Somatosensory evoked high-frequency oscillations recorded directly from the human cerebral cortex

Clinical Neurophysiology 111 (2000) 1916±1926

www.elsevier.com/locate/clinph

Somatosensory evoked high-frequency oscillations recorded directly from the human cerebral cortex Yoshihiro Maegaki a,c,*, Imad Najm a, Kiyohito Terada a,e, Harold H. Morris a, William E. Bingaman b, Norimasa Kohaya d, Atsumi Takenobu d, Yoko Kadonaga d, Hans O. LuÈders a a Department of Neurology, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, USA Department of Neurosurgery, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, USA c Division of Child Neurology, Institute of Neurological Sciences, Faculty of Medicine, Tottori University, 36-1 Nishi-machi, Yonago 683-8504, Japan d Division of Neurosurgery, Institute of Neurological Sciences, Faculty of Medicine, Tottori University, 36-1 Nishi-machi, Yonago 683-8504, Japan e Department of Brain Pathophysiology, Kyoto University School of Medicine, Shogoin, Sakyo-ku, Kyoto 606-01, Japan b

Accepted 21 August 2000

Abstract Objective: To elucidate the generator sources of high-frequency oscillations of somatosensory evoked potentials (SEPs), we recorded somatosensory evoked high-frequency oscillations directly from the human cerebral cortex. Subjects and methods: Seven patients, 6 with intractable partial epilepsy and one with a brain tumor, were studied. With chronically implanted subdural electrodes, we recorded SEPs to median nerve stimulation in all patients, and also recorded SEPs to lip and posterior tibial nerve stimulation in one. High-frequency oscillations were recorded using a restricted bandpass ®lter (500±2000 Hz). Results: For the median nerve oscillations, all oscillation potentials were maximum at the electrodes closest to the primary hand sensorimotor area. Most oscillations were distributed similar to or more diffusely than P20/N20. Some later oscillations after the peak of P20 or N20 were present in a very restricted cortical area similar to P25. We investigated the phase change of each oscillation potential around the central sulcus. One-third of the oscillations showed phase reversal around the central sulcus, while later oscillations elicited in a restricted cortical area did not. High-frequency oscillations to posterior tibial nerve and lip stimulation were also maximum in the sensorimotor areas. Most of the lip oscillations showed phase reversal around the central sulcus, but most of the posterior tibial nerve oscillations did not. Conclusion: High-frequency oscillations are generated near the primary sensorimotor area. There are at least two different generator mechanisms for the median nerve high-frequency oscillations. We suspect that most oscillations are derived from the terminal segments of thalamocortical radiations or from the primary sensorimotor cortex close to the generator of P20/N20, and some later oscillations from the super®cial cortex close to the generator of P25. q 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Somatosensory evoked potential; High-frequency oscillation; Subdural electrode; Thalamocortical radiation; Primary sensory cortex

1. Introduction For somatosensory evoked potentials (SEPs) following stimulation of the median nerve, the ®rst near-®eld potential (N20) is consistently recorded from the parietal region contralateral to the side of stimulation. N20 has been regarded as a cortical potential generated from the primary somatosensory cortex (area 3b) (LuÈders et al., 1983; Allison et al., 1991). Previous SEP studies have revealed several high-frequency oscillation potentials superimposed on the ascending and descending slopes of N20 (Cracco and Cracco, * Corresponding author. Tel.: 181-859-34-8038; fax: 181-859-34-8135. E-mail address: [email protected] (Y. Maegaki).

1976; Abbruzzese et al., 1978; Maccabee et al., 1983; Eisen et al., 1984; Emerson et al., 1988; Yamada et al., 1988; Emori et al., 1991; Sonoo et al., 1997). These small wavelets are better delineated by using a restricted bandpass ®lter with high-pass ®ltering of greater than 200 Hz, as a result of the attenuation of slower components such as N20 (Maccabee et al., 1983; Eisen et al., 1984; Green et al., 1986; Emori et al., 1991; Gobbele et al., 1998). The generator source of highfrequency oscillations is suspected to be different from that of N20, because they disappear or their amplitudes are signi®cantly reduced during sleep, while N20 remains almost constant (Emerson et al., 1988; Yamada et al., 1988). Although previous authors investigated possible generator sources of the controversial burst potentials, their conclu-

