Localization of evoked neuromagnetic 600 Hz activity in the cerebral somatosensory system

Electroencephalography and clinical Neurophysiology , 91 (1994) 483-487

483

© 1994 Elsevier Science Ireland Ltd. 0013-4694/94/$07.00

EEG94605

Short communication

Localization of evoked neuromagnetic 600 Hz activity in the cerebral somatosensory system Gabriel Curio

a,*,

Bruno-Marcel Mackert a, Martin Burghoff b, R o m a n Koetitz Klaus A b r a h a m - F u c h s c and Wolfgang H~irer c

b,

a Dept. o f Neurology, Benjamin Franklin Clinic, Freie Universitiit Berlin, Hindenburgdamm 30, 12200 Berlin (FRG), b Physikalisch-Technische Bundesanstalt, Berlin (FRG), c Siemens A G Medical Engineering, Erlangen (FRG) ( A c c e p t e d f o r p u b l i c a t i o n : 12 S e p t e m b e r 1994)

Summary U p o n electrical median nerve stimulation wide-band scalp SEP recordings show a burst of high-frequency low-amplitude wavelets of uncertain origin. Digital high-pass filtering (above 400 Hz) of the primary cortical response ("N20") can separate the burst from the underlying "N20 proper" which itself is known to be generated by excitatory postsynaptic potentials (EPSPs) in area 3b. Here, neuromagnetic multichannel recordings show a close correlation between the spatial field distributions of the magnetic burst and of the magnetic " N 2 0 m " proper. It is concluded that somatosensory evoked magnetic high-frequency (600 Hz) wavelets have generators at or near the primary somatosensory cortex. Possible modes of generation comprise repetitive discharges conducted in the terminal segments of thalamocortical axons and postsynaptic contributions from neocortical neurons. Key words: Magnetoencephalography; Somatosensory cortex; Median nerve; Evoked somatosensory responses; High-frequency wavelets; Oscillatory activity

The primary cortical response to electrical median nerve stimulation ("N20"), generated by area 3b pyramids in the postcentral gyrus (Allison et al. 1991), can be detected non-invasively in m a n using electrical potential or magnetic field recordings (Wood et al. 1985; Tiihonen et al. 1989). In somatosensory evoked potential (SEP) recordings a burst of low-amplitude high-frequency wavelets can be isolated from the underlying "N20 proper" by high-pass filtering of wide-band records above about 400 Hz (Eisen et al. 1984; Emori et al. 1991). A functional independence between these two activities was found during studies across the sleep-wake cycle, showing a burst amplitude reduction already during sleep stage II along with a persisting N20 proper (Yamada et al. 1988). Furthermore, in contrast to the source localization of the N20 proper at area 3b several subcortical and cortical sources were discussed for the high-frequency burst; at present, however, a definite conclusion with respect to the electrical burst generators cannot be drawn (for review see Curio et al. 1994a). Recently, new magnetic recording systems became available featuring a white-noise level two times lower than for the previous system generation (Drung and Koch 1993; Schneider et al. 1993). This allowed for the first time to detect the weak magnetic counterparts of the high-frequency SEP bursts with a reasonable n u m b e r of averages (Curio et al. 1993, 1994a,b). The following burst features could be derived from single channel evaluations: m a x i m u m magnetic burst duration 10 msec; peak n u m b e r between 4 and 7; burst amplitude 5 - 3 0 % of the subjects' N20 amplitude. Here, relying on neuromagnetic multichannel recordings, a spatial correlation of burst

* Corresponding author. Fax: +49-30-798-4141.

SSDI 0 0 1 3 - 4 6 9 4 ( 9 4 ) 0 0 2 5 2 - 5

and N20 proper field patterns is demonstrated indicating a close cortical colocalization of at least some of their respective generators.

