Methods A 50 year-old male patient with moderate-to-severe bilateral hearing loss (Fig.1) and longstanding bilateral tonal tinnitus underwent diagnostic electrocorticographic epilepsy monitoring. He had tinnitus unrelated to his seizure phenomenology. Coverage included high-impedance depth electrodes in left primary auditory cortex and grid electrodes over large parts of left auditory cortex (164 total intracranial electrodes). To investigate dynamic tinnitus correlates the subject’s tinnitus was repeatedly transiently suppressed using residual inhibition (RI) [1] following 30s masking stimuli (white noise). The subject provided feedback on tinnitus loudness at regular intervals (minimum 10 sec) for 3 blocks following each masker (Fig.1). For each trial spectral power was calculated in 10s epochs (sub-epoched into 1 s average) in 9 standard frequency bands (Fig.4f), following masker offset. For all analyses (oscillatory power, local cross-frequency coupling, within-frequency long-range coupling) Pearson product moment correlation coefficient was calculated between subjective tinnitus intensity and the measure of interest across trials for each frequency band for each electrode.

W. Sedley1,2, P.E. Gander1, S. Kumar2, H. Oya1, C.K. Kovach1, K.V. Nourski1, H. Kawasaki1, M.A. Howard III1, T.D. Griffiths1,2

Direct mapping of a cortical tinnitus network

References1 Roberts LE. Residual inhibition. Prog Brain Res 166:487–95, 2007.

2 de Ridder D, Vanneste S, Weisz N, Londero A, Schlee W, Elgoyhen AB, Langguth B. An integrative model of auditory phantom perception: Tinnitus as a unified percept of interacting separable subnetworks. Neurosci Biobehav Rev 44:16-32, 2013.

Introduction Tinnitus occurs when damage to the peripheral auditory system leads to

spontaneous brain activity that is interpreted as sound. Many types of brain activity show abnormalities in association with tinnitus, but it is not clear which of these relate to the phantom sound itself, as opposed to predisposing factors or secondary consequences. Direct demonstration of the core tinnitus correlates requires high-precision recordings of neural activity combined with a behavioral paradigm in which the perception of tinnitus is manipulated and accurately reported upon by the subject. This has thus far not been possible in animal or human research. Here we present extensive intracranial recordings from an awake, behaving tinnitus patient during short-term modifications in perceived tinnitus loudness, permitting a robust characterization of the core tinnitus brain network.

1Department of Neurosurgery, The University of Iowa, Iowa City, IA, USA, 2Institute of Neuroscience, Newcastle University Medical School, Newcastle-Upon-Tyne, UK

Conclusions In a tinnitus patient with typical symptomatology, we provide the first clear demonstration of a distributed cortical ‘tinnitus system’, which incorporates a large proportion of the cerebral cortex, and all of the major oscillatory frequency bands. Tinnitus modulations were associated with complex restructuring of widespread within- and between-region neural coupling. The extent and cortical areas found to be involved with tinnitus change are in general agreement with recent proposals of a tinnitus network of brain regions [2].

Local cross-frequency coupling changes with tinnitus suppression6

A pre-operative brain reconstruction is shown with electrode locations (black dots) estimated from the post-implantation CT and MRI scans. Circled electrodes denote significant power changes in a given frequency range following tinnitus suppression (see legend). a Left hemisphere. i) Coronal sections of Heschl’s gyrus electrode locations. Heschl’s gyrus is shaded dark grey. ii) Heschl’s gyrus reconstruction with electrode locations. b Ventral view. i) Axial section line tracings showing electrode trajectory to left (top) and right (bottom) amygdala (coloured regions). ii) Axial section line tracing showing posterior hippocampal electrode trajectory.

4 Power changes with tinnitus suppression

Contact informationWilliam Sedley, MD william.sedley@ncl.ac.ukPhillip Gander, PhD phillip-gander@uiowa.edu

Acknowledgements We thank Rich Tyler for clinical interview and Haiming Chen for help with data collection. This study was supported by Wellcome Trust Senior Clinical Fellowship WT091681, NIH R01-DC04290, R01-DC00657, UL1RR024979, and the Hoover Fund.