1388-2457/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S13 88-2457(00)0044 9-1

CLINPH 2000017

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

sions were quite inconsistent. These potentials were ®rst thought to originate from subcortical structures (Abbruzzese et al., 1978). Emori et al. (1991) investigated the recovery functions of high-frequency responses using paired stimuli and reported that they required a long interstimulus interval for full recovery. They suggested that these potentials were generated through a polysynaptic network in the cerebral cortex. Sonoo et al. (1997) suspected that all of these subcomponents were generated within area 3b, because they showed phase reversal between the frontal and parietal regions similar in the case of P20/N20. This hypothesis was supported by dipole source analysis involving magnetoencephalography. Curio et al. (1994) and Hashimoto et al. (1996) described somatosensory evoked magnetic oscillation responses similar to high-frequency oscillations, and determined their sources in the primary somatosensory cortex were very close to the source of the magnetic N20 peak. On dipole source analysis of the results of multichannel scalp SEP recording, on the other hand, Gobbele et al. (1998) proposed that early oscillation components originated from subcortical regions close to the thalamus, and that subsequent later components originated from the primary somatosensory cortex. In this study, we recorded high-frequency oscillations to median nerve stimulation using chronically implanted subdural electrodes, and examined the possible generator sources of oscillation potentials. High-frequency oscillations to lip and posterior tibial nerve stimulation were also recorded. The much better spatial resolution and less artifacts afforded by cortical surface recording allow better determination of the generator locations of these small potentials (Allison et al., 1991). To our knowledge this is the ®rst report of high-frequency oscillations recorded directly from the human cerebral cortex. 2. Materials and methods 2.1. Subjects Seven patients, 6 with medically intractable partial

1917

epilepsy and one with a brain tumor, were studied (Table 1). All patients were admitted to evaluate seizure localization and/or functional brain mapping for surgery using subdural electrodes according to the Cleveland Clinic Epilepsy Surgery Protocol (Morris, 1992). Informed consent approved by the Institutional Review Board of the Cleveland Clinic Foundation (Patients 1±6), and the Tottori University Hospital (Patient 7) was obtained from all patients prior to the study. All patients had no sensory disturbance. One patient (Patient 4) had mildly increased deep tendon re¯exes on both sides, predominantly in the lower extremities. MRI revealed no abnormality in the perirolandic regions and thalamus in all patients. The subdural electrode array covering the fronto-parietal convexity for SEP recording to median nerve and lip stimulation consisted of stainless steel electrodes in a 8 £ 8, 8 £ 5, or 8 £ 4 con®guration. Subdural strip electrodes covering the mesial frontal lobe were used for SEP recording to posterior tibial nerve stimulation in one patient (Patient 5). Fig. 1 shows the placement of subdural electrodes in each patient. Additional electrodes were placed on the temporal lobe or occipital lobe in 6 patients (Patients 2± 7). Each electrode was 3 mm in diameter and the centerto-center interelectrode distance was 1 cm. 2.2. Somatosensory evoked potentials and high-frequency oscillations The subjects lay supine on a bed and were instructed to remain relaxed during the recording. SEPs and highfrequency oscillations were recorded with an Excel (Cadwel, USA) in Patients 1±6, and with an MEB-2200 (Nihon Kohden, Japan) in Patient 7. The median and posterior tibial nerves contralateral to the side of the subdural electrodes were stimulated at the wrist and ankle, respectively. The stimulus electrodes were attached to the upper and lower lips for lip SEP recording. Because SEPs to lip stimulation are obscured by a stimulus artifact, stimulation of alternating polarity was used to cancel the

Table 1 Clinical and imaging characteristics of 8 patients for whom high-frequency oscillations were evaluated a Patient no. Age, sex Age at seizure onset (years)

Neurological ®ndings

Etiology

MRI

Side of subdural electrodes

1 2 3

37, M 29, F 21, M

37 2 5

Normal Normal Normal

Brain tumor Tubercular meningitis Bacterial meningitis

L R L

4

39, M

23

Head trauma

5 6 7

13, M 16, F 17, F

3 3 14

Spasticity (lower . upper extremities Normal Normal Normal

L. temporal lobe tumor Resection of R. anterior temporal lobe L. fronto-parietal encephalomalacia, L. mesial temporal sclerosis, L. temporal lobe cortical dysplasia Normal

a

L, left; R, right.

