Real-time fMRI feedback training may improve chronic tinnitus

Abstract

Tinnitus consists of a more or less constant aversive tone or noise and is associated with excess auditory activation. Transient distortion of this activation (repetitive transcranial magnetic stimulation, rTMS) may improve tinnitus. Recently proposed operant training in real-time functional magnetic resonance imaging (rtfMRI) neurofeedback allows voluntary modification of specific circumscribed neuronal activations. Combining these observations, we investigated whether patients suffering from tinnitus can (1) learn to voluntarily reduce activation of the auditory system by rtfMRI neurofeedback and whether (2) successful learning improves tinnitus symptoms.
Six participants with chronic tinnitus were included. First, location of the individual auditory cortex was determined in a standard fMRI auditory block-design localizer. Then, participants were trained to voluntarily reduce the auditory activation (rtfMRI) with visual biofeedback of the current auditory activation.
Auditory activation significantly decreased after rtfMRI neurofeedback. This reduced the subjective tinnitus in two of six participants.
These preliminary results suggest that tinnitus patients learn to voluntarily reduce spatially specific auditory activations by rtfMRI neurofeedback and that this may reduce tinnitus symptoms. Optimized training protocols (frequency, duration, etc.) may further improve the results.

Full text

Eur Radiol
DOI 10.1007/s00330-009-1595-z NEURO
Sven Haller
Niels Birbaumer
Ralf Veit
Received: 23 April 2009
Revised: 30 July 2009
Accepted: 5 August 2009
#European Society of Radiology 2009
Real-time fMRI feedback training may improve
chronic tinnitus
Abstract Objectives: Tinnitus
consists of a more or less constant
aversive tone or noise and is asso-
ciated with excess auditory activation.
Transient distortion of this activation
(repetitive transcranial magnetic
stimulation, rTMS) may improve
tinnitus. Recently proposed operant
training in real-time functional
magnetic resonance imaging (rtfMRI)
neurofeedback allows voluntary
modification of specific circum-
scribed neuronal activations.
Combining these observations, we
investigated whether patients
suffering from tinnitus can (1) learn to
voluntarily reduce activation of the
auditory system by rtfMRI neuro-
feedback and whether (2) successful
learning improves tinnitus symptoms.
Methods: Six participants with
chronic tinnitus were included. First,
location of the individual auditory
cortex was determined in a standard
fMRI auditory block-design localizer.
Then, participants were trained to
voluntarily reduce the auditory
activation (rtfMRI) with visual
biofeedback of the current auditory
activation. Results: Auditory activa-
tion significantly decreased after
rtfMRI neurofeedback. This reduced
the subjective tinnitus in two of six
participants. Conclusion: These
preliminary results suggest that
tinnitus patients learn to voluntarily
reduce spatially specific auditory
activations by rtfMRI neurofeedback
and that this may reduce tinnitus
symptoms. Optimized training
protocols (frequency, duration, etc.)
may further improve the results.
Keywords fMRI .BOLD .
Biofeedback .Tinnitus .
Neurofeedback
Abbreviations BOLD: blood
oxygenation level dependent .
DLPFC: dorso-lateral prefrontal
cortex .EEG:
electroencephalography .EPI:
echoplanar imaging .fMRI:
functional magnetic resonance
imaging .rtfMRI: real-time functional
magnetic resonance imaging .rTMS:
repetitive transcranial magnetic
stimulation .VLPFC: ventro-lateral
prefrontal cortex
Introduction
Tinnitus consists of the auditory perception of sounds or
noise not caused or triggered by external auditory stimuli,
and affects millions of people [1]. It is estimated that in 1–
3% of the general population tinnitus becomes chronic and
sufficiently intrusive to interfere with patients’quality of
life [2]. The underlying neuronal mechanism is only
partially understood, and treatment options are limited [3]. It
was shown that tinnitus is associated with over-activation
within the auditory network [4–6]. Further, it was shown that
repetitive transcranial magnetic stimulation (rTMS) over the
auditory area that temporarily disrupts neuronal activations
may alleviate tinnitus symptoms [6–8].