Delta-band (1-4Hz) inter-channel phase locking value (PLV) changes coinciding with tinnitus suppression. Significant PLVs were summarized using a principal component analysis with one positive (red) and two negative (blue, cyan) PCs. Direction of PLV change is shown in the legend. Each principal component indicates two separate sets of electrode weights (solid and dashed circles), the size of the circle denotes the magnitude of PLV change.a Left hemisphere. i) and ii) As describe in Fig.4. b Ventral view. i) and ii) As describe in Fig.4.

5 Delta phase locking value changes with tinnitus suppression

• Significant changes in local cross-frequency envelope coupling occurring with tinnitus suppression. Representation of a-c are described in e,f. Electrode sites with significant coupling changes are outlined with green boxes in Fig.4 (24 - posterior superior temporal gyrus; 164 - lateral Heschl’s gyrus; 173 - posterior medial temporal lobe). The plots are in polar co-ordinates on the complex plane, with the horizontal axis representing real-valued (non-lagged) coupling, and the vertical imaginary (phase-lagged) coupling. Distance from the origin is the magnitude of coupling and the angle the phase delay associated with the coupling. Coupling on the right-hand side is positive, and on the left is inverse or 'anti-coupling'. Above the origin denotes the lower frequency leading, and below the origin the higher frequency leading. Coupling was measured from the phase-locking value using the envelope data from the different frequencies. Significant changes in magnitude are shown with the size of the circle and changes in angle as rotation of the arrow. Red is baseline and black is during tinnitus suppression.

• The number by each plot indicates the electrode number at which coupling is illustrated, and the Greek letters show which frequency bands the coupling being illustrated is between. HG = Heschl’s gyrus. STG = superior temporal gyrus.

PHC = parahippocampal cortex. PLV = phase-locking value.

Tinnitus characteristics and experimental paradigm1

Tinnitus modulation single trial schematic

30s masker

. . .

pure tone or white noise

loudnessprompt

10s silence

loudnessprompt

10s silence

response time(variable)

loudnessprompt

10s silence

response time(variable)

response time(variable)

loudnessprompt

response time(variable)

Audiogram

Approximately half of the trials resulted in tinnitus modulation, almost all of which were in the direction of suppression (residual inhibition). The behavioural results were nearly identical on two separate days (upper and lower rows). RI was induced by white noise.

Experiment summary

Summary of findings and proposed tinnitus networks7

• Summary of findings shown in (a), a schematic of the notation used in a (b), the types of brain activity characterizing each tinnitus sub-network (c), and the anatomical distribution of each of these sub-networks according to this categorization scheme (d). White text boxes with solid and dashed outlines indicate superficial and deep areas respectively.

• Three tinnitus networks are proposed: Tinnitus Driving Network (TDN), Tinnitus Perception Network (TPN), Tinnitus Memory Network (TMN), with interaction hub sites linking them together.

• S1/M1 = primary somatosensory and motor cortex, IPC = inferior parietal cortex, STG = superior temporal gyrus, HG = Heschl’s gyrus, MTG = middle temporal gyrus, aMTL = anterior mesial temporal lobe, PHC = parahippocampal cortex, TP = temporal pole. See Fig.6f for frequency bands.

LNAmyg.

Hip.

1

32

8

25

129

136 137

144

65

69

70 81 86

9074

154153

214213

212219

218228 232 236

145

146147

148149150

151

152

33

209

210

211

215

216217

225

226227

229233

241

244

248

250

166

167

168

40

41

48

49

56

57

64

165

161162

163164

163

161

162

164

231235

16

24

9

17

170171

172

193

195196

197198

194

234230

249

247246 245 243 242

173

174

175

176

84

S

I

LR

R

L

PA

S

I

PA

L

R

PA

ai

aii

a

bbi

bii

10mm

0.2

0.4

0.6r

1-12Hz dec.

4-12Hz inc.

20-148Hz inc.

Cross-frequency coupling

Response to 5 kHz tone

LNAmyg.

Hip.