Unknown Normal Unknown Normal Cavernous hemangioma R. temporal encephalomalacia, R. mesial temporal sclerosis

L R R R

1918

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

Fig. 1. The locations of the subdural electrode array and functional brain mapping in each patient. SEPs and high-frequency oscillations were recorded from the electrodes enclosed by solid lines. Electrodes A7 and C4, and A4 were not used for recording because of disconnection of the wires in Patients 5 and 7, respectively. CS, central sulcus.

stimulus artifact (LuÈders et al., 1986; Ikeda et al., 1995). The stimulus was a constant-current square wave of 0.1 ms duration, delivered at a regular repetition rate of 2.31 or 2.79 Hz for Patients 1±6, and at an irregular repetition rate of 3 Hz for Patient 7. The stimulus intensity was adjusted to cause a vigorous non-painful twitch of the muscle for median and posterior tibial nerve SEP recording, and a subtle twitch of the lips for lip SEP recording. Recordings were usually performed during wakefulness (Patients 1±4 and 7) because high-frequency oscillations are attenuated during sleep (Emerson et al., 1988; Yamada et al., 1988). In two patients (Patients 5 and 6) SEPs were recorded during drowsiness or light sleep since they exhibited sleepiness. Median nerve SEPs and high-frequency oscillations were recorded from subdural electrodes covering the fronto-parietal convexity; 16 electrodes were used for the recording in Patient 1, 4 in Patient 2, 50 in Patient 3, 28 in Patient 4, 48 in Patients 5 and 6, and 39 in Patient 7 (Fig. 1). Fifty electrodes were used for lip SEP recording and 6

for posterior tibial nerve SEP recording (Patients 3 and 5, respectively). A subdural electrode distant from the recording electrodes was used as a reference (Fig. 1). A wide bandpass ®lter (30±2000 Hz) was used to con®rm the presence of cortical potentials; N20 (or N1), P20, P25 (or P2), and N30 (or N2) of the median nerve SEPs (LuÈders et al., 1983; Allison et al., 1991), N15 (or N16), P15 (or P16), N20 (or N21), and P20 of the lip SEPs (Baumgartner et al., 1992; McCarthy et al., 1993; Ikeda et al., 1995), and P40 and N40 of the posterior tibial nerve SEPs (LuÈders et al., 1986; Allison et al., 1996). In order to record the high-frequency oscillations, a restricted bandpass ®lter (500±2000 Hz) was used in all patients. We examined oscillation potentials using the analog ®ltering described in this article. In Patient 7, the oscillation potentials were also examined by digitally ®ltering raw SEPs (30±2000 Hz) from 500 to 2000 Hz, and compared with the oscillation potentials recorded on analog ®ltering. A fast Fourier transform analysis with a Hamming

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

1919

window was applied for digitally ®ltering. The analysis time was 50 or 100 ms, and the sampling rates were 24 kHz for the Excel and 125 kHz for the MEB-2200. Two trials of 200±1000 responses were averaged separately for con®rmation of the reproducibility of the waves. Positive and negative peaks were visually identi®ed on the computer display. Each component of the oscillation potentials was designated by polarity and peak latency (e.g. n20 and p20) in the SEPs recorded at the precentral electrodes closest to the primary sensorimotor areas, because all of the oscillation potentials were distributed around the primary sensorimotor areas and most of them were maximum at the precentral electrodes. Therefore, when positive and negative peaks are present at the same latency between the precentral and postcentral electrodes, the polarities are determined in the oscillation potentials recorded at the precentral electrodes in this article. We were able to con®rm some peaks at the postcentral electrodes while they were not present at the precentral electrode. In such cases, oscillation potentials were also designated by polarity and peak latency at the postcentral electrodes. The amplitude was measured to the peak of each oscillation potential from the baseline or from the preceding peak. The amplitudes of SEPs (N15, P15, N20, P20, N40, and P40) were also measured to the peak from the baseline.

