YNIMG-11264; No. of pages: 10; 4C:2, 4, 5, 6, 7, 9 NeuroImage xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

NeuroImage journal homepage: www.elsevier.com/locate/ynimg

F

O

5 6 7 8

a

9

a r t i c l e

10 11 12

Article history: Accepted 2 April 2014 Available online xxxx

13 14 15 16 17

Keywords: Neuronal oscillations EEG NIRS Brain hemodynamics

Neurophysics Group, Department of Neurology, Campus Benjamin Franklin, Charité, University Medicine, Hindenburgdamm 30, 12203 Berlin, Germany Berlin NeuroImaging Center, University Hospital Charité, Charitéplatz 1, 10117 Berlin, Germany Department of Neurology, Virchow Klinikum, University Medicine, Augustenburger Platz 1, 13353 Berlin, Germany d Bernstein Center for Computational Neuroscience, Philippstraße 12, 10115 Berlin, Germany

R O

b c

i n f o

a b s t r a c t

P

4

Vadim V. Nikulin a,d,⁎,1, Tommaso Fedele a,1, Jan Mehnert b, Axel Lipp c, Cornelia Noack c, Jens Steinbrink b, Gabriel Curio a,d

Previous studies demonstrated the presence of Monochromatic Ultra-Slow Oscillations (MUSO) in human EEG. In the present study we explored the biological origin of MUSO by simultaneous recordings of EEG, Near-Infrared Spectroscopy (NIRS), arterial blood pressure, respiration and Laser Doppler flowmetry. We used a head-up tilt test in order to check whether MUSO might relate to Mayer waves in arterial blood pressure, known to be enhanced by the tilting procedure. MUSO were detected in 8 out of 10 subjects during rest and showed a striking monochromatic spectrum (0.07–0.14 Hz). The spatial topography of MUSO was complex, showing multiple foci variable across subjects. While the head-up tilt test increased the relative power of Mayer waves, it had no effect on MUSO. On the other hand, the relative spectral power of 0.1 Hz oscillations in EEG, NIRS and blood pressure signals were positively correlated across subjects in the tilted condition. Eight subjects showed a coherence between MUSO and NIRS/arterial blood pressure. Moreover, MUSO at different electrode sites demonstrated coherence not reducible to volume conduction, thus indicating that MUSO are unlikely to be generated by one source. We related our experimental findings to known biological phenomena being generated at about 0.1 Hz, i.e.: arterial blood pressure, cerebral and skin vasomotion, respiration and neuronal activity. While no definite conclusion can yet be drawn as to an exact physiological mechanism of MUSO, we suggest that these oscillations might be of a rather extraneuronal origin reflecting cerebral vasomotion. © 2014 Published by Elsevier Inc.

D

3Q1

E

2

Monochromatic Ultra-Slow (~ 0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics

E

C

T

1

R

33 37 35 34

R

36

Introduction

39 40

Neuronal oscillations are involved in many brain processes including perceptual, motor and cognitive activity. EEG and MEG are particularly relevant for studying oscillations since they provide a direct and realtime measure of neuronal activity. The majority of previous studies have focused on theta (4–7 Hz), alpha (8–12 Hz), beta (15–12 Hz) and gamma (N30 Hz) oscillations. Ever since Hans Berger's discovery of alpha oscillations they are recognized as the most pronounced rhythmic activity in the human brain, being identified by a well defined spectral peak. In contrast, for EEG activity in the very low frequency range (b0.1 Hz), which has been topic of previous studies (Vanhatalo et al., 2004, 2005). In addition, very slow oscillations in a restricted frequency

45 46 47 48 49

U

43 44

N C O

38

41 42

18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

⁎ Corresponding author at: Campus Benjamin Franklin, Charité — University Medicine Berlin, Department of Neurology and Clinical Neurophysiology, Hindenburgdamm 30, D12203 Berlin, Germany. Fax: +49 30 8445 4264. E-mail address: [email protected] (V.V. Nikulin). 1 These authors equally contributed to the study.

range were found in invasive recordings in rabbits (7–9 cycles/min, Aladjalova, 1957) and in human scalp EEG recordings (~0.1 Hz, Girton et al., 1973; Trimmel et al., 1990). Although these oscillations were recorded with electrophysiological techniques, there is no unambiguous evidence that they were due to neuronal activity. In addition only one or few electrodes were used to record these slow oscillations, thus restricting the interpretation of the obtained results. Apart from neuronal activity, very slow fluctuations also occur in cerebral hemodynamics as detected with fMRI (Balduzzi et al., 2008; Biswal et al., 1995; Fox and Raichle, 2007; Khader et al., 2008; Magnuson et al., 2010; Mantini et al., 2007), however these fluctuations are not strictly periodic, and are well described by 1/f spectrum. On the other hand, hemodynamic activity recorded with Near-Infrared Spectroscopy (NIRS) (Obrig et al., 2000; Sirotin and Das, 2009), does show pronounced oscillations at 0.1 Hz. These 0.1 Hz oscillations are also present in fluctuations of arterial blood pressure and are called Mayer waves (Julien, 2006). Interestingly, NIRS signals and Mayer waves show pronounced coherence at 0.1 Hz range (Obrig et al., 2000; Pfurtscheller et al., 2011). Another prominent source of 0.1 Hz in hemodynamic fluctuations is due to cerebral vasomotion, as shown in

http://dx.doi.org/10.1016/j.neuroimage.2014.04.008 1053-8119/© 2014 Published by Elsevier Inc.

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69

113

Protocol

114

123

After EEG and NIRS preparation, the subject was placed in a supine position on a special table equipped with a foot support. A finger cuff was applied on the left-hand index finger and calibrated for blood pressure measurements. The subject was instructed to relax with eyes open. We performed four consecutive recording sessions of 10 min each, alternating supine and tilted positions: during the supine session the subject was lying on the bed; in the tilted position the bed was at 60° with respect to the horizontal plane. Recordings from the four different devices (EEG, NIRS, blood pressure, respiration) were synchronized by a continuous train of TTL pulses.

124

Experimental setup

125 126

Electroencephalography EEG was acquired with BrainAmp MR-plus (Brain Products, Germany) amplifiers, using 58 electrodes (Fig. 1, Ag/AgCl sintered ring electrodes, EasyCap, Germany) with nose being a reference. During data acquisition EEG signals were band-pass filtered between 0.016

98 99 100 101 102 103 104 105 106 107 108 109 110 111

115 116 117 118 119 120 121 122

127 128 129

C

96 97

E

95

R

89 90

R

87 88

O

85 86

C

83 84

N

81 82

U

79 80

F

112

We studied ten healthy male subjects (age: 29.3 ± 4.4 years, mean ± standard deviation). In order to address the origin of EEG MUSO we used a combination of different techniques such as: electroencephalography (EEG), Near Infrared Spectroscopy (NIRS), non-invasive blood pressure monitoring, measurement of respiration and Laser Doppler flowmetry. Below we will provide details on each of these techniques. The experimental protocol was approved by the Ethics Committee of the Charité — University Medicine Berlin, and written informed consent was provided by all subjects. We observed MUSO at a frequency of ~0.1 Hz thus being in the frequency range of Mayer waves, which are known to reflect fluctuation in arterial blood pressure (Julien, 2006). Therefore, we aimed at investigating how MUSO would be modulated by the experimental manipulations known to affect Mayer waves. One of the standard procedures, leading to the increase of the amplitude of 0.1 Hz oscillations in blood pressure, is a head-up tilt test (Cooke et al., 1999; Furlan et al., 2000) where the subject it tilted on a table by about 60–75°. In order to investigate whether changes in MUSO, recorded with EEG, can be due to the changes in Mayer waves, we utilized a passive head-up tilt test, tilting the head up to 60° on a special table providing a foot support.