1

32

8

25

129

136 137

144

65

69

70 81 86

9074

154153

214213

212219

218228 232 236

145

146147

148149150

151

152

33

209

210

211

215

216217

225

226227

229233

241

244

248

250

166

167

168

40

41

48

49

56

57

64

165

161162

163164

163

161

162

164

231235

16

24

9

17

170171

172

193

195196

197198

194

234230

249

247246 245 243 242

173

174

175

176

84

S

I

LR

R

L

PA

S

I

PA

L

R

PA

ai

aii

a

bbi

bii

10mm

Delta (1-4 Hz) coherence changes per .42 units tinnitus suppression

n mx

Decreased positive

Increased inverse

.1

.3

.5

Tinnituspitch match

MML = 60 dB HL

.125 .25 .5 1 1.5 2 3 4 6 8 9 10 11.2 12.5 14

-10

0

10

30

50

70

90

110

Frequency (kHz)

dB H

L

Left ear

Right ear

-1 (’Quiet’)

0 (’Normal’)

1 (’Loud’)

10 min

Day 1

Day 2

White noisemasker presentation

No task or stimulus(data used for analysis)

Rating prompt orstart/�nish time

Tinnitusratings

Partial tinnitusratings

a: Lat. HG

b: PHC

c: Post. STG d: IPC

164 164 164

δ<->α θ<->β2δ<->β2

β1<->γ1α<->β2θ<->β2

α<->β2δ<->αα<->β2

173 173 173

23 24 132

0(R+)

+/-π(R-)

π/2 (I+)

-π/2 (I-)

1/8

1/4

1/2

1/16

Synchronous

Inve

rse

Lower frequency leading

Higher frequency leading

e f

δ (delta) = 1-4 Hzθ (theta) = 4-8 Hzα (alpha) = 8-12 Hz

β1 (beta 1) = 12-20 Hzβ2 (beta 2) = 20-28 Hzγ1 (gamma 1) = 28-44 Hz

Baselinetinnitus

Suppressedtinnitus

Coup

ling

mag

nitu

deCo

uplin

gan

gle

Projected couplingtrajectory

with tinnitussuppression

S1/M1Somatic sensation/cross-modal input

TP (semantic/emotion)

MTG (auditory awareness/

comprehension)

aMTL(emotion/memory)

Auditory thalamus(cortical input)

HG(early auditory analysis)

IPC(auditory memory/

multimodal semantic/auditory awareness)

PHC (memory)

STG (auditory analysis)

δθα β1,2γδ<->αβ,θ<->βδθα β2γδ<->α<->β

α

δθα

θα θα<->β<->γ β1,2γ

δθα β2γ

δθ

δθ α β2γ

δθα β2γ

a

α<->β

β1,2

b

HG(early auditory analysis)

δθα β1,2γδ<->αβ,θ<->β

Area namePutative function

Low-frequencypower changes

Cross-frequencycoupling changes

High-frequencypower changes

Powerdecrease (δ +/- θα)

Powerincrease (β2γ)

Coherencedecrease (δ)

Tinnitus drivingnetwork (TDN):

Tinnitus perceptionnetwork (TPN):

Network interactionhubs:

<-> Cross-frequencycoupling change

c

Powerincrease (α +/- θ)

Tinnitus memorynetwork (TDN):

d

S1/M1 IPC

STG HG

aMTL

TPPHC

Thalamus

MTG

S1/M1 IPC

STG HG

aMTL

TPPHC

Thalamus

MTG

S1/M1 IPC

STG HG

aMTL

TPPHC

Thalamus

MTG

S1/M1 IPC

STG HG

aMTL

TPPHC

Thalamus

MTG

TDN

TMN

TPN

Interaction

2 Peri-masker power changes

Time-frequency representations are shown for the time period before, during, and after the presentation of 30s noise maskers (gray shaded region) used to elicit residual inhibition. a Power changes are compared for trials which resulted in a suppression of tinnitus (RI) to those in which the tinnitus was unchanged (Non-RI). Frequency bands are described in Fig.6f. i) RI trials with mean and SD of the time period used for analysis following the loudness rating. ii) Non-RI trials with mean and SD of the time period used for analysis following the loudness rating. iii) T-score difference of Non-RI subtracted from RI trials. b Trial by trial power changes sorted according to the time taken to give a rating. Each row represents one trial; the horizontal lines indicate times when either the masker was playing or the subject was considering or providing a rating; time periods without a horizontal line were used for analysis. i) Delta band power. Upper panel shows data for RI trials, with Non-RI trials on the lower panel. i) Delta band power. Upper panel shows data for RI trials, with Non-RI trials on the lower panel. ii) Gamma band power (20-148Hz). Upper panel shows data for RI trials, with Non-RI trials on the lower panel.