S. Haller (*)
Institute of Radiology,
Department of Neuroradiology,
University Hospital Basel,
Petersgraben 4,
CH 4031 Basel, Switzerland
e-mail: shaller@uhbs.ch
Tel.: +41-61-2652525
Fax: +41-61-2654908
S. Haller
Institute of Neuroradiology,
Department of Imaging and Medical
Informatics, Geneva University
Hospital,
Geneva, Switzerland
N. Birbaumer .R. Veit
Institute of Medical Psychology and
Behavioral Neurobiology,
University of Tübingen,
Tübingen, Germany
N. Birbaumer
Ospedale San Camillo, Instituto di
Ricovero e Cura a Carattere Scientifico,
Venezia, Italy

Biofeedback allows learning voluntary control over
otherwise autonomous physiological parameters by means
of operant training by providing real-time feedback of a
particular physiological change. Biofeedback was first
demonstrated for the autonomous nervous system (heart
rate and skin conductance) in the 1950s [9]. In the nervous
system, biofeedback (also called neurofeedback) classi-
cally uses electroencephalography (EEG), for example, to
restore communication in severely paralyzed patients [10,
11]. Recently, real-time functional magnetic resonance
imaging (rtfMRI) neurofeedback was introduced [12–15].
This non-invasive technique has a substantially higher
spatial resolution compared to EEG [13]. We reason that
this high spatial resolution of rtfMRI implies a substantial
clinical power because it is possible to learn voluntary
control over specific brain areas, while EEG feedback
allows modifications of large cortical areas only. The
majority of the few available rtfMRI biofeedback studies
investigated the principal feasibility of rtfMRI neurofeed-
back and the neuronal mechanisms of neurofeedback in
healthy volunteers [13,14,16]. To date, only one
controlled rtfMRI group study in patients is available that
demonstrates a beneficial effect of rtfMRI neurofeedback
in chronic pain patients [15].
Combining the above-mentioned observations, we
hypothesize that subjects with tinnitus can (1) learn to
voluntarily reduce the hyperactivity of auditory brain areas
by means of rtfMRI neurofeedback and that (2) this reduces
tinnitus symptoms.
Materials and methods
Subjects
The study was approved by the local ethics committee of
the Medical Faculty of the University of Tübingen,
Germany. Six subjects (three females, three males, age
36.0±14.2 years) gave their written informed consent prior
to inclusion. All subjects had continuous, non-pulsating
tinnitus (5 monaural, 1 binaural) for several years. Tinnitus
self-assessments [17,18] were performed before and after
the MRI session.
Task procedure
First, a standard fMRI auditory block-design paradigm was
performed to identify individual primary auditory cortices
(ROI1) with 20 s ON, 20 s OFF bilateral auditory stimula-
tion, with sine tone of 1,000 Hz pulsating at 6 Hz. This
stimulus is known to induce a strong and long-lasting
BOLD response [19]. Thereafter, we applied four rtfMRI
neurofeedback training sessions (each 4 min 24 s). Each
trial started with a 10-s baseline period followed by a
block-design alternating between down-regulation and no-
regulation periods lasting for 18 s each. To account for
unspecific and global BOLD changes, we used a second
ROI (ROI2) posterior and inferior to the primary auditory
area at the temporo-occipital junction and provided the
difference between the target auditory ROI1 and control
ROI2 as feedback. We used two criteria for selecting the
control region: (1) no activation during the auditory
localizer session and (2) the area not involved in tinnitus
or in the established rtfMRI experiments in healthy
controls.
In the down-regulation trials, visual feedback was
presented by means of thermometer bars [20]. During no-
regulation tasks, no feedback signal was given, and the
thermometer display showed no activation changes. During
the no-regulation condition, the subjects were asked to count
back silently. The participants were informed about the data
processing delay of 1.5 s and of the intrinsic physiological
hemodynamic response delay of about 6 s. The subjects were
not instructed to use a certain strategy for down-regulation,
and we recommended using a strategy that already helped
them to reducethe intensity of the tinnitus noise in their daily
life.