77 78

O

93 94

76

R O

Material and methods

74 75

P

92

72 73

Fig. 1. Layout of the measurement setup for EEG electrodes and NIRS probes. EEG electrodes positions from the 10–20 system are shown in black. Some of EEG cap plastic rings were used for the placement of the NIRS light sources (red circles) and detectors (green circles). Sites used for EEG and NIRS are indicated with bold circles. Blue dots represent the detection site for NIRS.

and 250 Hz and digitized at a rate of 1000 Hz. For later calculations 130 data were low-pass filtered at 80 Hz and down-sampled to 200 Hz. 131

T

91

animals (Golanov et al., 1994; Mayhew et al., 1996) and more recently in humans (Rayshubskiy et al., 2013). In the present study we combined multichannel electrophysiological and hemodynamic measures in order to investigate the biological origin of electroencephalographic Monochromatic (very narrow frequency range) Ultra-Slow Oscillations (MUSO) centered around 0.1 Hz. Given that EEG MUSO, found in the present study, peak at the same frequency (~0.1 Hz) as previously reported NIRS and blood pressure oscillations, we hypothesized that MUSO might relate to hemodynamic activity. In order to differentiate the contribution of systemic effect (blood pressure) and brain hemodynamics to the generation of MUSO, we performed recordings of EEG, arterial blood pressure and NIRS, before and after perturbation of hemodynamic activity with head-up tilt test (Cooke et al., 1999; Furlan et al., 2000). In particular, we conjectured that MUSO might be due to hemodynamic fluctuations having an electric counterpart. This idea would be in agreement with the recent findings showing that some EEG responses can be generated by cerebral hemodynamics modulating DC potential created by blood brain barrier (Vanhatalo et al., 2004; Voipio et al., 2003). If this were the case one should also observe a coupling between MUSO and oscillations in NIRS/blood pressure signals as extensively addressed in the present study.

D

70 71

V.V. Nikulin et al. / NeuroImage xxx (2014) xxx–xxx

E

2

Blood pressure Arterial blood pressure was measured non-invasively on a beat-tobeat basis by finger photoplethysmography (Finometer® midi, FMS, Amsterdam, Netherlands). The brachial arterial pressure was then reconstructed using a general inverse anti-resonance model correcting for the difference in pressure waveforms after individual forearm cuff calibration (Bogert and van Lieshout, 2005). Sampling frequency was set to 1 kHz. For later calculations data were low-pass filtered at 80 Hz and down-sampled to 200 Hz. In this paper we generally refer to brachial arterial pressure as blood pressure (BP).

132

NIRS The NIRS-System (NIRScout 8–16, NIRx Medizintechnik GmbH, Germany) consisted of 24 optical fibers (8 sources with wavelengths of 850 nm and 760 nm) and 16 detectors thus resulting in 26 measurement channels, as shown in Fig. 1. Concentration changes of hemoglobin were calculated according to the modified Lambert–Beer law on the NIRS data (differential path length factor of 5.98 (higher wavelength) and 7.15 (lower wavelength), extinction coefficients for HbO 2.5264/1.4866 (higher/lower wavelength) and HbR 1.7986/3.8437 (higher/lower wavelength), and an inter-optode-distance of 3 cm). As already used in Fazli et al. (2012), the extinction coefficients were calculated according to publicly available data (see e.g. http://www.spectros. com/fileadmin/user_upload/library/Data_Hemoglobin_Standards.pdf) provided by W.B. Gratzer, Med. Res. Council Labs, Holly Hill, London and N. Kollias, Wellman Laboratories, Harvard Medical School, Boston. The differential pathlength factors (DPF) are taken from Essenpreis et al. (1993). This procedure converts attenuation changes measured by the NIRS system into concentration changes of oxygenated (Oxy-Hb) and deoxygenated hemoglobin. In agreement with a previous study (Obrig et al., 2000), 0.1 Hz oscillations were detected primarily in Oxy-Hb and thus we restrict the following analysis to Oxy-Hb signals. Frontal,

142 143

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

133 134 135 136 137 138 139 140 141

144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162

V.V. Nikulin et al. / NeuroImage xxx (2014) xxx–xxx

173 174

Respiration The respiratory activity was monitored with a respiration belt (Brain Products, Germany) placed around the subject's torso, at the level of the stomach.

231 232

193

Analysis

194 195

For EEG data we used 1) nose-referenced potentials and 2) spatially filtered EEG potentials using Laplacian transformation (Graimann and Pfurtscheller, 2006; Hjorth, 1975) by subtracting activity from four neighboring electrodes from the activity in a given electrode. In the following text nose-referenced EEG is labeled as EEG-N and Laplaciantransformed EEG as EEG-L.

189 190

196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222

C

187 188

E

185 186

R

183 184

R

181 182

Spectral estimation For EEG, NIRS, blood pressure, Laser Doppler, and respiration we performed a power spectral density analysis. The data from the two tilt sessions and two supine sessions were combined for the analysis thus resulting in about 20 min of recordings for each condition. Power Spectral Density was estimated with 80-second segments (Hanning window) and a corresponding frequency resolution Δf = 0.0125 Hz. Then we identified the frequency bins at the base of the MUSO peak, the corresponding frequency bins are designated here as f1 and f2 and the integral (I1) of the spectral density was calculated in this range. In order to estimate how much a given peak stands above the 1/f part of the spectra we used a cubic interpolation taking into account only spectral power in f1 − 0.0375 and in f2 + 0.125 Hz frequency range and calculated an integral of the spectral power between f1 and f2 using interpolated spectrum (I2). Relative Power was calculated as I1/I2. The statistics for f1 and f2 were as follows (mean ± standard deviation): EEG-N: from 0.0725 ± 0.022 to 0.13 ± 0.021 Hz; EEG-L: from 0.08625 ± 0.019 to 0.14125 ± 0.029 Hz; NIRS: from 0.06 ± 0.015 to 0.135 ± 0.03 Hz; blood pressure: from 0.05625 ± 0.013 to 0.13375 ± 0.029 Hz. We then have chosen channels showing Relative Power N 1 and averaged corresponding spectra. On average 44 ± 12 (mean ± standard deviation) and 13 ± 4 channels had Relative Power N 1 for Nose

N C O

179 180

U

177 178

T

191 192

Laser Doppler flowmetry In a separate experimental session we recorded EEG (15 electrodes over the right hemisphere) in three subjects while simultaneously measuring cutaneous skin blood microcirculation with Laser Doppler technique (PeriFlux System 5000 by Perimed AB, Stockholm, Sweden). During the recording the subjects were sitting in a comfortable chair with their eyes being open. The Laser-Doppler instrument operated at 780 nm (red wavelength) and a short emitter detector distance (0.25 mm); hence the penetration depth was between 0.5 and 1 mm. Its probe (Probe 407 by Perimed AB, Stockholm, Sweden) was placed at three different locations during three 10-minute sessions: at the forehead close to the position of AF4 electrode, at a central position close to a position of C4 electrode and at a parietal position close to a P4 electrode. The probe was placed into a holder integrated in the EEG cap and fixed at the subject's scalp by an adhesive strip. The instrument and subjects were acclimatized for around 30 min before the actual measurement in the same room and the instrument was calibrated just before the start of the experiment.

