Periodic amplitude modulation of EEG

Neuroscience Letters, 136 (1992) 213-215

213

© 1992 Elsevier Scientific Publishers Ireland Ltd. All rights reserved 0304-3940/92/$ 05.00

NSL 08439

Periodic amplitude modulation of EEG P. N o v a k a, V. Lepicovska a and C. Dostalek b ~Research Center, H6pital du Saer~-Coeur de Montreal, Department of Medicine, UniversitOde MontrOal, MontrOal, Que. (Canada) and blnstitute of Physiological Regulations, Czechoslovak Academy of Sciences, Hospital Bulovka, Prague (Czechoslovakia) (Received 20 August 1991; Revised version received 4 December 1991; Accepted 5 December 1991)

Key words: EEG; Amplitude modulation; Slow brain wave; Spectral analysis Time variation of the EEG spectral parameters was analyzed during a 10 min resting period in 40 healthy subjects. Spectral band powers over the theta and alpha bands were calculated for each non-overlapping 2.5 s long EEG segment. The time variation of the band powers was further analyzed by computing the power spectra. The results showed that both theta and alpha band powers oscillate at an average frequency 0.024 Hz and 0.057 Hz. This indicates, that the background EEG activity is modulated by periodical slow components. We hypothesize that this modulation reflects spontaneous periodic changes of cortical excitability with control at the brainstem level.

Occasional synchrony between respiration and spectral parameters of EEG band alpha together with the 0.1 Hz modulation of alpha spectral powers were observed by Pfurtscheller [6]. In the present paper the time variations of spectral parameters were quantified by spectral analysis. Forty subjects, aged from 18 to 42 years, without the neurological history participated on the study. The experimental protocol consisted in 10 min resting EEG recordings in the sitting position with eyes closed. The EEG signal was recorded on the EEG mapping system Brain Imager I/s, Neuroscience Ltd. Standard ten-twenty system with the monopolar recording was used and linked ears served as a common reference. The position was maintained using the electrode cap (ElectroCap Int.). The cylindric silver-silver chloride electrodes were filled with the EEG gel and the maximum impedance level was kept under 4 kQ. Data were checked for the presence of artifacts during recording and repeatedly after the experiment. The EEG signal was filtered with bandpass 0.3-40 Hz and digitized by sampling frequency 200 Hz with 12 bit resolution per channel. A fifth-order low-pass Butterworth filter with cutoff 30 Hz was used for further suppression of the high frequencies. The fast Fourier transform (FFT) was calculated for the 2.5 s long EEG segment multiplied by the Tukey window [l]. Resultant Correspondence." P. Novak, Research Center, H6pital du Sacr6Coeur de Montr6al, 5400w boul. Gouin, H4J1C5 Montr6al, Que., Canada.

power spectrum was divided into the delta (0.39-3.9 Hz), theta (4.3-7.8 Hz), alpha (8.2-11.7 Hz), beta I (12.1-16 Hz) and beta II (16.4-30 Hz) bands. The theta and alpha band powers were obtained as the integral of power spectrum over theta, respectively alpha band. Then the FFT of the next segment was taken until the band powers from all 240 segments were obtained. The midline electrodes from the frontal (Fz), central (Cz) and occipital (Oz) regions were chosen for further analysis. The time variation of the band powers was further evaluated by computing the power spectra as follows. The theta and alpha band powers were considered as time series with the sampling rate 2.5 s. The baseline trend was removed by the third order moving polynomial with the 75 samples window length. The frequencies above half the Nyquist rate were filtered by a finite impulse response low-pass filter. The Blackman [1] window was used to prevent spectral leakage. Then the power spectrum was computed using the FFT. The possibility of the EEG machine evoked artifacts was tested by recording the artificial signal: input of preamplifiers was open, input was shortcut by connecting all channels with the ground, the input channels were connected with the ground through 5 lc~ resistors, and finally input was connected with the signal generator producing sinusoid 10 Hz signal of an amplitude of 50 pV for each channel. The electrode impedance stability and galvanic skin response interference was tested on signals from non-scalp recordings from thigh. These signals were then evaluated in the same way as EEG signal. Time series, constructed from the successive theta and

214

alpha bands powers revealed considerable periodicity (Fig. la). The FFT spectra disclosed three main distinct peaks in both band powers and leads. The first dominant and the largest peak was detected in the 0.015-0.028 Hz range in 38 subjects (98%) and the second peak in 0.040.07 Hz range in 35 subjects (95%). The third detected peak - in the 0.09-0.15 Hz range, was identified in 22 subjects (65%) and had much smaller magnitude (Fig. l b). The average values of the frequencies of the peaks with the standard deviations are summarized in Table I. No spectra similar to EEG's were obtained from nonscalp recordings. This suggests, that neither machine, electrode nor galvanic skin resistance are responsible for the observed results. The results showed the periodic rhythmicity of band powers. Consequently, the background EEG activity is modulated by spontaneous slow periodic activity. Observed modulation of the theta and alpha activity suggests spontaneous periodic changes of cortical excitability. Rhythms with frequency below 0.08 Hz, typically at

TABLE 1 THE MEAN VALUES AND THE STANDARD DEVIATIONS (S.D.) OF THREE SPECTRAL PEAKS DETECTED IN THE POWER SPECTRA FROM THE THETA AND ALPHA TIME SERIES FROM THE ELECTRODE Fz AND Oz

1. Peak (Hz)

2. Peak (Hz)

3. Peak (Hz)

Electr, Band

Mean

S.D,

Mean

S,D.