Patient 1, and C2 and D2 in Patient 7. P20 and N30 were only evoked in 4 patients (Patients 3±6). Phase reversal of P15/N15 of the lip SEPs, analogous to P20/N20 of the median nerve SEPs, was observed between electrodes H5 and H6 in Patient 3. Phase reversal of P40/N40 of the posterior tibial nerve SEPs, analogous to N20/P20 of the median nerve SEPs, was observed between electrodes L and N in Patient 5. Based on the results of SEP recording, cortical stimulation, and intrasurgical observation of the brain surface along with or without 3-dimensional MRI reconstruction, we made a functional map of the brain and de®ned the central sulcus in each patient (Fig. 1).

2.3. Electrical stimulation of the cortex

3.3. Morphology of high-frequency oscillations to median nerve stimulation

Electrical stimulation of the cortex via subdural electrodes was performed according to a standard protocol (LuÈders et al., 1987) in all except for Patient 2. Positive and negative motor, sensory and language areas were systematically identi®ed. 3. Results 3.1. Functional mapping of the brain Both the positive hand motor and sensory areas were de®ned in one patient (Patient 7), and only the positive hand motor area was identi®ed in two patients (Patients 5 and 6). The positive face motor and sensory areas were de®ned in 3 patients (Patients 1, 3 and 5), and only the positive face motor area was identi®ed in two patients (Patients 4 and 6). In Patient 5, both the positive foot motor and sensory areas were identi®ed. For cortical SEPs to median nerve stimulation, P20 and N30 are evoked diffusely in cortical regions anterior to the central sulcus, and N20 in cortical regions posterior to the central sulcus, while P25 is elicited in restricted cortical regions just behind the central sulcus (LuÈders et al., 1983, 1986; Allison et al., 1991). Both P20 and N20 were recorded in 3 patients (Patients 1, 2 and 7); phase reversal of P20/N20 was seen between electrodes H1 and H2 in Patient 1, between electrodes A1 and A3 in Patient 2, and between electrodes C1 and C2 in Patient 7. P25 was seen in two patients (Patients 1 and 7); at electrodes H2 and H3 in

3.2. Analog ®ltering and digital ®ltering (Patient 7) A fast Fourier transform analysis of the wide-band records showed 3 peaks of signal energy; one below 300 Hz, one between 300 and 500 Hz, and one above 500 Hz. The energy of the high-frequency oscillation potentials ranged from 560 to 840 Hz, with the peak at 640 Hz. This frequency range was similar to that reported previously (Curio et al., 1994; Hashimoto et al., 1996). Although all oscillations were identical and their latencies were almost the same among the 3 different modes of ®ltering, they were most clearly isolated by digital ®ltering.

Fig. 2 shows typical oscillation potentials to median nerve stimulation. Although most high-frequency oscillation potentials could be seen as a notch on the slope of the SEPs with the wide bandpass ®lter, they were clearly distinguishable from the background noise and the SEPs with the restricted ®lter. A few oscillations (5 of 79 oscillations, 6.3%) were only evident with the restricted ®lter. The ®rst oscillation potential appeared just at the onset of the P20 or N20 slope in 6 of the 7 patients, and before it in the other one (Patient 4). Early oscillations before the peak of P20 were constantly seen in all patients. Subsequent later oscillations varied in each patient. The peak latencies of oscillations were similar between the two bandpass ®lters and the latency differences between the two bandpass ®lters were minimal (less than 0.6 ms; a mean value, 0.11 ms) (Fig. 2). 3.4. Distribution of high-frequency oscillations to median nerve stimulation We examined the cortical distribution of each oscillation potential in each patient (Figs. 3 and 4). The cortical distributions of high-frequency oscillations were almost similar to those of P20/N20 or P25. High-frequency oscillations were maximum at the electrodes closest to the primary sensorimotor area. There were at least two different types of oscillations in distribution. Most oscillations were distributed diffusely around the hand sensorimotor area, and their distributions