175 176

Testing for multiple sources in the EEG In order to determine whether MUSO were generated by one or multiple sources, the following analysis was performed. For EEG-L signals, we used synchronization index (SI, Rosenblum et al., 2001) and Phase Lag Index (PLI, Stam et al., 2007). The phase of MUSO was extracted from the Hilbert transform of the signal after band-pass filtering centered at individual peak frequencies (band width 0.06 Hz). SI indicates the strength of the phase coupling and PLI shows whether phase synchronization between the two signals is due to phase lag being 0 (thus most likely reflecting a volume conduction effect). Significant non-zero value of PLI indicates the presence of interacting sources with a phase lag different from 0. The significance of SI and PLI was validated also with permutation tests. The following logic was applied to deduce whether there was more than one source of MUSO: Given that a few channels show a peak at MUSO range, the following situations were possible.

244 245

1. SI is 0, PLI is 0. This case indicates the presence of more than one MUSO source because more than one EEG channel shows a spectral peak around 0.1 Hz. Had it been due to volume conduction, SI would not have been 0. 2. SI is not 0, but PLI is 0. This indicates that while there is a phase synchronization, the phase difference is symmetric around zero, thus most likely indicating a volume conduction. Thus this case is not indicative for the presence of multiple MUSO generators. 3. SI is not 0 and PLI is not 0. This case like in (1) indicates the presence of more than one MUSO source. However, different from (1) these sources also demonstrate a genuine phase synchronization with time lag. 4. SI is 0 but PLI is not 0. We include this case only for the completeness of the association between SI and PLI. However, theoretically this result is unlikely to happen since non-zero PLI should also indicate phase synchronization and thus SI should be non-zero as well. It was indeed the case in our data as we observed this situation in less than 5% of the data, which is most likely due to statistical fluctuation in the calculation of SI and PLI.

260

F

171 172

O

170

225

R O

169

Coherence estimation In order to study the relation between MUSO in EEG and NIRS, blood pressure and Laser Doppler signals, we used magnitude squared coherence. The following coherence measures were calculated: EEG-N/L vs. NIRS, EEG-N/L vs. blood pressure, NIRS vs. blood pressure, and EEG-N/ L vs. Laser Doppler. For assessing the significance of coherence we used permutation tests where 30-second segments were permuted in one of the vectors thus destroying the true order of segments between the two signals while preserving their spectral content. Such permutation was performed 200 times. The significance was determined as a percentage of coherence values (from permuted sequence) exceeding the coherence value corresponding to not permuted data. In the present study the threshold for critical significance was set to 1%. The coherence estimation was calculated for the tilted condition only since the spectral peak for blood pressure was clearly detected in all subjects only in this condition. The significance of multiple coherence calculations was assessed with binomial testing (Montez et al., 2009). Using this approach we calculated a probability of obtaining m pairs with significant peak coherences (at P b 0.01) out n of tested pairs.

P

167 168

Reference and Laplacian recordings, respectively. Relative Power of 223 the average spectrum was then used for across-subjects statistics. 224

D

165 166

motor and parietal areas of the head were covered with optodes as shown in Fig. 1. The sampling frequency was 6.25 Hz (for further calculations involving EEG we up-sampled the data to 200 Hz). NIRS probes and EEG electrodes were integrated in a standard EEG cap (extended 10–20 system) with inter-optode distances 2–3 cm. The optical probes were constructed such that they fit into the ring of standard electrodes. This allowed us to place NIRS fibers at the same locations as EEG electrodes.

E

163 164

3

226 227 228 229 230

233 234 235 236 237 238 239 240 241 242 243

246 247 248 249 250 251 252 253 254 255 256 257 258 259 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278

As it follows from the description, the presence of cases described in 279 (1) and (3) is associated with multiple sources of MUSO. 280 Results

281

We detected EEG-MUSO in 8 of 10 subjects. Fig. 2A shows a time 282 course of 10 min from unfiltered Laplace-transformed EEG recordings 283

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

V.V. Nikulin et al. / NeuroImage xxx (2014) xxx–xxx

P

R O

O

F

4

295 296 297 298 299 300

T

C

E

R

293 294

R

291 292

O

289 290

C

288

were no apparent similarities in the topographies of 0.1 Hz oscillations in EEG and Oxy-Hb recordings. In addition we have not observed similarities in the spatial patterns of EEG-MUSO across subjects. In some subjects electrodes showing strong MUSO were located over the frontal areas, while in other subjects strongest MUSO were detected over the parietal areas.

301 302 303 304 305 306

Effect of head-up tilt on 0.1 Hz oscillations in EEG, NIRS and blood pressure 307 We compared changes in the Relative Power of 0.1 Hz oscillations in EEG, NIRS, and blood pressure between supine and tilted conditions. Fig. 3B shows that only for blood pressure 0.1 Hz oscillations were significantly increased in amplitude in the tilted condition (Wilcoxon signed rank test) while no amplitude changes between the two conditions were detected in EEG or NIRS. In addition, we also compared peak frequencies of oscillations between the conditions. We observed significant increase in peak frequency for Oxy-Hb signal in

N

286 287

(EEG-L, see Methods). Fig. 2B provides an enlarged segment of MUSO in EEG-N and in EEG-L. In addition, the three lower panels of Fig. 2B show 0.1 Hz oscillations obtained in NIRS (Oxy-Hb), blood pressure and respiration. All traces are shown only with low-pass filtering at 80 Hz (see Methods section Electroencephalography) in order to demonstrate that the oscillatory nature of the signal is not due to the narrow bandpass filtering. In general we observed that MUSO are best detectable after the Laplacian transformation compared to nose-referenced recordings as visible from comparing traces in corresponding panels of Fig. 2B. MUSO showed waxing-and-waning behavior but in some cases the oscillations could last without a considerable attenuation for a hundred of seconds (Fig. 2A). Fig. 2C shows corresponding spectra for the activities presented in Fig. 2B. Note the well-shaped and pronounced spectral peaks at about 0.1 Hz for EEG-N and EEG-L. Fig. 2D shows a topographic map of the spectral power density of oscillations in EEG-N, EEG-L, and Oxy-Hb recordings, respectively. A characteristic feature of MUSO in EEG was their relatively sparse power topography. In general, there

U

284 285

E

D

Fig. 2. Example of 0.1 Hz oscillations in different measurement systems in one subject. A. 10 minute unfiltered EEG-L recording from one of the channels. B. Time courses for EEG-N (electrode C5), EEG-L (electrode C5), NIRS (source 8/detector 15, close to PO1), Blood Pressure (BP) and respiration. C. Power spectral density (PSD) in logarithmic scale in the low frequency range (0–0.25 Hz). Horizontal green lines indicate the peak frequency range. D Scalp power spatial distribution of the PSD for the frequency peak at about 0.1 Hz for EEG-N, EEG-L and NIRS. The crosses indicate positions of sensors with the corresponding activity, depicted in panel A.

Fig. 3. Effect of the tilt test. A. Peak frequencies. B. Relative Power in the tilted and supine condition. Non-parametric Wilcoxon test, * P b 0.05. Error bars indicate standard error of the mean.