We used the Turbo BrainVoyager software package
(www.brainvoyager.com) in combination with in-house
Matlab (www.mathworks.com) scripts for real-time data
analysis.
fMRI data acquisition
Functional images were acquired on a 3-T whole body MR
scanner with a standard 12-channel head coil (Siemens
Magnetom Trio Tim, Siemens Erlangen, Germany). A
standard Echo-Planar Imaging sequence was used [EPI; TR
(repetition time)=1.5 s, matrix size =64 ×64, effective echo
time TE (echo time)=30 ms, flip angle α=70°, bandwidth=
1.954 kHz/pixel; 16 slices; voxel size= 3.3× 3.3× 5.0 mm
3
,
slice gap=1 mm]. Additionally, we acquired an anatomical
T1-weighted whole brain scan (MPRAGE, matrix size=
256× 256, 160 partitions, 1 mm
3
isotropic voxels, TR=
2,300 ms, TE=3.93 ms, TI (inversion time)=1,100 ms,
α=8°).
Offline post-hoc fMRI data analysis
The post-hoc, off-line data analysis was done with the
SPM5 statistical parametric mapping software package
(Wellcome Department of Imaging Neuroscience, London).
Processing included spatial data smoothing (8-mm Gaussian
kernel), temporal drift removal (0.0088-Hz high-pass
filtering) and spatial normalization to the Montreal Neuro-
logical Institute (MNI) space. The six movement regressors
were used as confounds to reduce movement-related
variance. All conditions were modeled with a canonical
hemodynamic response function (HRF) using standard SPM

5 settings. The following contrasts were analyzed: down-
regulation vs. no-regulation and no-regulation vs. down-
regulation. We performed two major analyses.
At the single subject level, we tested the within-subject
learning effects by estimating a linear decrease in activity
over sessions (Table 1). The reported single-subject p-values
were not corrected for multiple comparisons, because we
describe single-subject effects. All identified significant
voxels were within the ROIs of the auditory localizer runs.
At the group level, a fixed-effects group analysis was
performed using the last session of each subject testing for
the amount of successful down-regulation at the end of the
training, i.e., contrasting the down-regulation trials versus
the no-regulation tasks (Fig. 2a,b,e,f,g, Table 2). Effects
were considered as significant using a whole-brain familiy-
wise error rate (FWE) of p<0.001. Additionally, we
performed a linear regression t-test analysis of the
individual BOLD responses in the auditory areas separately
for the right and left auditory areas with session number as
the independent and the parameter estimates of each
subject as the dependent variable (Fig. 2c,d). This test was
performed one-tailed under the strict a priori hypothesis of
decreasing activations over training sessions.
Results
Single subject analysis
Five of the six included subjects successfully learned to
down-regulate their activations in the auditory ROI. An
example subject is illustrated in Fig. 1. The individual peak
areas with linear signal decrease over time are listed in
Table 1.
Group analysis
Regression analyses revealed a significant linear decrease
in the auditory activations over the training sessions
(p<0.05 bilaterally; Fig. 2c, d). The group analysis of
the last training session shows a significant decrease in
activation after rtfMRI neurofeedback training in bilateral
auditory areas (Fig. 2a, b; Table 2). Moreover we found
deactivations in the occipital lobe and in areas comprising
the default brain network during resting states (prefrontal
cortex, precuneus, inferior parietal lobe). Note that this
decrease in activation is spatially restricted to the above-
mentioned areas, not a general decrease in neuronal
activations of the whole brain. Increased activations during
rtfMRI neurofeedback training were found in bilateral
insula extending to the ventro-lateral prefrontal cortex
[VLPFC, the right dorso-lateral prefrontal cortex (DLPFC)
anatomical labeled as middle frontal gyrus and right
occipito-temporal junction].
Behavioral data
Subjects reported no change (N=4) or mild improvement
(N=2) in tinnitus symptoms after rtfMRI neurofeedback
(follow-up questionnaire approximately 2 weeks after
training). None of the patients reported increasing tinnitus
symptoms.
