-based framework, a resting state
paradigm lends itself to standardization in data acquisition and may be
analyzed in multiple ways, at least one of which may maintain parity
with other rs-fc studies of tinnitus. Of the techniques reviewed in this
article, it is our opinion that seed-based rs-fc is the simplest
technique and has the fewest assumptions. The major assumption made in
seed-based analysis is the choice of the seed region. If several rs-fc
studies chose the same seed region, it would allow for assessment of
replicability and a quantitative meta-analysis of studies. We recommend
that different cortical networks, apart from the auditory network (seeds
in the primary auditory cortex), such as the DAN (seeds in the frontal
and parietal cortices), and the DMN (seeds in the posterior cingulate
cortex and medial prefrontal cortex), be routinely assessed in future
studies because there is gathering evidence of the role of
extra-auditory networks in tinnitus from several brain imaging studies.
Additionally, it is important to account for heterogeneity by carefully
choosing the test population, by controlling for variables such as
degree of hearing loss, age, gender, and various aspects of tinnitus.
Variability in the tinnitus profiles may relate to lateralization, age
of onset, chronicity, duration of tinnitus, loudness and pitch of the
percept and subjective measures of distress.
7. Conclusion
Rs-fc
is in an exploratory stage in unraveling the networks subserving
tinnitus. We have shown that the DMN-limbic and the auditory-limbic
functional connections are altered in tinnitus and may be correlated
with tinnitus-related distress. Although, the auditory-limbic link is
known from task-based fMRI and structural MRI studies, the DMN-limbic
link is unique to rs-fc. The third set of functional links implicated in
tinnitus includes those of the attentional network. A promising avenue
for further research is to focus on a particular network and to detail
the specific aspects of their alteration in tinnitus, which may be
invariant across sub-groups or show distinctions between sub-groups. We
are optimistic about the usage of rs-fc as an objective imaging
biomarker of tinnitus and its uses as a diagnostic tool. This in turn
has important applications in understanding the pathophysiology of the
disorder across different sub-groups, longitudinally, and also in
testing the efficacy of different therapies.
Acknowledgments
We wished to acknowledge the support of Tinnitus Research Consortium to FTH and of the NeuroEngineering
NSF IGERT (Integrative Graduate Education and Research Traineeship) to
SAS. We are grateful to Michelle Hampson for her comments on a version
of this manuscript.
References
- Adamchic et al., 2012
- Psychometric evaluation of visual analog scale for the assessment of chronic tinnitus
- Am. J. Audiol., 21 (2012), pp. 215–225
- |
- Agosta et al., 2012
- Resting state fMRI in Alzheimer's disease: beyond the default mode network
- Neurobiol. Aging, 33 (2012), pp. 1564–1578
- | |
- Bartels et al., 2008
- The additive effect of co-occurring anxiety and depression on health status, quality of life and coping strategies in help-seeking tinnitus sufferers
- Ear Hear., 29 (2008), pp. 947–956
- |
- Bauer, 2004
- Mechanisms of tinnitus generation
- Curr. Opin. Otolaryngol. Head Neck Surg., 12 (2004), pp. 413–417
- |
- Binnewijzend et al., 2012
- Resting-state fMRI changes in Alzheimer's disease and mild cognitive impairment
- Neurobiol. Aging, 33 (2012), pp. 2018–2028
- | |
- Biswal et al., 1995
- Functional connectivity in the motor cortex of resting human brain using echo-planar MRI
- Magn. Reson. Med., 34 (1995), pp. 537–541
- |
- Bluhm et al., 2007
- Spontaneous low-frequency fluctuations in the BOLD signal in schizophrenic patients: anomalies in the default network
- Schizophrenia Bull., 33 (2007), pp. 1004–1012
- |
- Britz et al., 2010
- BOLD correlates of EEG topography reveal rapid resting-state network dynamics
- Neuroimage, 52 (2010), pp. 1162–1170
- | |
- Burton et al., 2012
- Altered networks in bothersome tinnitus: a functional connectivity study
- BMC Neurosci., 13 (2012), p. 3
- Carpenter-Thompson
- Carpenter-Thompson, J.R., Akrofi, K., Schmidt, S.A., Husain, F.T. Affective sound processing illustrates differences in limbic system response of those with and without tinnitus, unpublished data.