1920

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

Fig. 2. Typical examples of high-frequency oscillations to median nerve stimulation recorded with a restricted bandpass ®lter of 500±2000 Hz compared with SEPs recorded with a wide bandpass ®lter of 30±2000 Hz. The SEPs and high-frequency oscillations were recorded at the same precentral electrodes (A1 in Patient 2 and A5 in Patient 5). Note the better isolated oscillation potentials on restricted ®ltering as a result of the attenuation of slower SEP components. Most of the oscillation potentials can be identi®ed with both bandpass ®lters. p22 can only be seen on restricted bandpass ®ltering in Patient 2. The latencies of oscillations differed by 0.11 ms for the two different bandpass ®lters.

were similar to or diffuser than that of P20/N20. Some later oscillations after the peak of P20/N20 were only evoked in a restricted cortical region, and their distributions were similar to that of P25 (n21 and p22 in Patient 5 (Fig. 3), and n18, p18, and n19 in Patient 7 (Fig. 4)). We examined the polarity change of the oscillations around the central sulcus. The phase changes varied in each patient. The phase reversals were seen in 5 of the 7 patients; 27 of the 79 oscillations (34.2 %) reversed phase. The locations of phase reversal were near the primary sensorimotor area in all oscillations. The sites of phase reversal were usually located between the precentral and postcentral electrodes, similar to that of P20/N20. There were subtle differences in location of phase reversal in each patient. In Patient 1, 5 of the 13 oscillations showed phase reversal across the central sulcus, similar to P20/N20, while others did not (Fig. 5B). In Patients 2 and 4, all oscillations did not show any phase reversal in the cortical area examined. In Patient 3, two oscillations showed phase reversal across the central sulcus while phase reversal of P20/N20 was not observed in the cortical area examined. In Patient 5, 4 oscillations showed phase reversal across the central sulcus, similar to P20/N20. In the other 4 oscillations, phase reversals were present in the precentral area. In Patient 6, two oscillation potentials showed phase reversal in the precentral area (Fig. 6B). In Patient 7, 4 oscillations showed phase reversal across the central sulcus in the hand sensorimotor area, as P20/N20 did (Fig. 7B). In this area, other potentials were all in phase or did not show any phase reversal. On the other hand, 3 oscillations showed phase reversal in a cortical region inferior to the hand sensorimotor area (Fig. 7C).

3.5. High-frequency oscillations to lip stimulation (Patient 3) There were 10 oscillation peaks around the primary face sensorimotor area. The ®rst peak just appeared at the onset of the P15/N15 slope, similar to that seen for the median nerve oscillations. The amplitudes of oscillation potentials were maximum at the electrodes closest to the central sulcus (H5 and H6), where phase reversal of P15/N15 was observed (Fig. 1). Each oscillation potential showed a similar distribution to P15/N15. Seven of the 10 oscillations showed phase reversal across the central sulcus. The other 3 peaks did not show any phase change around the central sulcus. 3.6. High-frequency oscillations to posterior tibial nerve stimulation (Patient 5) Twenty-four oscillation potentials were observed to posterior tibial nerve stimulation. The ®rst positive peak appeared at the onset of the P40/N40 slope. Early oscillations, which were present on the ascending and descending slopes of P40, were recorded at electrodes L, M and N, but subsequent later components were only seen at electrode L. Some oscillations were only seen at electrodes M or N. No oscillation potentials except for one showed any phase change across the central sulcus, while phase reversal of P40/N40 was seen between electrodes L and N. 4. Discussion High-frequency oscillations were isolated from the background noise and slow cortical components on restricted

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

1921

Fig. 3. Cortical distributions of SEPs and high-frequency oscillations to median nerve stimulation in Patient 5. (A) Typical high-frequency oscillation potential recorded at electrode A5. (B) The location of recording electrodes. (C) Cortical distributions of the SEPs and high-frequency oscillations. P20/N20 are distributed diffusely around the primary hand sensorimotor area, while P25 is elicited in a restricted cortical area. Most oscillation potentials show a cortical distribution similar to that of P20/N20. Two later oscillations (n21 and p22) are elicited in a restricted cortical area similar to P25.