Discussion
The present investigation is based on the combination of
three previous findings: (1) tinnitus is associated with
Table 1 depicts the test of linear signal decrease over training
sessions at the individual level: peak activation of the activation
cluster in MNI standard space, corresponding t-value and p-value
(uncorrected). Except for subject 3, all subjects successfully
decreased the activations by means of rtfMRI neurofeedback
training
Subject Side MNI coordinates t-value p-value
xyz
1 Left −39 −33 15 t=7.05 p<0.0001
Right 48 −24 9 t =6.65 p <0.0001
2 Left −48 −30 9 t =7.68 p <0.0001
Right 60 −21 12 t=8.09 p<0.0001
3 Left Not significant
Right Not significant
4 Left −39 −33 18 t=2.61 p=0.005
Right 48 −30 9 t =3.13 p =0.001
5 Left −36 −27 27 t=3.30 p=0.001
Right Not significant
6 Left −45 −21 21 t=3.12 p=0.001
Right 42 −27 21 t=2.92 p=0.002

excess auditory activation [4–6]; (2) transient reduction or
interruption of this activation (repetitive transcranial mag-
netic stimulation, TMS)may improve tinnitus [6–8]; (3) real-
time fMRI neurofeedback allows learning the voluntary
control of localized neuronal activations through operant
training [12–15].
In accordance with our hypothesis, patients with tinnitus
successfully learned to reduce the auditory activations by
means of rtfMRI neurofeedback. This improved the subjec-
tive tinnitus symptoms in some patients. The current proof of
principle results justify future rtfMRI neurofeedback studies
in tinnitus in larger cohorts and adequate controls.
Concerning the brain activations, rtfMRI neurofeedback
training induced a significant decrease in the activations of
the auditory cortex despite the small sample size of only six
participants, which implies a high potential effect of rtfMRI
neurofeedback. Our data support findings of recent studies
showing that with rtfMRI-neurofeedback people can learn
to control changes of their BOLD signal in circumscribed
brain areas within a few sessions [14,15,20]. Only one of
six participants did not manage to down-regulate the
activations in the primary auditory cortex.
Although the task of the present study was to decrease
activations within the auditory cortex as discussed above,
the post-hoc data analysis identified additional and
spatially separated regions with decreasing activations,
although these areas were not included in the feedback
display of the rtfMRI training: the medial frontal cortex, the
precuneus and the angular gyrus extending to the inferior
parietal lobe. These areas are part of the ‘default brain
network’[21,22] and consistently active during resting
states or show deactivations during cognitively demanding
tasks [22,23]. Conversely, increasing activations were
found in the insula extending to the bilateral VLPFC and
the right DLPFC. The increasing activation within these
regions during down-regulation of the primary auditory
area might suggest a functional role of these areas in the
down-regulation process. These areas were reported during
self-regulation of emotional responses using cognitive
reappraisal strategies or during diverting attention from
emotional arousal stimuli [24,25]. It was proposed that the
insula might play a critical role in mediating the influence of
peripheral autonomic arousal on consciously experienced
emotional states [26,27]. Correspondingly, the subjects in
the present study primarily used positive events (remember-
ing pleasurable events like a holiday) or relaxation-related
strategies (‘autogenic training’) to decrease the activation in
the auditory cortex.
Concerning the behavioral effects, we cannot expect that
in this proof-of-concept study four training sessions within
a single day will evoke a strong effect, in particular because
tinnitus is a chronic condition lasting for several years in
Table 2 Lists those areas with a significant decrease in activation
(top) or increase (bottom) as a result of the rtfMRI neurofeedback
training sessions. Brain region of the decreasing/increasing activa-
tion, peak coordinate x,y,z in Montreal Neurological Institute (MNI)
space and t-value (family-wise error, FWE, rate corrected, p<0.001)
Brain regions MNI coordinates t-value
xyz
Decreasing activation during rtfMRI training
Occipital lobe R 27 −96 6 8.99
Occipitallobe L −30 −93 −3 8.24
Precuneus 0 −54 42 8.74
Angular gyrus R 45 −69 39 8.21
Angular gyrus L −42 −69 39 6.80
Superior medial frontal cortex L −9 60 33 7.13
Superior medial frontal cortex R 0 66 24 6.65
Heschl gyrus R 39 −27 18 6.15
Heschl gyrus R 60 −12 15 5.51
Heschl gyrus L −57 −12 18 5.05
Middle temporal gyrus L −60 −48 −9 5.24
Increasing activation during rtfMRI training
Insula R 33 27 0 7.96
Middle temporal gyrus R 39 −66 15 7.18
Frontal Inferior opercularis R 51 9 21 6.98
Middle frontal gyrus R 36 42 21 5.69
Insula L −36 18 −6 6.26
Middle occipital L −30 −75 21 5.55

our patients. Correspondingly, the mean activation
continuously decreased within the first four training
sessions (Fig. 2c,d). On the other hand, the observed
positive minor behavioral effect despite only 1 training
day suggests a substantial potential effect of rtfMRI
neurofeedback in tinnitus. Necessary optimizations of
rtfMRI training include adjustment of the number,
duration and frequency of the rtfMRI training sessions.