- Chang and Glover, 2010
- Time-frequency dynamics of resting-state brain connectivity measured with fMRI
- Neuroimage, 50 (2010), pp. 81–98
- | |
- Chen et al., 2011
- Classification of Alzheimer disease, mild cognitive impairment, and normal cognitive status with large-scale network analysis based on resting-state functional MR imaging
- Radiology, 259 (2011), pp. 213–221
- |
- Cole et al., 2010
- Advances and pitfalls in the analysis and interpretation of resting-state FMRI data
- Front Syst. Neurosci., 4 (2010), p. 8
- Cordes et al., 2000
- Mapping functionally related regions of brain with functional connectivity MR imaging
- AJNR Am. J. Neuroradiol., 21 (2000), pp. 1636–1644
- Crippa et al., 2010
- A diffusion tensor imaging study on the auditory system and tinnitus
- Open Neuroimaging J., 4 (2010), pp. 16–25
- |
- Davis and Rafaie, 2000
- Tinnitus Handbook
- R.S. Tyler (Ed.), Epidemiology of Tinnitus, Singular, San Diego, CA (2000)
- Eggermont and Roberts, 2004
- The neuroscience of tinnitus
- Trends Neurosci., 27 (2004), pp. 676–682
- | |
- Ferreira and Busatto, 2013
- Resting-state functional connectivity in normal brain aging
- Neurosci. Biobehav Rev., 37 (2013), pp. 384–400
- | |
- Fox and Raichle, 2007
- Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging
- Nat. Rev. Neurosci., 8 (2007), pp. 700–711
- |
- Fox et al., 2005
- The human brain is intrinsically organized into dynamic, anticorrelated functional networks
- Proc. Natl. Acad. Sci. U. S. A., 102 (2005), pp. 9673–9678
- |
PDF (3385 K)- Fuggetta and Noh, 2012
- A neurophysiological insight into the potential link between transcranial magnetic stimulation, thalamocortical dysrhythmia and neuropsychiatric disorders
- Exp. Neurol. (2012)
- Gaab et al., 2007
- Assessing the influence of scanner background noise on auditory processing. I. An fMRI study comparing three experimental designs with varying degrees of scanner noise
- Hum. Brain Mapp., 28 (2007), pp. 703–720
- |
- Giaquinto and Nolfe, 1988
- The resting EEG: an useful monitor of brain function in experimental aging research
- Bollettino della Societa italiana di biologia sperimentale, 64 (1988), pp. 109–128
- Golm et al., 2013
- Neural correlates of tinnitus related distress: an fMRI-study
- Hear Res., 295 (2013), pp. 87–99
- | |
- Greicius, 2008
- Resting-state functional connectivity in neuropsychiatric disorders
- Curr. Opin. Neurol., 21 (2008), pp. 424–430
- |
- Greicius et al., 2004
- Default-mode network activity distinguishes Alzheimer's disease from healthy aging: evidence from functional MRI
- Proc. Natl. Acad. Sci. U. S. A., 101 (2004), pp. 4637–4642
- |
- Gu et al., 2010
- Tinnitus, diminished sound-level tolerance, and elevated auditory activity in humans with clinically normal hearing sensitivity
- J. Neurophysiol., 104 (2010), pp. 3361–3370
- |
- Guldenmund et al., 2012
- A default mode of brain function in altered states of consciousness
- Arch. italiennes de biologie, 150 (2012), pp. 107–121
- Hafkemeijer et al., 2012
- Imaging the default mode network in aging and dementia
- Biochim. biophys. acta, 1822 (2012), pp. 431–441
- | |
- Hall et al., 1999
- “Sparse” temporal sampling in auditory fMRI
- Hum. Brain Mapp., 7 (1999), pp. 213–223
- |
- Hampson et al., 2012
- Functional connectivity MR imaging
- S.H. Faro, F.B. Mohamed, M. Law, J.T. Ulmer (Eds.), Functional Neuroradiology: Principles and Clinical Applications, Springer Science+Business Media, LLC, New York, NY (2012), pp. 355–372
- Hampson et al., 2002
- Detection of functional connectivity using temporal correlations in MR images
- Hum. Brain Mapp., 15 (2002), pp. 247–262
- |
- Henry et al., 2005
- General review of tinnitus: prevalence, mechanisms, effects, and management
- J. Speech Lang. Hear Res., 48 (2005), pp. 1204–1235
- |
- Horwitz, 2003
- The elusive concept of brain connectivity