bandpass ®ltering. The waveform of oscillations recorded from the subdural electrodes was quite similar to that previously obtained on the scalp recording (Eisen et al., 1984; Yamada et al., 1988; Emori et al., 1991). Analog ®ltering with a narrow bandpass induces a phase shift of the potentials, especially broad potentials, while a phase shift of high-frequency oscillations does not occur, or if it does, it is only minimal (Rossini et al., 1981; Maccabee et al., 1983). This was con®rmed in this study. We examined the wave difference between the analog and digital ®ltering in one patient. Most oscillation potentials were identical between the two different modes of ®ltering. Each oscillation potential on analog ®ltering corresponded to that one in the raw SEPs. Therefore, the analog ®ltering performed in this study allowed us to examine the generator source of high-frequency oscillations. 4.1. Generator source of high-frequency oscillations to median nerve stimulation There were at least two populations of high-frequency oscillations. The ®rst population of oscillations was distributed diffusely around the primary hand sensorimotor area

and their distributions were similar to or diffuser than that of P20/N20. About one third of these oscillations showed phase reversal around the primary sensorimotor area. The second population of oscillations was only present in the restricted cortical area and their distributions were similar to P25. These oscillations were present after the peak of P20 or N20 and phase reversals were mostly not evident. These results imply that the generator mechanisms of these two populations of oscillations are different. Ozaki et al. (1998) closely examined high-frequency oscillations by means of 8-channel SEP recording from the centro-parietal scalp and reported that early oscillations superimposed on P20/N20 showed phase reversal over the centro-parietal area, but later oscillations superimposed on P22 did not. Mochizuki et al. (1999) recorded highfrequency oscillations in patients with Parkinson's disease and myoclonus epilepsy, and found two different populations of oscillation potentials. They reported that the early oscillations around N20 were enlarged in patients with Parkinson's disease, and the later oscillations around P20± N33 were enhanced in patients with myoclonus epilepsy. Gobbele et al. (1998) studied high-frequency oscillations by dipole source analysis of multichannel scalp SEP record-

1922

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

Fig. 4. Cortical distributions of SEPs and high-frequency oscillations to median nerve stimulation in Patient 7. (A) Typical high-frequency oscillation potential recorded at electrode C1. (B) The location of recording electrodes on the 3-dimensional MRI reconstruction. (C) Cortical distributions of the SEPs and highfrequency oscillations. Most oscillation potentials are distributed similar to or more diffusely than P20/N20. Three later oscillations (n18, p18 and n19) are elicited in a restricted cortical area similar to P25.

ing, and found two main oscillation bursts. They suspected the ®rst bursts seen in the time window between the brainstem P14 and the cortical N20 originated from deep axon segments of thalamocortical ®bers, and the subsequent bursts timed around N20 from the primary sensory cortex. Because the ®rst population of oscillations was distributed more diffusely around the primary hand sensorimotor area similar to P20/N20 in the present study, their generator sources would be near or slightly deeper than the generators of P20/N20. The later oscillations, which were present at the restricted cortical area similar to the P25 distribution (just behind the central sulcus) and mostly did not show phase reversal, appear to be generated from the super®cial radial sources. Therefore, we suspect that most oscillations are generated from the terminal segments of thalamocortical radiations or the deep primary sensorimotor cortex such as area 3b, and some later oscillations from super®cial sensorimotor cortex such as area 1. High-frequency responses to median nerve stimulation have been recorded from the human thalamus using depth electrodes (Katayama and Tsubokawa, 1987; Morioka et al.,

1989). Katayama and Tsubokawa examined each potential in detail and suspected that several components (peak latency, around 15 ms or more later) originated from thalamocortical radiations in and running through the thalamus. In an animal, burst potentials were also recorded from the thalamus (Arezzo et al., 1979; Rasmusson 1996). Such thalamic burst potentials themselves probably do not account for high-frequency oscillations recorded from the subdural electrodes due to their electrically closed ®eld structure. Thalamic burst potentials were reported to be signi®cantly decreased in amplitude when the recording site was moved rostrally outside the thalamus, and they were not easily identi®ed at the scalp (Katayama and Tsubokawa, 1987). Arezzo et al. (1981) recorded median nerve burst potentials within the monkey sensorimotor cortex and subcortical white matter using a high-pass ®lter above 500 Hz, and observed two different series of burst potentials; the initial bursts were seen in both intracortical and subcortical portions, and the subsequent bursts exclusively within the cortex. They suspected that the former represented the thalamocortical input to the cortex and the latter the initial ®ring