Methods Nine volunteer subjects (3 f, 6 m; 24-35 years; without any neurological deficit) were laid in a relaxed lateral position. The median nerve was transcutaneously electrostimulated at the right or left wrist (8.1/sec monophasic square-wave pulses of 0.1 msec width, constant current 6-12 mA: for each subject clearly above the motor threshold; n = 8000 epochs). Two planar recording systems were used: inside the Berlin magnetically shielded room a new 6-channel D C - S Q U I D m a g n e t o m e t e r system (white noise level 2.2 iT/HvrH-z- operated in an axial electronic gradiometer mode with 70 m m baseline; D r u n g and Koch 1993) was positioned coplanarly to the scalp near the anterior maximum of the N20 field distribution which was individually determined by checking pilot records (200 epochs) obtained over the hemisphere contralateral to the stimulated median nerve. For 64 msec poststimulus magnetic field data were acquired from 6 hexagonally arranged channels (7 m m diameter of S Q U I D pick-up area; 30 m m center-tocenter interchannel distance) with 0.5-1500 Hz hardware filtering and 4000 Hz A D conversion. A digital binomial high-pass filter (cut-off frequency 423 Hz; Link and T r a h m s 1992) was applied off-line. The area covered by the 6-channel gradiometer system allowed the observation of some distinct field amplitude variation, but could provide only partial covering of the field distribution required for the determination of source locations. Hence, for one subject an additional 32-channel field mapping was performed (10

G. C U R I O ET AL.

484

/4n-r

stimuli/sec, 0.2 msec pulse width, n = 4000 epochs) using an experimental prototype (Siemens, Erlangen; white noise level 2.5 f T / H T ~ ; hexagonal 32-sensor array with 26 mm center-to-center interchannel distances; 196 mm array diameter; Schneider et al. 1993) inside a commercial magnetically shielded room.

A

10

8 Results 6

With the present experimental noise level (about 4 iT peak-topeak) the averaged magnetic records contained significant highfrequency activity for 7 of the 9 subjects: a windowed FFT analysis (Fig. 1) of a typical orginal wide-band record showed two frequency ranges carrying the main signal energy, one below 400 Hz (with energy contributed by the N20m proper), the other one with a substantially weaker peak near 600 Hz (contributed by the burst activity). This allows to study the field distributions of N20m proper and burst separately after digital low- or high-pass filtering, respectively, with a cut-off frequency set at around 400 Hz. Fig. 2 displays two versions of a characteristic 6-channel record: version A documents the wide-band averaged data showing the N20m activity with a signal-to-noise ratio allowing to identify a series of low-amplitude, high-frequency inflections superimposed on the slopes of the N20m proper. Version B displays the same data after application of a digital high-pass filter (423 Hz cut-off frequency). Evidently, high-frequency activity above noise level occurred only within a burst at about 20 msec. When comparing the field amplitude across the 6 channels between these two data sets, it was found that the larger the N20m proper (which dominated the response in the wide-band records) the larger also the isolated burst. Based on this spatial correlation of the low- and high-frequency activities, a close colocalization of the generators for the N20m proper and the magnetic burst may be assumed. In a test of this hypothesis 1 of the 7 subjects showing the magnetic high-frequency burst could be studied

0 500

1000

1500

Hz Fig. 1. Windowed FFT analysis (20 msec Hanning window centered at 20 msec) of a neuromagnetic wide-band record (0.5-1500 Hz; 6 kHz sampling rate; n = 4000 epochs; computed at the Siemens AG Medical Engineering, Basic Development, Erlangen, subject 1). The data for the spectral amplitude density were obtained near the anterior field maximum of the magnetic N20 correlate ("N20m") upon electrical median nerve stimulation at the contralateral wrist. Note the major low-frequency peak A (contributed by the N20m proper) and the second (weaker) peak B at 600 Hz; this allowed to isolate the high-frequency burst by means of a digital high-pass filter with a corner frequency at about 400 Hz (cf. Fig. 2).

B: high-pass (> 423 Hz)

A: wide-band (0.5-1500 Hz)

in

T A: 200 fT / B: 40 IT

1

out

I

I

10

t

I

20

I

3!