Mean

S,D,

Fz

0.0271 0.0245 0.0247 0.0198

0.0079 0.0067 0.0070 0.0095

0.0624 0.0528 0.0598 0.0546

0,0164 0,0132 0.0143 0.0155

0.1272 0.1232 0.1293 0.1299

0.0204 0.0189 0.0186 0.0198

theta

alpha Oz

theta

alpha

0.05 Hz ('20-s rhythm') and at 0.016 Hz (' l-min rhythm') were recorded from the brainstem cardiovascular and respiratory centres responsible for blood pressure regulation and for involuntary or automatic respiratory movement in animals [2-5, 7]. Viewing these facts, control of the slow rhythms

@

b

37

T h e t a , Fz

~-. 31

650

52O 7" .~390

> 25

/

1

if2 j5

~ 260 o

o

o-13

m

130 0 0.0

T h e t Q , Cz

48

0.05 0.10 0,15 0.20 0.25

Frequency (Hz)

750 =

6OO

4O c.a

z-

45o _ III

>~2 ~24

~o300 m 150

0

<16

0

0.00 0,05 0.10 0,15 0.20 0.25

Frequency (Hz)

600 -

39 53

480 i ~

>~26

~360

~2o o

m13 7

0- 120 i

0

i

60

i

u

120

i

i

180

i

1

240

i

i

i

I

300 360 Time (s)

i

i

420

i

I

i

i

480 540

i

i

600

0

-'

0.00 0.05 0.10 0.15 0.20 0.25 Frequency (Hz)

Fig. 1. The theta and alpha time series (a) and corresponding power spectra (b) from subject No, 12. Three main peaks labelled as 1, 2 and 3 indicate the periodic fluctuations of the spectral parameters, The peaks are Iocalised at 0.02 Hz (0.01 Hz), 0.049 Hz (0,049 Hz) and 0.1 Hz (0.12 Hz) in the theta (alpha) spectra. The dotted line indicates the level of white noise.

215

(3

40

32 ff~24 0

ck 8 I

0

0

I

60

I

I

120

I

l,

180

I

I

240

I

I

I

I

300 560 Time (s)

I

I

420

I

I

I

480

I

540

I

I

600

0

0.00

,

0.05

0.10 0.15 0.20 Frequency (Hz)

i

0.25

Fig. 2. The alpha band power time series (a) and its power spectrum (b). In this subject two main peaks were identified at 0.0166 Hz and at 0.073 Hz.

occurs at the brainstem level and affects the functional behaviour of cortical activity by excitatory and inhibitory influences from both functional and anatomical connection between the brainstem, the thalamus and the cortex. We suppose that this integrative mechanism results in the slow modulation of EEG. The approach in this paper is considered to be a valuable non-invasive tool for studying the function of the deeper brain structures, more particularly, for elucidating the brainstem activity in the pathologies with presumably involvement of the central autonomic nervous system. We would like to thank Madaus Medizin Elektronik, Freiburg, FRG for financial support and Mr. F. Eickmann for technical assistance.

3

4

5

6

7 1 Harris, F.J., On the use of windows for harmonic analysis with the discrete Fourier transform, Proc. IEEE, 66 (1978) 51-83. 2 Hukuhara, T., Discharge properties of respiratory modulated brainstem reticular neurons and their relation to slow arterial pressure fluctuation in the rabbit. In K. Miyakawa, H.P. Koepchen and

C. Polosa (Eds.), Mechanisms of Blood Pressure Waves, Japan Sci. Soc. Press/Springer, Tokyo/Berlin, 1984, pp. 305-316. Koepchen, H.P., History of studies and concepts of blood pressure waves. In K. Miyakawa, H.P. Koepchen and C. Polosa (Eds.), Mechanisms of Blood Pressure Waves, Japan Sci. Soc. Press/Springer, Tokyo/Berlin, 1984, pp. 3-23. Langhorst, P., Schulz, G., Lambertz, M. and Camerer, H., Dynamic characteristic of the 'unspecific brain stem system.' In H.P. Koepchen, S.M. Hilton and A. Trzebski (Eds.), Central Interaction Between Respiratory and Cardiovascular Control Systems, Springer, Berlin, 1980, pp. 30-39. Langhorst, P., Schulz, G. and Lambertz, M., Oscillating neuronal network of the 'common brainstem system'. In K. Miyakawa, H.P. Koepchen and C. Polosa (Eds.), Mechanisms of Blood Pressure Waves, Japan Sci. Soc. Press/Springer, Tokyo/Berlin, 1984, pp. 257275. Pfurtscheller, G., Ultralangsame Schwankungen innerhalb der rhythmischen Aktivitaet im Alpha-Band und deren moegliche Ursachen, Pflugers Arch., 367 (1976) 55-66. Polosa, C., Hemorrhage and blood pressure waves. In K. Miyakawa, H.P. Koepchen and C. Polosa (Eds.), Mechanisms of Blood Pressure Waves, Japan Sci. Soc. Press/Springer, Tokyo/Berlin, 1984, pp. 147166.