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

308 309 310 311 312 313 314 315

V.V. Nikulin et al. / NeuroImage xxx (2014) xxx–xxx

326 327 328 329 330 331 332 333 334 335

For the tilted condition, where the 0.1 Hz oscillations were detected in blood pressure, EEG and NIRS recordings, we performed a correlation of Relative Power between signals from different measuring modalities and across subjects. Fig. 4A shows that Relative Power in EEG-N and blood pressure were positively correlated (Spearman correlation, R = 0.71, P b 0.05). We also observed a significant positive correlation between Relative Power in EEG-N and Oxy-Hb (Fig. 4B, Spearman correlation, R = 0.82, P b 0.05). In addition we found a significant correlation between Relative Power of 0.1 Hz oscillations in Oxy-Hb and blood pressure (Fig. 4C, Spearman correlation, R = 0.7, P b 0.05). No correlation was detected between Relative Power of 0.1 Hz oscillations in EEG-L and NIRS/blood pressure recordings. Coherence analysis between 0.1 Hz oscillations in EEG, NIRS and blood pressure

338

For this analysis we calculated all pairwise coherences between channels belonging to different measurement techniques. For instance for 26 NIRS channels and 58 EEG-N channels in a given subjects we would obtain 26 × 58 pairs of signals which can be used for the calculation of coherences. Fig. 5 shows an example of coherence spectra for EEG-N vs. NIRS and EEG-N vs. blood pressure. The figure shows a pronounced statistically significant peak at about 0.1 Hz for both EEG-N vs. NIRS and EEG-N vs. blood pressure coherences. In 8/10 subjects we detected significant values of coherence at about 0.1 Hz between EEG-N/EEG-L and blood pressure signals and between EEG-N/EEG-L and Oxy-Hb signals. Fig. 6A shows the average percentage of EEG channels (nose referenced or Laplace-transformed) showing at least one significant (at α = 0.01) coherence value at about 0.1 Hz with Oxy-Hb or blood pressure signals. Although for EEG-N/EEG-L vs NIRS the number m of significant pairs (out of a total number of tested pairs n) varied across subjects, the binomial testing for multiple comparisons (see Methods) showed that even for the subject with the smallest number of significant pairs, the probability of multiple significant

348 349 350 351 352 353 354 355

C

E

347

R

345 346

R

343 344

N C O

341 342

367

358 359 360 361 362 363 364 365 366

In order to clarify whether MUSO originate from more than one source, we used a combination of Synchronization Index (SI) and Phase Lag Index (PLI). Table 1 shows the percentage of interactions between EEG-L channels at MUSO frequency range for each of the combinations of these two measures. In total 39% of the pairwise interactions had SI and PLI not being significantly different from zero (Case 1, see Methods). This indicates that there were different sources of MUSO in EEG not showing any interactions but indicating that MUSO is not due to one source, since spectral peaks were detected in multiple channels. In 32% of all pair-wise interactions both, SI and PLI were significantly different from zero (Case 3). This in turn indicated that there might have been multiple MUSO sources also showing some phase locking which was not due to the volume conduction. In 25% of the data SI was significantly different from 0, but PLI was not. This indicates that the interactions corresponding to these pairs of EEG channels are most likely due to the volume conduction.

368 369

3.5. Laser Doppler flowmetry and EEG

384

In three subjects we performed additional recordings measuring skin blood flow with Laser Doppler technique in combination with EEG. On average we observed a peak in the Laser Doppler signal at ~0.096 Hz. Fig. 7 shows an example of spectra of EEG and Laser Doppler signals for one subject. For all three subjects we calculated coherence between EEG-N/EEG-L channels and Laser Doppler signals and used permutation tests to check for the significance. In two subjects we observed a significant coherence between the Laser Doppler signals and EEG-L. In both subjects the coherence peak was lower than the corresponding peaks in Laser Doppler. In one of these subjects (Fig. 7), we detected a significant interaction between EEG-N and Laser Doppler.

385 386

U

339 340

T

336 337

3.4. EEG-MUSO are generated by more than one source

F

324 325

3.2. Relationship between Relative Power of 0.1 Hz oscillations in EEG and NIRS/blood pressure

O

323

R O

322

356 357

P

320 321

coherences was at most P b 5e− 09. For coherences between EEG-N/ EEG-L and BP binomial testing was at most P b 0.001. While Fig. 6A shows how many EEG channels showed coherence, another interesting question is how many EEG channels were coherent with a given NIRS or blood pressure signal. Fig. 6B shows the average percentage of EEG-N/EEG-L channels demonstrating significant coherence with the activity in a given NIRS channel or with blood pressure signal. These percentages were quite high, usually N 40% and considerably higher than what would be expected for the signals with random coherences at α = 0.01, e.g. only about 1% of the cases would show the significance.

D

318 319

NIRS recordings (Fig. 3A, P b 0.05) in supine compared to the tilted condition, while no changes in peak frequency were detected in EEG and blood pressure. Head-up tilt test also did not change the frequency of respiration, which was on average ~0.25 Hz, being significantly larger (by a factor of 2) than the frequency of MUSO in EEG (P b 0.001).

E

316 317

5

Fig. 4. Correlation between Relative Power values of blood pressure, NIRS and EEG across different subjects (Spearman's correlation coefficient, P b 0.05). A. BP vs. EEG, B. NIRS vs. EEG, C. BP vs. NIRS.

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

370 371 372 373 374 375 376 377 378 379 380 381 382 383

387 388 389 390 391 392 393 394 395

V.V. Nikulin et al. / NeuroImage xxx (2014) xxx–xxx

R O

O

F

6

401

Discussion

402 403

This study demonstrated the presence of monochromatic EEG oscillations at about 0.1 Hz and their coherence with blood pressure and NIRS signals. Importantly, spectral/coherency peaks of MUSO clearly stand out above the 1/f part of the spectrum, thus representing a truly oscillatory phenomenon in EEG. MUSO was detected in 8 out 10 subjects. We hypothesize that the absence of MUSO in two subjects could be due to insufficient signal-to-noise ratio. The main idea of the present study was to describe MUSO in EEG recordings and to show their possible relation to the hemodynamic responses. While more experiments are required in order to address further aspects of MUSO, here we used six different recordings techniques (EEG/NIRS/blood pressure/Laser Doppler/Respiration) as well as head-up tilt test to characterize the nature of MUSO. Although a

414

T

C

E

R

R

412 413

O

410 411

C

408 409

N

406 407

416 417 418 419

Two previous studies presented an evidence for the presence of MUSO in human scalp EEG (Girton et al., 1973; Trimmel et al., 1990). There was no spectral analysis performed in those studies, but the presented time traces showed oscillatory EEG activity at about 0.1 Hz. The amplitude and frequency of MUSO in those studies were comparable to the parameters obtained in our study. In addition, another study in rabbits, using invasive cortical recordings, suggested that these oscillations were generated directly by brain tissue (Aladjalova, 1957). No attempt was made in any of these studies to test the hypothesis that MUSO might be linked to hemodynamic activity. Interestingly, the authors (Girton et al., 1973) also commented on the fact that none of the stimuli (tactile or visual) as well as none of other types of mental activity (mental arithmetic calculation, meditation, memory recollections) had any effect upon eliciting MUSO.