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

1923

Fig. 5. Phase changes of P20/N20 and each oscillation potential to median nerve stimulation around the central sulcus in Patient 1. (A) Phase reversal of P20/ N20 can be observed between electrodes H1 and H2, and P25 can be seen at electrodes H2 and H3. (B) The ®rst 6 oscillations (p18±n22) are all in phase around the central sulcus. Five later components (p22, n23, p24, n24, and p25) show phase reversal across the central sulcus, while the other 3 oscillations (p27, n28 and n30) do not.

of the cortical neurons in response to the thalamocortical input to the primary sensory cortex. These burst potentials were also seen on the slope of the P10 cortical potential (corresponding to human P20) from the monkey epidural electrode overlying the central sulcus, similar to our results (Peterson et al., 1995). Therefore, it is most likely that the ®rst main oscillations represent repetitive activity in the terminal segments of the thalamocortical radiations initiated by the thalamus. Some authors reported that all of the oscillation potentials were generated from tangential sources similar in the case of P20/N20 (Sonoo et al., 1997; Hashimoto et al., 1996). On scalp SEP recording, Sonoo et al. (1997) found that all of the N20 subcomponents (corresponding to high-frequency oscillations) showed phase reversal between the midfrontal (Fz) and contralateral centro-parietal (CPc) electrodes. Ozaki et al. (1998) reported that oscillation potentials superimposed on P20/N20 showed phase reversal around the centro-parietal scalp. Hashimoto et al.

(1996) reported that all of the magnetic high-frequency oscillations showed phase reversal medio-laterally similar to magnetic N20, and that their estimated dipole sources were very close to that of magnetic N20. On the other hand, Yamada et al. (1988) reported that only some of the oscillation potentials showed an out of phase relationship between the frontal (F3/F4) and contralateral parietal (2 cm behind C3/C4) electrodes. In the present study, the phase change across the central sulcus varied in each oscillation. These inconsistent data may be due to the tilted dipole of the potential as to the perpendicular axis as to the cortical surface. When a radial dipole exhibits some degree of inclination as to the perpendicular axis, only the horizontal component as to the cortical surface may be detected as a tangential dipole on magnetoencephalography, and phase reversal may be seen on scalp recording because the frontal electrode is located anteriorly and the parietal electrode is located posteriorly several centimeters distant from the generator source.

1924

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

Because there were several different locations of phase reversal in the oscillations (Fig. 7), they may have different generator mechanisms. In the monkey, Arezzo et al. (1979, 1981) reported that later burst potentials were seen within not only area 3 but also areas 1, 2 (posterior bank and crown of the postcentral gyrus), 5 (anterior bank of the postcentral or intraparietal sulcus), and 4 (anterior wall of the central sulcus). Areas 1, 2 and 4 as well as 3 receive thalamocortical projections directly from VPL, and area 5 receives projections from the primary sensory area (Parent, 1996). Therefore, high-frequency oscillations may originate from not only area 3 but also areas 1, 2, 4, and 5. Electrophysiologically, Yamada et al. (1988) examined the characteristics of oscillations in man, and reported that those in frontal and parietal scalp regions showed different amplitude changes during the waking-sleeping cycle. Emori et al. (1991) examined the recovery functions of oscillation potentials by applying paired stimuli of various interstimulus intensities, and found that the recovery function varied with each oscillation. These ®ndings support the hypothesis that oscillation potentials are derived through several different physioanatomical mechanisms.