I

0

I

40

I

I

[

5(I ms

I

(

| I0

I

~

I

20

I

30

I

I

40

1

I

I

50 ms

I

M A G N E T I C 600 Hz C E R E B R A L S O M A T O S E N S O R Y FIELDS

485

high-pass (> 450 Hz)

low-pass (< 450 Hz) 20.67 Ins

cm

!

xr" f \

~fsJ I

1K"

step: 2 IT 22.50 ms

,..~

~,~

,

~<....A ~ - ' ~ . . . . ...\ f~¢' I K" \\

@

L~ i

x

..'"

.....~S_2: •. . . . .

/ ,""[//'/"~"/'~~'~'~k~"-3

O-

/I x

-2": / /

/*

: x

-4 -

,/'I

*,

-6-

I

I

-6

-4

x

I

I

I

I

I

I

I

I

I

I

I

I

-2

0

2

4

6

-6

-4

-2

0

2

4

6

A

/ iL

[Fig. 3. Large-array mapping (32-channel m a g n e t o m e t e r prototype, 196 m m array diameter, Siemens A G Medical Engineering, Erlangen, subject 1) of somatosensory fields evoked upon contralateral median nerve stimulation. U p p e r two panels: maps obtained (left) at the peak latency of the N20m proper (averaged data digitally low-pass filtered at 450 Hz) and (right) at one of the burst peaks (high-pass filtered at 450 Hz); note map similarity and dipolar aspect. Lower 4 panels show lateral (A, C) and axial bottom (B, D) views of the spherical head model (distance between ancillary lines: 1, resp. 2 cm); dots indicate Iocalizations of equivalent current dipoles fitted at 10 data points around the peak of the N20 proper (low-pass; A, B) and dipoles fitted to the field distributions around the burst peaks (high-pass; C, D). An iterative least-squares fit algorithm was used for a spherical volume conductor model (using an estimated sphere center); hence only relative, but not absolute, 3D head coordinates may be recognized. Note that all equivalent current dipoles for both activities (N20m proper and burst peaks) were found to be closely colocalized, overlapping in a superficial volume element of about 1 cm 3.

in a second session using this time a 32-channel m a g n e t o m e t e r prototype (Siemens A G Medical Engineering, Erlangen). Indeed, strikingly similar dipolar m a p s (Fig. 3) of evoked fields at the peak

latency of the N20m proper and at one of the burst peaks were obtained for wide-band multichannel records which had been processed off-line by m e a n s of either low- or high-pass digital filtering at

Fig. 2. Characteristic (7 of 9 subjects) 6-channel record of somatosensory cerebral fields evoked upon contralateral electrical median nerve stimulation (inset: channel array; interchannel distance 30 mm; obtained at the PTB Laboratory, Berlin, subject 2). A: original wide-band records using 0.5 Hz and 1500 Hz analog filters; the signal-to-noise ratio allowed to identify a series of low-amplitude, high-frequency inflections superimposed on the N20m proper; channels were displayed with decremental amplitude of the N20m from the top to bottom trace; the inflections were identified by dashed ancillary lines. B: same data set after application of a digital high-pass filter with a cut-off frequency of 423 Hz (display gain increased by factor 5); note the evoked high-frequency burst occurring around 20 msec; dashed ancillary lines as in A; n = 8000 epochs except for trace 5 which displays superimposed data from two averages with n = 4000 epochs for demonstration of curve reproducibility. On comparison of wide-band (A) and high-pass filtered data (B) a spatial correlation of the N20m proper and burst amplitude across the sensor array is evident, suggestive of a close generator colocalization which was corroborated by an additional large-array mapping (cf. Fig. 3).

486 450 Hz. Based on such mappings a conventional least-squares dipole fit procedure was run through the data around the peak latency of the N20m proper as well as the burst peaks. Since in this pilot study only the spatial relation between the sources of the N20m proper and burst activities and not their absolute 3-dimensional coordinates in a head frame of reference was of interest, it was sufficient for the fit to assume an ideally spheric volume conductor approximating the subject's head in relation to the dewar bottom as derived from a photograph of the recording situation. Given this model assumption all equivalent current dipoles for both activities (N20m proper and burst peaks) were found to be closely colocalized, i.e., they overlapped in a volume element of about 1 cm 3 (Fig. 3).