421 422

Neural generators

435

When one uses EEG, the main assumption is that it is a real time measure of electrical neuronal activity in contrast to purely hemodynamic responses measured for instance with fMRI/NIRS. Current wisdom holds that apart from easily identifiable muscle artifacts and eyes movements, EEG primarily reflects a neuronal activation. Therefore, if someone records

436 437

Table 1 Evidence for the presence of more than one source of MUSO.SI (Synchronization Index), and PLI (Phase Lag Index) for EEG-L channels with Relative Power N1. Case 1 and case 3 indicate the presence of more than one MUSO source. Case 2 relates to volume conduction effects.

Fig. 6. Descriptive statistics of coherence between EEG and hemodynamic measurements. A. Average percentage of EEG-N and EEG-L channels across all subjects with at least one significant interaction with NIRS (dark gray) and BP (light gray). B. Average percentage of EEG-N/EEG-L channels showing significant interactions with the activity in a given NIRS channel (dark gray) or with blood pressure signal (light gray).

415

420

Previous reports on MUSO

U

404 405

complete understanding of MUSO origin is still pending, we found an unequivocal evidence for their relation to NIRS and blood pressure signals thus providing a basis for further experiments in this area. Using our findings and the evidence reported in other studies, we consider different interpretations for the generation of MUSO.

E

399 400

Finally, in the third subject with a peak at about 0.1 Hz in Laser Doppler and EEG, no significant peak in coherence was observed. Thus only in one out of three subjects and only in EEG-N we observed a relatively small peak of coherence corresponding to a spectral peak in Laser Doppler and MUSO.

397 398

D

396

P

Fig. 5. Coherences between different measuring modalities. A. Coherence between blood pressure and EEG-N signals in a representative subject. B. Coherence between EEG-N and 26 NIRS signals (Oxy-Hb). Red dashed line: the highest confidence interval (P b 0.01) for the shuffled data. C and D correspond to A and B where only three largest coherences are shown. In panel C the blue, green and red traces correspond to the interaction of BP with EEG-N electrodes Fz, FC3, C2, respectively. In panel D the coherence is between EEG-N and NIRS such as: electrode PO2 vs source 7/detector 13 (close to Pz, blue), electrode O1 vs source 2/detector 2 (close to F2, green), electrode O1 vs source 3/detector 7 (close to FC1, red).

Significant SI Significant PLI Occurrence %

423 424 425 426 427 428 429 430 431 432 433 434

438 439 440 t1:1 t1:2 t1:3 t1:4 t1:5

Case 1 No synchr no phase lag

Case 2 Synchr no phase lag

Case 3 Synchr phase lag

t1:6

− − 39

+ − 25

+ + 32

t1:7 t1:8 t1:9

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

7

R O

O

F

V.V. Nikulin et al. / NeuroImage xxx (2014) xxx–xxx

441 442

P

Fig. 7. Example of the interaction of EEG with Laser Doppler perfusion signal. The first three rows show Power Spectral Density of EEG-N, EEG-L and Laser Doppler signal. The last two rows show the coherence: EEG-N vs. Laser Doppler and EEG-L vs. Laser Doppler. The three columns indicate the position of the Laser Doppler probe in three different sessions (Frontal: AF4, Central: C4; Parietal: P4). The EEG-N and EEG-L channels' subset are color-coded with respect to the spatial location on the scalp, as described in the legend. A significant coherence at the peak frequency corresponding to the peak frequency of Laser Doppler and EEG is marked by an arrow in the parietal session. For coherence plots magenta line indicates confidence interval.

478 479

513

461

Respiration

462 463

473

One of the most straightforward explanations for the presence of MUSO might have been due to respiration. In this scenario slight movement of the body during breathing might have lead to the movement of electrodes and correspondingly to the changes in EEG potentials. However, the frequency of respiration was twice larger than the frequency for MUSO, e.g. 0.25 Hz vs. 0.12 Hz. The mean respiratory rate, obtained in the present study (~15 breaths/min or 0.25 Hz), was similar to the mean normative rate (~ 16 breaths/min) reported in Tobin et al. (1983). Moreover, the presence of multiple sources and the complexity of MUSO topographies with multiple foci rule out a possibility for their generation through a single common source such as a respiratory activity.

particularly in large ones, the presence of local vasomotion indicates that not all changes in vessel tone are due to systemic blood flow changes (Nilsson and Aalkjaer, 2003). In agreement with this statement it was shown that the low-frequency oscillations in the microvasculature in response to a vasoconstrictive challenge were not a consequence of oscillations in heart rate or systemic blood pressure (Silverman and Stout, 2002). Interestingly, Laser Doppler flowmetry also demonstrated that the highest perfusion values were seen on the face and forehead with much lower perfusion being detected at the extremities and at the trunk (Stücker et al., 2001). In line with this, it was found that oscillations at about 0.13 Hz were strongest on the forehead compared to forearm skin (Bari et al., 2005), as also reported previously (Lossius and Eriksen, 1995). This indicates that vasomotion is particularly pronounced in scalp skin from where EEG signals are usually recorded. Given that there is a trans-epidermal potential (De Talhouet and Webster, 1996) mechanical fluctuations due to skin vasomotion might give rise to electric potentials oscillating at the frequency of vasomotion, potentially recordable with EEG. The coherence between EEG and NIRS at 0.1 Hz is an indication that MUSO might have a contribution from skin vasomotion, since in addition to cerebral blood flow NIRS also detects skin hemodynamics (Kirilina et al., 2012). Also the presence of multiple sources of EEG MUSO as well as their complex topographies would be in agreement with spatially diverse activation of vasomotion oscillations (Nilsson and Aalkjaer, 2003). Yet, in more specific flow-metric measurements we detected only some cases of significant coherence between MUSO and Laser-Doppler signals where the frequency of interaction was not necessarily at the peak frequency of MUSO and there was only one marginally significant instance of coherence corresponding to the frequency range of EEG MUSO. Notwithstanding, we cannot completely exclude the possibility that MUSO might have some contribution from skin potentials, and as almost all EEG studies are performed with surface electrodes, the understanding of electrical phenomena in such condition is imperative for a proper interpretation of the obtained results.

474

4.4. Skin vasomotion

Mayer waves in head-up tilt test

475 476

Low-frequency oscillations at about 0.1 Hz are known to occur in skin blood flow due to the vasomotion (Bari et al., 2005; Silverman and Stout, 2002). Although Mayer waves can also be observed in vessels,

Oscillations in arterial blood pressure at about 0.1 Hz (Mayer waves) 514 are prominent in humans and animals and reflect baroreflex activity in 515 the cardiovascular system (see for review Julien, 2006). A head-up tilt 516

453 454 455 456 457 458 459

464 465 466 467 468 469 470 471 472

477

E

T

C

E

451 452

R

449 450

R

447 448

N C O

445 446

U

443 444

D

460

oscillations with EEG, then these are habitually interpreted in neuronal terms. Ultra-slow EEG fluctuations (at and below 0.1 Hz) have been previously reported (Aladjalova, 1957; Girton et al., 1973; Trimmerl et al., 1990; Vanhatalo et al., 2004, 2005). Although the possibility of a neuronal interpretation of the obtained data is mentioned in those studies, the unimodal nature of such recordings did not allow for an unequivocal evidence of primarily neuronal generation. Notably, some invasive intracortical recordings in cats showed that ~ 0.1 Hz hemodynamic activity can coincide with rhythmic bursting (intra-burst frequency 5 Hz), recorded with ECoG activity, which is likely to reflect neuronal processing (Golanov et al., 1994). Large amplitude of MUSO would be in agreement with the fact that slower oscillations are usually generated by the greater number of neurons (Hughes et al., 2011). In addition, the fact, that our analysis revealed a presence of multiple generators of EEG MUSO, might also be considered as an evidence for distributed neuronal sources. Although such interpretations are possible we consider below extraneuronal sources of slow oscillations which might have contributed to the generation of MUSO.