Learning is difficult in the absence of guidelines for
mental strategies and can lead to a drop of motivation,
especially in the uncomfortable environment of the MRI
scanner [15,28]. The rtfMRI biofeedback algorithm has
Fig. 1 Illustrates the experi-
mental setup for a single subject.
First, a standard fMRI block-
design auditory localizer in
combination with real-time data
analysis was used to identify the
individual auditory areas (a,
screen-shot of the real-time data
analysis using Turbo Brain-
voyager, www.brainvoyager.
com). The right auditory area
(green square) and left auditory
area (red square) illustrate a
clear task-related BOLD re-
sponse (graphs on the right-hand
side with baseline periods in
blue and auditory stimulation
periods in green). Note the
hemodynamic delay of the
BOLD response. The third,
lowest graph on the right-hand
side illustrates the on-line motion
correction. In a second step, the
individually defined auditory
areas were used for the rtfMRI
neurofeedback. A thermometer
bar was visually presented to the
subjects inside the MRI scanner
(b) that indicated the current
BOLD activation in the auditory
region of interest. Subjects trained
to down-regulate this activation.
The presented example illustrates
a current down-regulation
(thermometer below the mean).
The post-hoc data analysis of a
single subject depicts those areas
that are down-regulated after the
rtfMRI biofeedback training (c).
Additionally, the evolution of the
BOLD activations, illustrated as
BETA estimate contrast (in
arbitrary units) over the four
training sessions, is illustrated for
left (d, at locations X −39, Y −33,
Z 12 in Montreal Neurological
Institute MNI space) and right
(e, at locations X 48, Y −24, Z 9)
auditory areas

to be optimized, and the training protocol should be
adapted to the individual patient.
Previous rTMS studies improved tinnitus [6–8]; however,
the effect is transient and depends on the presence of the
rTMS device. The rtfMRI neurofeedback technique has
the advantage that once a participant has learned to control
the individual auditory activations, this strategy can be used
in every day life to improve tinnitus. Previous EEG
biofeedback studies in tinnitus successfully improved
tinnitus symptoms by up-regulating alpha-activity and
down-regulating beta-activity [29] or by enhancement of
tau activity within the alpha frequency range and concom-
itant reduction in delta power [30]. As compared to these
EEG studies, the major advantage of rtfMRI biofeedback is
the much higher spatial specificity that allows for selective
reduction of the auditory activation only.
This proof of principle study aims to illustrate the
clinical potential of rtfMRI neurofeedback. Tinnitus was
chosen as an example because of the existing clear apriori
hypotheses and the localized target region. The presented
principle of rtfMRI neurofeedback represents a novel
‘therapeutic’instead of the usual diagnostic application of
Fig. 2 Illustrates the effect of the
rtfMRI neurofeedback training of
the group analysis (N =6).
Decreasing activations after
rtfMRI neurofeedback are present
in the bilateral auditory area
(yellow circles in aand b)andin
the default brain network
(prefrontal cortex, precuneus,
inferior parietal lobe). The
additional linear regression
analysis of the activations in the
left (c)andright(d) auditory area
demonstrates a significant
decrease over the four training
sessions. Increasing activations
(E-F) were present in bilateral
insula extending to the bilateral
VLPFC and right DLPFC, and
right occipito-temporal junction.
SPM convention, left hemisphere
on left hand side

MRI. This principle might be transferred to other diseases
with presumed excess neuronal activations, such as auditory
or visual hallucinations. Likewise, rtfMRI biofeedback might
also be used to increase neuronal activations, e.g., in stroke
patients with motor disorders, as increased motor activity
might improve motor dysfunctions [31].