- Neuroimage, 19 (2003), pp. 466–470
- | |
- Horwitz and Rowe, 2011
- Functional biomarkers for neurodegenerative disorders based on the network paradigm
- Prog. Neurobiol., 95 (2011), pp. 505–509
- | |
- Horwitz et al., 1987
- Intercorrelations of regional cerebral glucose metabolic rates in Alzheimer's disease
- Brain Res., 407 (1987), pp. 294–306
- | |
- Horwitz et al., 2005
- Investigating the neural basis for functional and effective connectivity. Application to fMRI
- Philos. Trans. R Soc. Lond B Biol. Sci., 360 (2005), pp. 1093–1108
- |
- Husain et al., 2011a
- Neuroanatomical changes due to hearing loss and chronic tinnitus: a combined VBM and DTI study
- Brain Res., 1369 (2011), pp. 74–88
- | |
- Husain et al., 2011b
- Discrimination task reveals differences in neural bases of tinnitus and hearing impairment
- PLoS One, 6 (2011), p. e26639
- Jastreboff, 1990
- Phantom auditory perception (tinnitus): mechanisms of generation and perception
- Neurosci. Res., 8 (1990), pp. 221–254
- | |
- Kaltenbach et al., 2005
- Tinnitus as a plastic phenomenon and its possible neural underpinnings in the dorsal cochlear nucleus
- Hear Res., 206 (2005), pp. 200–226
- | |
- Karbasforoushan and Woodward, 2012
- Resting-state networks in schizophrenia
- Curr. Top. Med. Chem., 12 (2012), pp. 2404–2414
- |
- Keilholz et al., 2013
- Dynamic properties of functional connectivity in the rodent
- Brain Connect., 3 (2013), pp. 31–40
- |
- Kim and Horwitz, 2009
- How well does structural equation modeling reveal abnormal brain anatomical connections? An fMRI simulation study
- Neuroimage, 45 (2009), pp. 1190–1198
- | |
- Kim et al., 2012
- Alteration of functional connectivity in tinnitus brain revealed by resting-state fMRI? A pilot study
- Int. J. Audiol., 51 (2012), pp. 413–417
- |
- Koch et al., 2012
- Diagnostic power of default mode network resting state fMRI in the detection of Alzheimer's disease
- Neurobiol. Aging, 33 (2012), pp. 466–478
- | |
- Kuk et al., 1990
- The psychometric properties of a tinnitus handicap questionnaire
- Ear Hear., 11 (1990), pp. 434–445
- |
- Langers and Melcher, 2011
- Hearing without listening: functional connectivity reveals the engagement of multiple nonauditory networks during basic sound processing
- Brain Connect., 1 (2011), pp. 233–244
- |
- Langers and van Dijk, 2011
- Robustness of intrinsic connectivity networks in the human brain to the presence of acoustic scanner noise
- Neuroimage, 55 (2011), pp. 1617–1632
- | |
- Laufs, 2010
- Multimodal analysis of resting state cortical activity: what does EEG add to our knowledge of resting state BOLD networks?
- Neuroimage, 52 (2010), pp. 1171–1172
- | |
- Laufs et al., 2008
- Recent advances in recording electrophysiological data simultaneously with magnetic resonance imaging
- Neuroimage, 40 (2008), pp. 515–528
- | |
- Laufs et al., 2003
- Electroencephalographic signatures of attentional and cognitive default modes in spontaneous brain activity fluctuations at rest
- Proc. Natl. Acad. Sci. U. S. A., 100 (2003), pp. 11053–11058
- |
- Leaver et al., 2012
- Cortico-limbic morphology separates tinnitus from tinnitus distress
- Front Syst. Neurosci., 6 (2012), p. 21
- Lee et al., 2007
- Evaluation of white matter structures in patients with tinnitus using diffusion tensor imaging
- J. Clin. Neurosci., 14 (2007), pp. 515–519
- | |
- Levine, 1999
- Somatic (craniocervical) tinnitus and the dorsal cochlear nucleus hypothesis
- Am. J. Otolaryngol., 20 (1999), pp. 351–362
- | |
- Li et al., 2012
- Attention-related networks in Alzheimer's disease: a resting functional MRI study
- Hum. Brain Mapp., 33 (2012), pp. 1076–1088
- |
- Lockwood et al., 2002
- Tinnitus
- New Engl. J. Med., 347 (2002), pp. 904–910
- |
- Lorenz et al., 2009
- Loss of alpha power is related to increased gamma synchronization-A marker of reduced inhibition in tinnitus?