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512

575

Cerebral vasomotion and DC potentials due to blood brain barrier

576 577

Another source of 0.1 Hz oscillations in the brain is due to cerebral vasomotion as shown in animals (Golanov et al., 1994; Mayhew et al., 1996) and humans (Rayshubskiy et al., in press). Invasive Laser Doppler

542 543 544 545

551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573

578

C

540 541

E

538 539

R

536 537

R

534 535

O

532 533

C

530 531

N

528 529

U

526 527

F

574

We also assume that skin vasomotion or Mayer waves alone are unlikely to explain MUSO due to pure mechanical movement-related factors. In case of mechanical artifacts, we should have observed an increase of MUSO amplitude in the tilted condition, since this condition significantly increased blood-pressure related oscillations at 0.1 Hz (Fig. 3). Yet, the amplitude of MUSO was not different between the supine and tilted conditions. Moreover, for the control of skin vasomotion we performed simultaneous Laser-Doppler recordings and EEG and we have not observed a consistent coherence between skin blood flow and EEG signals. However, such coherence should be very high and easily replicable in different subjects if MUSO were due to mechanical artifacts caused by the skin vasomotion, the latter being fully described by Laser Doppler technique which has a penetration depth of 0.5–1 mm, thus being primarily sensitive to skin blood flow. In addition, in another study (Girton et al., 1973) the authors directly tested whether MUSO can be produced by deliberate moving of the body with frequency 5–8 cycles/min, yet no oscillatory contribution could be detected in EEG from these maneuvers. Moreover, in the same study the authors showed that MUSO cannot be due to the changes in skin potential by showing that the oscillations were absent when activity was recorded from the hand. We also believe that large scale pulsatile brain movements are unlikely to explain the generation of MUSO. This is because a presence of multiple interacting MUSO sources as well as non-coherent MUSO sources are rather incongruent with a strong common movement source.

524 525

O

549 550

523

R O

MUSO and mechanical artifacts

521 522

P

548

519 520

measurements in rats demonstrated spatial heterogeneity in 0.1 Hz oscillations even on sub-millimeter scale, which in turn indicated that cerebral vasomotion could not be reduced to a common general systemic influence or a motion artifact (Mayhew et al., 1996). Recently it was also found that sinusoidal hemodynamic oscillations at 0.1 Hz in the human brain were spatially localized and exhibited wave-like propagation thus further confirming that they were not a result of systemic blood pressure (Rayshubskiy et al., in press). As vasomotion affects blood flow, it is also likely to be reflected in NIRS signals in addition to fluctuations in blood pressure, since NIRS measures blood volume in arterial and venous vascular beds (Watzman et al., 2000). Here, we would like to emphasize a theoretically plausible consequence for EEG signal generation when one considers a combination of vasomotion and electrical potentials created at the blood brain barrier (BBB). Previous studies showed the existence of an electrical near-DC potential between cerebrospinal fluid and blood measuring up to a few millivolts (Kjallquist, 1970; Sorensen et al., 1978) due to partial permeability of plasma solutes across BBB. A model was suggested (Voipio et al., 2003) where BBB forms a voltage source that generates a volume current flowing through the distributed resistance, coupling brain potential to the surrounding extracortical tissue layers, and also through the distributed resistance of the layers between brain surface and skin surface thus giving rise to a voltage drop which can be measured on the scalp. This mechanism was hypothesized to be responsible for large (hundreds of microvolts) slow DC oscillations in EEG recorded during bilateral jugular vein compression, head-up tilt, Valsalva maneuver (Vanhatalo et al., 2003). Interestingly the authors also showed that the time course of slow EEG responses was paralleled by a similar activity of hemodynamic measures such as total hemoglobin, recorded with NIRS. The authors concluded that hemodynamic changes in the brain in combination with their modulation of BBB potential may give rise to extraneuronal generation of EEG signals. Here we extend this idea of generation of EEG through changes in BBB potentials due to cerebral hemodynamics also to spontaneous hemodynamic oscillations occurring in the brain due to local vasomotion and changes in system blood pressure. As these are primarily generated at about 0.1 Hz, as we discussed above, corresponding changes in EEG should also appear in the same frequency range and, specifically, a coherence between NIRS and EEG signals should be observed, which, indeed, was detected in abundance in the present study. Moreover, our analysis indicated that EEG-MUSO oscillations are generated by multiple sources. In many cases these sources were coupled, showing a strong phase synchronization which is not reducible to volume conduction. In other cases MUSO in different electrodes were not synchronous, thus also indicating a presence of independent generators. Such patterns of MUSO in EEG as well as often mosaic-like MUSO topographies would also be in agreement with locally specific and diverse patterns of cerebral vasomotion (Mayhew et al., 1996; Rayshubskiy et al., in press). Interestingly, the fact that we observed coherence between EEG-N/ EEG-L MUSO and NIRS/blood pressure, indicates that MUSO consist not only of spatially restricted but also of globally distributed 0.1 Hz oscillations. This is because EEG-N recordings reflect rather more global brain activation without spatial high-pass filtering of Laplacian transformation. Another evidence for a global character of MUSO is reflected in the finding that Relative Power of NIRS and systemic blood pressure was correlated only with the Relative Power of EEG-N but not with the Relative Power of EEG-L recordings. Taken together our analysis shows that EEG MUSO most likely contain both global and local features. Fig. 8 summarizes the main findings related to the interpretation of EEG MUSO. A respiration is unlikely to be involved in the generation of MUSO. On the other hand, there were no experimental findings contradicting the generation of MUSO by the cerebral vasomotion. At the same time skin vasomotion and arterial blood pressure (Mayer waves) had evidence for and against their involvement in the generation of MUSO. Ironically, although we used EEG, we do not have a direct

T

546 547

test was used in the present study as Mayer waves have also a monochromatic spectrum peaking at about 0.1 Hz and thus could be related to MUSO in EEG. In agreement with previous studies we showed that the head-up tilt test resulted in an increase in the amplitude of 0.1 Hz oscillations in blood pressure (Cooke et al., 1999; Furlan et al., 2000). This increase in the amplitude of 0.1 Hz oscillations is most likely due to high sympathetic outflow during the head-up test (Taylor and Eckberg, 1996). The amplitude of 0.1 Hz oscillations in Oxy-Hb signals was not affected by the head-up tilt, but the frequency of these oscillations was significantly smaller in the tilted than supine position. The absence of increase of Oxy-Hb peak at 0.1 Hz in NIRS might be due to cerebral autoregulation, which maintains blood flow at appropriate levels during the changes in systemic arterial blood pressure. However, one can still observe coherence between fluctuations in blood pressure and cerebral blood flow measured with transcranial Doppler ultrasound (Diehl et al., 1995; Panerai et al., 2006). This in turn might be a reason for the coherence between blood pressure and NIRS signals observed in the previous studies (Obrig et al., 2000; Pfurtscheller et al., 2011) and also in the present study. Because of the coupling of cerebral blood flow and arterial pressure, we also observed a positive correlation between Relative Power in NIRS and blood pressure signals. Interestingly, despite the changes in the amplitude of 0.1 Hz oscillations in blood pressure and frequency changes in NIRS signals, the amplitude and frequency of MUSO were not significantly affected by the head-up tilt either in EEG-N or in EEG-L recordings. In addition, the presence of multiple generators of EEG MUSO would also be at odds with a hypothesis about a single systemic factor, such as arterial blood pressure, generating MUSO. Although this finding might indicate that MUSO are not directly related to changes in blood pressure, other aspects of blood-pressure vs. EEG, namely correlations between mean amplitudes and coherences provide a complementary perspective.