A major limitation of the presented proof of concept
investigation evidently is the small sample size.
The frequency of the auditory localizer stimulus was
identical in all subjects despite differences in the subjective
tinnitus frequency. Given the established tonotopic orga-
nization of the auditory cortex, [32], we reason that the
auditory stimulus of the functional localizer experiment
should ideally match the subjective tinnitus to detect the
exact sub-region of the auditory area. Due to the com-
parably small distance of the tonotopic distribution in
relation to size of the volumes-of-interest, future studies are
needed to determine whether the additional effort of
generating individually tuned auditory stimuli outweigh
the more convenient use of a standardized “fixed”auditory
stimuli, in particular with respect to potential clinical
applications of the presented technique.
A general concern in rtfMRI neurofeedback is related to
the spatial specificity of the learned modification of brain
activations. The BOLD response is influenced, for
example, by respiration and the related blood level of
carbon dioxide [33]. Subjects might simply hyperventilate
to reduce the BOLD activations that also—but not
specifically—includes the auditory regions. To control
for global effects, we provided feedback as the difference
between the target auditory area against a distant control
region. Additionally, the post-hoc group comparison of
brain activations after rtfMRI training versus prior to
training (Fig. 2) confirms that the subjects learned a region-
specific down-regulation of the auditory areas.
Another limitation is the absence of control groups. The
only available clinical rtfMRI group study found a beneficial
effect of neurofeedback in patients (N=8) with chronic pain
[15]. The authors applied pain (noxious thermal stimulus) in
healthy controls using ‘true’rtfMRI neurofeedback as well as
(1) training, without rtfMRI information, (2) purely
behavioral training, (3) rtfMRI information derived from a
brain region not involved inaffective pain processing and (4)
rtfMRI information of other subjects (false or sham
feedback). None of these healthy control groups without
the true rtfMRI neurofeedback was able to reduce the
perceived pain in response to the noxious thermal stimulus.
Conclusions
In conclusion, we could prove the principle that patients
with tinnitus successfully learned to reduce the auditory
activations by real-time functional magnetic resonance
imaging neurofeedback and that this improved tinnitus
symptoms in at least two out of six patients.
Acknowledgements We thank all subjects for participation in the
study.
Conflict of interest No conflicts of interest.
References
1. Heller AJ (2003) Classification and
epidemiology of tinnitus. Otolaryngol
Clin North Am 2:239–248
2. Dobie RA (2003) Depression and
tinnitus. Otolaryngol Clin North Am
2:383–388
3. Dobie RA (1999) A review of
randomized clinical trials in tinnitus.
Laryngoscope 8:1202–1211
4. Muhlnickel W, Elbert T, Taub E et al
(1998) Reorganization of auditory
cortex in tinnitus. Proc Natl Acad Sci
U S A 17:10340–10343
5. Andersson G, Lyttkens L, Hirvela C et
al (2000) Regional cerebral blood flow
during tinnitus: a PET case study with
lidocaine and auditory stimulation.
Acta Otolaryngol 8:967–972
6. Kleinjung T, Eichhammer P, Langguth
B et al (2005) Long-term effects of
repetitive transcranial magnetic stimu-
lation (rTMS) in patients with chronic
tinnitus. Otolaryngol Head Neck Surg
4:566–569
7. Rossi S, De Capua A, Ulivelli M et al
(2007) Effects of repetitive transcranial
magnetic stimulation on chronic
tinnitus. A randomised, cross over,
double blind, placebo-controlled study.
J Neurol, Neurosurg Psychiatry
8. Plewnia C, Reimold M, Najib A et al
(2007) Dose-dependent attenuation of
auditory phantom perception (tinnitus)
by PET-guided repetitive transcranial
magnetic stimulation. Hum Brain Mapp
3:238–246
9. Miller NE (1975) Clinical applications
of biofeedback: Voluntary control of
heart rate, rhythm, and blood pressure.