3 Power changes with tinnitus suppression across days

Oscillatory power changes correlating with tinnitus suppression are shown for two experimental sessions completed on subsequent days. Three data matrices are displayed, corresponding to (left to right) days 1 and 2 of the experiment, and the combined results thresholded at p < 0.05 corrected. Rows in the matrices represent individual electrodes (numbers correspond to those in Figure 4) and columns individual frequency bands (Hz). Color values denote the correlation coefficients (Pearson’s r) between partial tinnitus suppression and power in any specific electrode/frequency combination. Cool colors indicate power decreases, with tinnitus suppression, and hot colors power increases. In the thresholded plot, gray colors indicate the absence of significant power change. S1/M1 = primary somatosensory/motor cortex, TP = temporal pole, HG = Heschl’s gyrus, aMTL = anterior mesial temporal lobe, PHC = parahippocampal cortex.

ai: RI

−0.4

−0.2

0

0.2

0.4

aii: Non−RI

−0.4

−0.2

0

0.2

0.4

aiii: RI vs Non−RI

Mask on Mask off

−6−4−20246

ERSP (dB)

ERSP (dB)

T score

First RI data block

First non-RI data block

5 2510 2015Post-masker time (s)

DeltaThetaAlphaBeta1Beta2

Gamma2Gamma1

Gamma3Gamma4

DeltaThetaAlphaBeta1Beta2

Gamma2Gamma1

Gamma3Gamma4

DeltaThetaAlphaBeta1Beta2

Gamma2Gamma1

Gamma3Gamma4

−1

−0.5

0

0.5

1

bi: Delta

bii: Gamma

−0.2−0.100.10.20.3

−0.3

ERSP (dB)

ERSP (dB)

RI t

rials

(by

resp

onse

tim

e)N

on-R

I tria

ls(b

y re

spon

se ti

me)

RI t

rials

(by

resp

onse

tim

e)N

on-R

I tria

ls(b

y re

spon

se ti

me)

Posteriortemporal,

occipital

1

3233

64

Posteriorinferior

temporal

Temporal,S1/M1

65

90

Parietal,occipital

129

144

Temporal pole145

154HG (161-4)

(R) TP (165-8)(R) aMTL (169-72)

PHC (173-6)aMTL (193-200)

Lesion (201-4)

Temporalpole

209

236

(R) Temporal pole241

250

Del

ta (1

-4)

Thet

a (4

-8)

Alph

a (8

-12)

Beta

1 (1

2-20

)

Beta

2 (2

0-28

)

Gam

ma1

(28-

44)

Gam

ma2

(44-

60)

Gam

ma3

(60-

92)

Gam

ma4

(92-

148)

Del

ta (1

-4)

Thet

a (4

-8)

Alph

a (8

-12)

Beta

1 (1

2-20

)

Beta

2 (2

0-28

)

Gam

ma1

(28-

44)

Gam

ma2

(44-

60)

Gam

ma3

(60-

92)

Gam

ma4

(92-

148)

Del

ta (1

-4)

Thet

a (4

-8)

Alph

a (8

-12)

Beta

1 (1

2-20

)

Beta

2 (2

0-28

)

Gam

ma1

(28-

44)

Gam

ma2

(44-

60)

Gam

ma3

(60-

92)

Gam

ma4

(92-

148)

0

- 0.2

- 0.4

- 0.6

- 0.8

- 1

0.2

0.4

0.6

0.8

1

Pearson’s r

Day 1 Day 2 Combined (thresholded)