D

517 518

V.V. Nikulin et al. / NeuroImage xxx (2014) xxx–xxx

E

8

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644

9

P

R O

O

F

V.V. Nikulin et al. / NeuroImage xxx (2014) xxx–xxx

evidence for the involvement of genuine neuronal electrical activity in the generation of MUSO, although we cannot exclude such possibility either.

648

Hemodynamics and EEG

649 650

670

One possible conclusion from the present data is that not only hemodynamic activity might reflect neuronal activity, but that EEG responses can also reflect hemodynamic fluctuations, while not being the cause for these hemodynamic responses — this idea being in line with a similar hypothesis suggested earlier, but for the case of special experimental maneuvers (Vanhatalo et al., 2003). This hypothesis would imply that EEG MUSO might represent an electric counterpart of the hemodynamic responses. This situation would call for a consideration in combined EEG and fMRI/NIRS studies, as it would indicate that the last two techniques would not exclusively measure neuro-vascular coupling, and thus one would rather study an association between the two manifestations of the same hemodynamic activity. We studied oscillations in 0.1 Hz as they were the common and easily identifiable binding factor across different recoding modalities. However, it is also known that hemodynamic responses as well as EEG/MEG/LFP signals have a 1/f part of spectra (Fox and Raichle, 2007; Pan et al., 2013; Thompson et al., 2014). This in turn potentially might imply that, if the idea of hemodynamic responses having an electric counterpart is feasible, other parts of the spectra of EEG signals might also have a hemodynamic contributions thus complicating the conventional interpretation of EEG/MEG/LFP signals reflecting primarily electrical neuronal activity.

671 Q3

Uncited reference

657 658 659 660 661 662 663 664 665 Q2 666 667 668 669

672

C

E

R

R

655 656

N C O

653 654

U

651 652

Linkenkaer-Hansen et al., 2001

673

Acknowledgements

674 675 Q4

V.N. and G.C. were funded by the Bernstein Center for Computational Neuroscience project B1. T.F. was funded by the Bernstein Focus for Neurotechnology, Project B1.

676

References

T

646 647

E

645

D

Fig. 8. A diagram summarizing the experimental evidence for the involvement of different physiological phenomena in the generation of EEG MUSO. Green links show that there were analysis techniques supporting the contribution of a given physiological phenomenon to the generation of EEG MUSO while red links indicates evidence to the contrary. Corresponding measures are specified in green and red squares, respectively, see the legend below.

Aladjalova, N., 1957. Infra-slow rhythmic oscillations of the steady potential of the cerebral cortex. Nature 179 (4567), 957–959. Balduzzi, D., Riedner, B.A., Tononi, G., 2008. A BOLD window into brain waves. Proc. Natl. Acad. Sci. U. S. A. 105 (41), 15641–15642. Bari, F., Toth-Szuki, V., Domoki, F., Kalman, J., 2005. Flow motion pattern differences in the forehead and forearm skin: age-dependent alterations are not specific for Alzheimer's disease. Microvasc. Res. 70 (3), 121–128. Biswal, B., Yetkin, F.Z., Haughton, V.M., Hyde, J.S., 1995. Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magn. Reson. Med. 34 (4), 537–541. Boger, L.W., van Lieshout, J.J., 2005. Non-invasive pulsatile arterial pressure and stroke volume changes from the human finger. Exp. Physiol. 90 (4), 437–446. Cooke, W.H., Hoag, J.B., Crossman, A.A., Kuusela, T.A., Tahvanainen, K.U., Eckberg, D.L., 1999. Human responses to upright tilt: a window on central autonomic integration. J. Physiol. 517 (2), 617–628. De Talhouet, H., Webster, J.G., 1996. The origin of skin-stretch-caused motion artefacts under electrodes. Physiol. Meas. 17 (2), 81–93. Diehl, R.R., Linden, D., Lücke, D., Berlit, P., 1995. Phase relationship between cerebral blood flow velocity and blood pressure. A clinical test of autoregulation. Stroke 26 (10), 1801–1804. Essenpreis, M., Cope, M., Elwell, C.E., Arridge, S.R., van der Zee, P., Delpy, D.T., 1993. Wavelength dependence of the differential pathlength factor and the log slope in timeresolved tissue spectroscopy. Adv. Exp. Med. Biol. 333, 9–20. Fazli, S., Mehnert, J., Steinbrink, J., Curio, G., Villringer, A., Muller, K.R., Blankertz, B., 2012. Enhanced performance by a hybrid NIRS–EEG brain computer interface. Neuroimage 59 (1), 519–529. Fox, M.D., Raichle, M.E., 2007. Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nat. Rev. Neurosci. 8 (9), 700–711. Furlan, R., Porta, A., Costa, F., Tank, J., Baker, L., Schiavi, R., Robertson, D., Malliani, A., Mosqueda-Garcia, R., 2000. Oscillatory patterns in sympathetic neural discharge and cardiovascular variables during orthostatic stimulus. Circulation 101 (8), 886–892. Girton, D.G., Benson, K.L., Kamiya, J., 1973. Observation of very slow potential oscillations in human scalp recordings. Electroencephalogr. Clin. Neurophysiol. 35 (6), 561–568. Golanov, E.V., Yamamoto, S., Reis, D.J., 1994. Spontaneous waves of cerebral blood flow associated with a pattern of electrocortical activity. Am. J. Physiol. 266, 204–214. Graimann, B., Pfurtscheller, G., 2006. Quantification and visualization of event-related changes in oscillatory brain activity in the time–frequency domain. Prog. Brain Res. 159, 79–97. Hjorth, B., 1975. An on-line transformation of EEG scalp potentials into orthogonal source derivations. Electroencephalogr. Clin. Neurophysiol. 39 (5), 526–530. Hughes, S.W., Lorincz, M.L., Parri, H.R., Crunelli, V., 2011. Infraslow (b0.1 Hz) oscillations in thalamic relay nuclei basic mechanisms and significance to health and disease states. Prog. Brain Res. 193, 145–162. Julien, C., 2006. The enigma of Mayer waves: facts and models. Cardiovasc. Res. 70 (1), 12–21. Khader, P., Schicke, T., Roder, B., Rosler, F., 2008. On the relationship between slow cortical potentials and BOLD signal changes in humans. Int. J. Psychophysiol. 67 (3), 252–261.