- Neurosci. Lett., 453 (2009), pp. 225–228
- | |
- Lowe et al., 1998
- Functional connectivity in single and multislice echoplanar imaging using resting-state fluctuations
- Neuroimage, 7 (1998), pp. 119–132
- | |
- Lu et al., 1992
- Generator sites of spontaneous MEG activity during sleep
- Electroencephalogr Clin. Neurophysiol., 82 (1992), pp. 182–196
- | |
- Mantini et al., 2007
- Electrophysiological signatures of resting state networks in the human brain
- Proc. Natl. Acad. Sci. U. S. A., 104 (2007), pp. 13170–13175
- |
- Maudoux et al., 2012a
- Connectivity graph analysis of the auditory resting state network in tinnitus
- Brain Res., 1485 (2012), pp. 10–21
- | |
- Maudoux et al., 2012b
- Auditory resting-state network connectivity in tinnitus: a functional MRI study
- PLoS One, 7 (2012), p. e36222
- Meikle et al., 2011
- The tinnitus functional index: development of a new clinical measure for chronic, intrusive tinnitus
- Ear Hear. (2011)
- Melcher et al., 2000
- Lateralized tinnitus studied with functional magnetic resonance imaging: abnormal inferior colliculus activation
- J. Neurophysiol., 83 (2000), pp. 1058–1072
- Melcher et al., 2009
- The auditory midbrain of people with tinnitus: abnormal sound-evoked activity revisited
- Hear Res., 257 (2009), pp. 63–74
- | |
- Morcom and Fletcher, 2007
- Does the brain have a baseline? Why we should be resisting a rest
- Neuroimage, 37 (2007), pp. 1073–1082
- | |
- Muhlau et al., 2006
- Structural brain changes in tinnitus
- Cereb. Cortex, 16 (2006), pp. 1283–1288
- Muller et al., 2013
- rTMS induced tinnitus relief is related to an increase in auditory cortical alpha activity
- PLoS One, 8 (2013), p. e55557
- Musso et al., 2010
- Spontaneous brain activity and EEG microstates. A novel EEG/fMRI analysis approach to explore resting-state networks
- Neuroimage, 52 (2010), pp. 1149–1161
- | |
- Newman et al., 1996
- Development of the tinnitus handicap inventory
- Arch. Otolaryngol. Head Neck Surg., 122 (1996), pp. 143–148
- Northoff and Qin, 2011
- How can the brain's resting state activity generate hallucinations? A ‘resting state hypothesis’ of auditory verbal hallucinations
- Schizophr Res., 127 (2011), pp. 202–214
- | |
- Pan et al., 2010
- Simultaneous FMRI and electrophysiology in the rodent brain
- J. Vis. Exp. JoVE (2010)
- Perrachione and Ghosh, 2013
- Optimized design and analysis of sparse-sampling FMRI experiments
- Front. Neurosci., 7 (2013), p. 55
- Raichle and Snyder, 2007
- A default mode of brain function: a brief history of an evolving idea
- Neuroimage, 37 (2007), pp. 1083–1090 discussion 1097-9
- | |
- Raichle et al., 2001
- A default mode of brain function
- Proc. Natl. Acad. Sci. U. S. A., 98 (2001), pp. 676–682
- |
- Rauschecker et al., 2010
- Tuning out the noise: limbic-auditory interactions in tinnitus
- Neuron, 66 (2010), pp. 819–826
- | |
- Roberts et al., 2010
- Ringing ears: the neuroscience of tinnitus
- J. Neurosci., 30 (2010), pp. 14972–14979
- |
- Salmelin and Hari, 1994
- Characterization of spontaneous MEG rhythms in healthy adults
- Electroencephalography Clin. Neurophysiol., 91 (1994), pp. 237–248
- | |
- Schlee et al., 2009
- Abnormal resting-state cortical coupling in chronic tinnitus
- BMC Neurosci., 10 (2009), p. 11
- |
- Schmidt, in press
- Schmidt, S.A., Akrofi, K., Carpenter-Thompson, J.R., Husain, F.T., Default mode and dorsal attention networks exhibit differential functional connectivity in tinnitus and hearing loss. PLoS ONE, in press.