Discussion

In this multichannel MEG study a burst of low-amplitude ( < 35 fT), high-frequency (600 Hz) magnetic wavelets was found in 7 of 9 subjects; it was superimposed on the N20m proper of the primary cortical response evoked upon conventional electrical shock stimulation of the median nerve at the wrist. A close spatial correlation of the field amplitude distributions was found between the N20m proper and the superimposed burst. Since it is known that the N20m proper is generated by EPSPs in pyramids of area 3b (Allison et al. 1991), it may be concluded that the magnetic burst (and the corresponding fraction of the electrical burst, see below) is generated at or near the 3b hand area. In SEP recordings the N20 proper may be differentiated not only from the high-frequency wavelet burst (by means of digital high-pass filtering; Eisen et al. 1984; Yamada et al. 1988; Emori et al. 1991) but also from other partially overlapping centro-parietal components such as P22 with a mainly radial and P24 with a tangential generator orientation (e.g. by means of changing the stimulus rate; Garc[a-Larr e a e t al. 1992). For MEG recordings (which are sensitive to tangential sources) the radial P22 source will not contribute substantially (but a minor magnetic correlate clue to a small tangential component remains possible; cf. Tiihonen et al. 1989). Hence, the magnetic signal in the original wide-band recordings contained activity mainly from the tangential source components (such as the sequential N20 and P24). With regard to the magnetic burst traces presented here (Fig. 2), however, both of these were removed due to the high-pass filtering (cut-off > 423 Hz). Remarkably, additional simultaneous electrical SEP and magnetic recordings in 4 subjects had shown that in two of them the magnetic burst was of different wave shape or even of shorter duration than the electrical one (Curio et al. 1994a,b). This might indicate that the magnetic gradiometer recording (coplanar to the scalp) selectively samples from a (tangential and superficial, i.e., area 3b) subset of several sources, the activities of which were overlapping in time and were all incorporated in SEP recordings; in particular, the SEP wavelets but not their magnetic counterpart could comprise major contributions from radial generators, both superficial (area 1) and deep (thalamocortical radiation fibers). Notably, the tangential magnetic burst generator (at or near the generator locus of the N20m proper in area 3b) may itself also be of heterogeneous origin with possible contributions from presynaptic action potentials (i.e., repetitive discharges conducted in the terminal segments of thalamocortical projection neurons; Katayama and Tsubokawa 1987), neocortical fast non-NMDA EPSPs (Stern et al. 1992), burst discharges of the first-order postsynaptic cortical cell population (Amassian 1953; McCormick et al. 1993; Swadlow 1993), rapid sequential activation of second- and higher-order neurons (Bode-Greuel et al. 1987) or even reverberatory activity within the local cortical microcircuitry (Langdon and Sur 1990). In any case, neuromagnetic recordings based on an advanced instrumentation (Drung and Koch 1993; Schneider et al. 1993) provide a fast and easy access to a high-frequency response mode in the

G. C U R I O ET AL. cerebral somatosensory system. This mode is either related to the thalamocortical projection to the primary somatosensory cortex or represents an intrinsically neocortical response. Its variability might be interesting to explore for diagnostic approaches to basal ganglia affections or neocortical diseases such as dementias. References