In: Russel HI (ed) New horizons in
cardiovascular practice. University
Park Press, Baltimore, pp 239–249
10. Kubler A, Kotchoubey B, Hinterberger
T et al (1999) The thought translation
device: a neurophysiological approach
to communication in total motor
paralysis. Exp Brain Res 2:223–232
11. Birbaumer N, Ghanayim N,
Hinterberger T et al (1999) A spelling
device for the paralysed. Nature
6725:297–298
12. Weiskopf N, Veit R, Erb M et al (2003)
Physiological self-regulation of
regional brain activity using real-time
functional magnetic resonance imaging
(fMRI): methodology and exemplary
data. Neuroimage 3:577–586
13. Weiskopf N, Scharnowski F, Veit R et
al (2004) Self-regulation of local brain
activity using real-time functional
magnetic resonance imaging (fMRI). J
Physiol Paris 4–6:357–373

14. Weiskopf N, Sitaram R, Josephs O et al
(2007) Real-time functional magnetic
resonance imaging: methods and
applications. Magn Reson Imaging
6:989–1003
15. deCharms RC, Maeda F, Glover GH et
al (2005) Control over brain activation
and pain learned by using real-time
functional MRI. Proc Natl Acad Sci
U S A 51:18626–18631
16. deCharms RC (2007) Reading and
controlling human brain activation
using real-time functional magnetic
resonance imaging. Trends Cogn Sci
11:473–481
17. Zenner HP, De Maddalena H (2005)
Validity and reliability study of three
tinnitus self-assessment scales:
loudness, annoyance and change. Acta
oto-laryngologica 11:1184–1188
18. Goebel G, Hiller W (1994) [The
tinnitus questionnaire. A standard
instrument for grading the degree of
tinnitus. Results of a multicenter study
with the tinnitus questionnaire]. HNO
3:166–172
19. Seifritz E, Esposito F, Hennel F et al
(2002) Spatiotemporal pattern of neural
processing in the human auditory
cortex. Science 5587:1706–1708
20. Caria A, Veit R, Sitaram R et al (2007)
Regulation of anterior insular cortex
activity using real-time fMRI. Neuro-
image 3:1238–1246
21. Biswal B, Yetkin FZ, Haughton VM et
al (1995) Functional connectivity in the
motor cortex of resting human brain
using echo-planar MRI. Magn Reson
Med 4:537–541
22. Damoiseaux JS, Rombouts SA,
Barkhof F et al (2006) Consistent
resting-state networks across healthy
subjects. Proc Natl Acad Sci U S A
37:13848–13853
23. Raichle ME, MacLeod AM, Snyder AZ
et al (2001) A default mode of brain
function. Proc Natl Acad Sci U S A
2:676–682
24. Ochsner KN, Ray RD, Cooper JC et al
(2004) For better or for worse: neural
systems supporting the cognitive down-
and up-regulation of negative emotion.
Neuroimage 2:483–499
25. Eippert F, Veit R, Weiskopf N et al
(2007) Regulation of emotional
responses elicited by threat-related
stimuli. Hum Brain Mapp 5:409–423
26. Critchley HD, Wiens S, Rotshtein P et
al (2004) Neural systems supporting
interoceptive awareness. Nat Neurosci
2:189–195
27. Critchley HD, Melmed RN,
Featherstone E et al (2002) Volitional
control of autonomic arousal: a
functional magnetic resonance study.
Neuroimage 4:909–919
28. Sitaram R, Caria A, Veit R et al (2007)
FMRI brain-computer interface: a tool
for neuroscientific research and treat-
ment. Comput Intell Neurosci 25487
29. Gosepath K, Nafe B, Ziegler E et al
(2001) [Neurofeedback in therapy of
tinnitus]. HNO 1:29–35
30. Dohrmann K, Weisz N, Schlee W et al
(2007) Neurofeedback for treating
tinnitus. Prog Brain Res 473–485
31. Nelson LA (2007) The role of
biofeedback in stroke rehabilitation:
past and future directions. Top Stroke
Rehabil 4:59–66
32. Bilecen D, Scheffler K, Schmid N et al
(1998) Tonotopic organization of the
human auditory cortex as detected by
BOLD-FMRI. Hear Res 1–2:19–27
33. Cohen ER, Ugurbil K, Kim SG (2002)
Effect of basal conditions on the
magnitude and dynamics of the blood
oxygenation level-dependent fMRI
response. J Cereb Blood Flow Metab
9:1042–1053