- Shore, 2011
- Plasticity of somatosensory inputs to the cochlear nucleus–implications for tinnitus
- Hear Res., 281 (2011), pp. 38–46
- | |
- Shulman et al., 1997
- Common blood flow changes across visual Tasks: II. Decreases in cerebral cortex
- J. Cogn. Neurosci., 9 (1997), pp. 648–663
- |
- Simonyan et al., 2009
- Functional but not structural networks of the human laryngeal motor cortex show left hemispheric lateralization during syllable but not breathing production
- J. Neurosci., 29 (2009), pp. 14912–14923
- |
- Smith et al., 2012
- Temporally-independent functional modes of spontaneous brain activity
- Proc. Natl. Acad. Sci. U. S. A., 109 (2012), pp. 3131–3136
- |
- Soddu et al., 2011
- Identifying the default-mode component in spatial IC analyses of patients with disorders of consciousness
- Hum. Brain Mapp, 33 (2011), pp. 778–796
- Tagliazucchi et al., 2012
- Dynamic BOLD functional connectivity in humans and its electrophysiological correlates
- Front. Hum. Neurosci., 6 (2012), p. 339
- Tass et al., 2012
- Counteracting tinnitus by acoustic coordinated reset neuromodulation
- Restorative Neurol. Neurosci., 30 (2012), pp. 137–159
- Tyler et al., 2008
- Identifying tinnitus subgroups with cluster analysis
- Am. J. Audiol., 17 (2008), pp. S176–S184
- |
- Vanneste and De Ridder, 2011
- Bifrontal transcranial direct current stimulation modulates tinnitus intensity and tinnitus-distress-related brain activity
- Eur. J. Neurosci., 34 (2011), pp. 605–614
- |
- Vanneste et al., 2011
- Different resting state brain activity and functional connectivity in patients who respond and not respond to bifrontal tDCS for tinnitus suppression
- Exp. Brain Res., 210 (2011), pp. 217–227
- |
- Vanneste et al., 2010a
- The differences in brain activity between narrow band noise and pure tone tinnitus
- PLoS One, 5 (2010), p. e13618
- Vanneste et al., 2010b
- The neural correlates of tinnitus-related distress
- Neuroimage, 52 (2010), pp. 470–480
- | |
- Vernon, 1997
- Introduction
- J.A. Vernon (Ed.), Tinnitus: Treatment and Relief, Allyn & Bacon, Needham Heights, MA (1997)
- Vincent et al., 2007
- Intrinsic functional architecture in the anaesthetized monkey brain
- Nature, 447 (2007), pp. 83–86
- |
- Weisz et al., 2007a
- The relevance of spontaneous activity for the coding of the tinnitus sensation
- Prog. Brain Res., 166 (2007), pp. 61–70
- | |
- Weisz et al., 2007b
- The neural code of auditory phantom perception
- J. Neurosci., 27 (2007), pp. 1479–1484
- |
- Wineland et al., 2012
- Functional connectivity networks in nonbothersome tinnitus
- Otolaryngol. Head Neck Surg., 147 (2012), pp. 900–906
- |
- Zhang et al., 2010
- Resting brain connectivity: changes during the progress of Alzheimer disease
- Radiology, 256 (2010), pp. 598–606
- |
- Zhao et al., 2012
- Disrupted small-world brain networks in moderate Alzheimer's disease: a resting-state FMRI study
- PLoS One, 7 (2012), p. e33540
- Zhou et al., 2007
- Functional disintegration in paranoid schizophrenia using resting-state fMRI
- Schizophr Res., 97 (2007), pp. 194–205
- | |
- ☆
- This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-No Derivative Works License, which permits non-commercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
- Corresponding author. 901 S. Sixth Street, Champaign, IL 61821, USA. Tel.: +1 217 333 7561; fax: +1 217 244 2235.
- 1
- For the purpose of this review, we will not consider ‘objective’ tinnitus, which often results from physical or vascular reasons and can be objectively heard by the ear or by using a stethoscope.
Copyright © 2013 The Authors. Published by Elsevier B.V. All rights reserved.