Allison, T., McCarthy, (3., Wood, C.C. and Jones, S.J. Potentials evoked in human and monkey cerebral cortex by stimulation of the median nerve: a review of scalp and intracranial recordings. Brain, 1991, 114: 2465-2503. Amassian, V.E. Evoked single cortical unit activity in the somatic sensory areas. Electroenceph. clin. Neurophysiol., 1953, 5: 415438. Bode-(3reuel, K.M., Singer, W. and Aldenhoff, J.B. A current source density analysis of field potentials evoked in slices of visual cortex. Exp. Brain Res., 1987, 69: 213-219. Curio, G., Mackert, B.-M., Burghoff, M., Koetitz, R., Drung, D., Abraham-Fuchs, K., Haerer, W. and Marx, P. Neuromagnetic detection of high-frequency (600 Hz) oscillations evoked in the human primary somatosensory cortex. Eur. J. Neurosci., 1993, Suppl. 6: 228. Curio, (3., Mackert, B.-M., Ahraham-Fuchs, K. and H~irer, W. Highfrequency activity (600 Hz) evoked in the human primary somatosensory cortex - a review of electrical and magnetic recordings. In: C. Pantev, Th. Elbert and B. Liitkenh6ner (Eds.), Oscillatory Event Related Brain Dynamics. NATO ASI Series Volume. 1994a: in press. Curio, Co., Mackert, B.-M., Burghoff, M., Koetitz, R., Drung, D. and Marx, P. Neuromagnetic detection of cerebral high-frequency activity (600 Hz) superimposed on the N20-activity from SI in man. In: L. Deecke, C. Baumgartner, (3. Stroink and S. Williamson (Eds.), Biomagnetism: Fundamental Research and Clinical Applications. 1994b: in press. Drung, D. and Koch, H. An electronic second order gradiometer for biomagnetic applications in clinical shielded rooms. IEEE Trans. Appl. Superconduct., 1993, 3: 2594-2597. Eisen, A., Roberts, K., Low, M., Hoirch, M. and Lawrence, P. Questions concerning the sequential neural generator theory of the somatosensory evoked potential raised by digital filtering. Electroenceph. clin. Neurophysiol., 1984, 59: 388-395. Emori, T., Yamada, T., Seki, Y., Yasuhara, A., Ando, K., Honda, Y., Leis, A.A. and Vachatimanont, P. Recovery functions of fast frequency potentials in the initial negative wave of median SEP. Electroenceph. clin. Neurophysiol., 1991, 78: 116-123. (3arcia-Larrea, L., Bastuji, H. and Maugui~re, F. Unmasking of cortical SEP components by changes in stimulus rate: a topographic study. Electroenceph. clin. Neurophysiol., 1992, 84: 71-83. Katayama, Y. and Tsubokawa, T. Somatosensory evoked potentials from the thalamic sensory relay nucleus (VPL) in humans: correlations with short latency somatosensory evoked potentials recorded at the scalp. Electroenceph. din. Neurophysiol., 1987, 68: 187-201. Langdon, R.B. and Sur, M. Components of field potentials evoked by white matter stimulation in isolated slices of primary visual cortex: spatial distribution and synaptic order. J. Neurophysiol., 1990, 64: 1484-1501. Link, A. and Trahms, L. Highpass filters for detecting late potentials. In: Computers in Cardiology. IEEE Computer Society Press, Washington, DC, 1992: 159-162. McCormick, D . A , Gray, C. and Wang, Z. Chattering cells: a new physiological subtype which may contribute to 20-80 Hz oscillations in cat visual cortex. Soc. Neurosci. Abst., 1993, 19: 869. Schneider, S., Abraham-Fuchs, K., Seifert, H. and Hoenig, H.E., Current trends in biomagnetic instrumentation. Appl. Superconduct., 1993, 1: 1791-1812.

MAGNETIC 600 Hz CEREBRAL SOMATOSENSORY FIELDS Stern, P., Edwards, F.A. and Sakmann, B. Fast and slow components of unitary EPSCs on stellate cells elicited by focal stimulation in slices of rat visual cortex. J. Physiol. (Lond.), 1992, 449: 247-278. Swadlow, H.A. The sequence of activation of VB thalarnic afferents, cortical efferent neurons and putative interneurons in rabbit somatosensory cortex. Soc. Neurosci. Abst., 1993, 19: 1704. Tiihonen, J., Hari, R. and H~im~il~iinen, M. Early deflections of cerebral magnetic responses to median nerve stimulation. Electroenceph, clin. Neurophysiol., 1989, 74: 290-296.

487 Wood, C.C., Cohen, D., Cuffin, B.N., Yarita, M. and Allison, T. Electrical sources in human somatosensory cortex: identification by combined magnetic and potential recordings. Science, 1985, 227: 1051-1053. Yamada, T., Kameyama, S., Fuchigami, Y., Nakazumi, Y., Dickins, Q.S. and Kimura, J. Changes of short latency somatosensory evoked potential in sleep. Electroenceph. clin. Neurophysiol., 1988, 70: 126-136.