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725

P

R O

O

F

Rosenblum, M.G., Pikovsky, A.S., Kurths, J., Schäfer, C., Tass, P., 2001. Phase synchronization: from theory to data analysis. In: Moss, F., Gielen, S. (Eds.), Neuro-informatics and Neural Modeling. In: Hoff, A.J. (Ed.), Handbook of Biological Physics, vol. 4. Elsevier Science, pp. 279–321 (Chapter 9). Silverman, D.G., Stout, R.G., 2002. Distinction between atropine-sensitive control of microvascular and cardiac oscillatory activity. Microvasc. Res. 63 (2), 196–208. Sirotin, Y.B., Das, A., 2009. Anticipatory haemodynamic signals in sensory cortex not predicted by local neuronal activity. Nature 457 (7228), 475–479. Sorensen, E., Olesen, J., Rask-Madsen, J., Rask-Andersen, H., 1978. The electrical potential difference and impedance between CSF and blood in unanesthetized man. Scand. J. Clin. Lab. Invest. 38 (3), 203–207. Stam, C.J., Nolte, G., Daffertshofer, A., 2007. Phase lag index: assessment of functional connectivity from multi channel EEG and MEG with diminished bias from common sources. Hum. Brain Mapp. 28 (11), 1178–1193. Stücker, M., Steinberg, J., Memmel, U., Avermaete, A., Hoffmann, K., Altmeyer, P., 2001. Differences in the two-dimensionally measured laser Doppler flow at different skin localisations. Skin Pharmacol. Appl. Ski. Physiol. 14 (1), 44–51. Taylor, J.A., Eckberg, D.L., 1996. Fundamental relations between short-term RR interval and arterial pressure oscillations in humans. Circulation 93 (8), 1527–1532. Thompson, G.J., Pan, W.J., Magnuson, M.E., Jaeger, D., Keilholz, S.D., 2014. Quasi-periodic patterns (QPP): Large-scale dynamics in resting state fMRI that correlate with local infraslow electrical activity. Neuroimage 84, 1018–10131. Tobin, M.J., Chadha, T.S., Jenouri, G., Birch, S.J., Gazeroglu, H.B., Sackner, M.A., 1983. Breathing patterns. 1. Normal subjects. Chest 84 (2), 202–205. Trimmel, M., Mikowitsch, A., Groll-Knapp, E., Haider, M., 1990. Occurrence of infraslow potential oscillations in relation to task, ability to concentrate and intelligence. Int. J. Psychophysiol. 9, 167–170. Vanhatalo, S., Tallgren, P., Becker, C., Holmes, M.D., Miller, J.W., Kaila, K., Voipio, J., 2003. Scalp-recorded slow EEG responses generated in response to hemodynamic changes in the human brain. Clin. Neurophysiol. 114 (9), 1744–1754. Vanhatalo, S., Palva, J.M., Holmes, M.D., Miller, J.W., Voipio, J., Kaila, K., 2004. Infraslow oscillations modulate excitability and interictal epileptic activity in the human cortex during sleep. Proc. Natl. Acad. Sci. U. S. A. 101 (14), 5053–5057. Vanhatalo, S., Voipio, J., Kaila, K., 2005. Full-band EEG (FbEEG): an emerging standard in electroencephalography. Clin. Neurophysiol. 116 (1), 1–8. Voipio, J., Tallgren, P., Heinonen, E., Vanhatalo, S., Kaila, K., 2003. Millivolt-scale DC shifts in the human scalp EEG: evidence for a nonneuronal generator. J. Neurophysiol. 89 (4), 2208–2214. Watzman, H.M., Kurth, C.D., Montenegro, L.M., Rome, J., Steven, J.M., Nicolson, S.C., 2000. Arterial and venous contributions to near-infrared cerebral oximetry. Anesthesiology 93 (4), 947–953.

D

Kirilina, E., Jelzow, A., Heine, A., Niessing, M., Wabnitz, H., Brühl, R., Ittermann, B., Jacobs, A.M., Tachtsidis, I., 2012. The physiological origin of task-evoked systemic artefacts in functional near infrared spectroscopy. Neuroimage 61, 70–81. Kjallquist, A., 1970. The CSF-blood potential in sustained acid–base changes in the rat. With calculations of electrochemical potential differences for H+ and HCO3. Acta Physiol. Scand. 78 (1), 85–93. Linkenkaer-Hansen, K., Nikouline, V.V., Palvam, J.M., Ilmoniemim, R.J., 2001. Long-range temporal correlations and scaling behavior in human brain oscillations. J. Neurosci. 21 (4), 1370–1377. Lossius, K., Eriksen, M., 1995. Spontaneous flow waves detected by laser Doppler in human skin. Microvasc. Res. 50 (1), 94–104. Magnuson, M., Majeed, W., Keilholz, S.D., 2010. Functional connectivity in blood oxygenation level-dependent and cerebral blood volume-weighted resting state functional magnetic resonance imaging in the rat brain. J. Magn. Reson. Imaging 32 (3), 584–592. Mantini, D., Perrucci, M.G., Del Gratta, C., Romani, G.L., Corbetta, M., 2007. Electrophysiological signatures of resting state networks in the human brain. Proc. Natl. Acad. Sci. U. S. A. 104 (32), 13170–13175. Mayhew, J.E.W., Askew, S., Zheng, Y., Porrill, J., Westby, G.W.M., Redgrave, P., Rector, D.M., Harper, R.M., 1996. Cerebral vasomotion: a 0.1-Hz oscillation in reflected light imaging of neural activity. NeuroImage 4, 183–193. Montez, T., Poil, S.S., Jones, B.F., Manshanden, I., Verbunt, J.P., van Dijk, B.W., Brussaard, A.B., van Ooyen, A., Stam, C.J., Scheltens, P., Linkenkaer-Hansen, K., 2009. Altered temporal correlations in parietal alpha and prefrontal theta oscillations in early-stage Alzheimer disease. Proc. Natl. Acad. Sci. U. S. A. 106 (5), 1614–1619. Nilsson, H., Aalkjaer, C., 2003. Vasomotion: mechanisms and physiological importance. Mol. Interv. 3 (2), 79–89. Obrig, H., Neufang, M., Wenzel, R., Kohl, M., Steinbrink, J., Einhaupl, K., Villringer, A., 2000. Spontaneous low frequency oscillations of cerebral hemodynamics and metabolism in human adults. Neuroimage 12, 623–639. Pan, W.J., Thompson, G.J., Magnuson, M.E., Jaeger, D., Keilholz, S., 2013. Infraslow LFP correlates to resting-state fMRI BOLD signals. Neuroimage 74, 288–297. Panerai, R.B., Eames, P.J., Potter, J.F., 2006. Multiple coherence of cerebral blood flow velocity in humans. Am. J. Physiol. Heart Circ. Physiol. 291 (1), H251–H259. Pfurtscheller, G., Klobassa, D.S., Altstätter, C., Bauernfeind, G., Neuper, C., 2011. About the stability of phase shifts between slow oscillations around 0.1 Hz in cardiovascular and cerebral systems. IEEE Trans. Biomed. Eng. 58 (7), 2064–2071. Rayshubskiy, A., Wojtasiewicz, T.J., Mikell, C.B., Bouchard, M.B., Timerman, D., Youngerman, B.E., McGovern, R.A., Otten, M.L., Canoll, P.D., McKhann II, G.M., Hillman, E.M., 2014. Direct, intraoperative observation of ~ 0.1 Hz hemodynamic oscillations in awake human cortex: implications for fMRI. Neuroimage (in press).

T

726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 Q5 764 765 766

V.V. Nikulin et al. / NeuroImage xxx (2014) xxx–xxx

E

10

U

N

C

O

R

R

E

C

808

Please cite this article as: Nikulin, V.V., et al., Monochromatic Ultra-Slow (~0.1 Hz) Oscillations in the human electroencephalogram and their relation to hemodynamics, NeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.04.008

767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807