4.2. Generator source of high-frequency oscillations to lip and posterior tibial nerve stimulation High-frequency oscillations were also identi®ed after stimulation of the lips and the posterior tibial nerve. The distributions of these oscillations were quite similar to those of the ®rst cortical responses recorded with a wide bandpass ®lter, namely P15/N15 of the lip SEPs and P40/N40 of the posterior tibial nerve SEPs. Based on the results regarding the phase changes of oscillations around the central sulcus, the majority of oscillations to lip stimulation would be generated from tangential sources, but that to posterior tibial nerve stimulation from radian sources. For the lip SEPs, the dipole direction of P15/N15 might be parallel to the cortical surface, because the peak amplitude of P15 (H5) was equal to that of N15 (H6). For the posterior tibial nerve SEPs, because the peak amplitude of P40 (K and L) was 4 times higher than that of N40 (N), we suspect that the dipole of P40/N40 has a large radial direction and a small tangential one. Therefore, the inclination of the lip sensorimotor cortex as to the brain surface might be different to that of the foot sensorimotor cortex; the former might have a perpendicular

Fig. 6. Phase changes of P20/N20 and each oscillation potential to median nerve stimulation around the central sulcus in Patient 6. (A) P20 can be seen at electrode A4 and phase reversal of P20/N20 can not be observed in a cortical area evaluated. (B) Two oscillations (n19 and p23) show phase reversal at the precentral area (between electrodes A4 and B6).

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

1925

Fig. 7. Phase changes of P20/N20 and each oscillation potential to median nerve stimulation around the central sulcus in Patient 7. (A) Phase reversal of P20/ N20 can be observed between electrodes C1 and C2, and P25 can be seen at electrode C2. (B) Four oscillations (p15, n16, p17, and p22) show phase reversal across the central sulcus (between electrodes C1 and C3). In this area, other oscillations are all in phase or do not show any phase reversal. (C) Three oscillations (p13, n14, and p18) show phase reversal between electrodes D3 and F3.

relationship to the brain surface but the latter a relatively parallel relationship. Most of the later oscillations to posterior tibial nerve stimulation after P40 were restricted to one electrode (L). These responses might originate from a very super®cial cortical region, similar to P25 in the median nerve SEPs. 4.3. Conclusion This study demonstrates that high-frequency oscillations are possibly generated from the primary sensorimotor area and the terminal segments of the thalamocortical radiations. These oscillations showed a somatotopic organization similar to the ®rst cortical potentials such as P15/N15 to lip stimulation, P20/N20 to median nerve stimulation, and P40/N40 to posterior tibial nerve stimulation. Notably, the generator of the high-frequency oscillations was not single but heterogeneous. We could de®ne at least two spatiotemporally distinct generator mechanisms for the median nerve SEPs. Some limitations of this study should be noted. (1) The number of patients studied was very small, especially for lip and posterior tibial nerve stimulation. In the cortical

stimulation study, we de®ned the primary motor and sensory areas of the hand in only one patient. (2) Although most of the oscillations were identi®ed with both the wide and narrow bandpass ®lters, and the peak latency difference between the two ®lters was minimal, some peaks were only identi®ed with the restricted bandpass ®lter. We cannot rule out the possibility that these oscillations were modi®ed by the ®ltering itself. Further examinations are necessary to con®rm the generator mechanisms and the electrophysiological characteristics of such oscillation potentials. References Abbruzzese M, Favale E, Leandri M, Ratto S. New subcortical components of the cerebral somatosensory evoked potential in man. Acta Neurol Scand 1978;57:325±332. Allison T, McCarthy G, Wood CC, Jones SJ. Potentials evoked in human and monkey cerebral cortex by stimulation of the median nerve: a review of scalp and intracortical recordings. Brain 1991;114:2465±2503. Allison T, McCarthy G, Luby M, Puce A, Spencer DD. Localization of functional regions of human mesial cortex by somatosensory evoked potential recordings and by cortical stimulation. Electroencephalogr Clin Neurophysiol 1996;100:126±140.

1926

Y. Maegaki et al. / Clinical Neurophysiology 111 (2000) 1916±1926

Arezzo J, Legatt AD, Vaughan Jr HG. Topography and intracranial sources of somatosensory evoked potentials in the monkey. I. Early components. Electroencephalogr Clin Neurophysiol 1979;46:155±172. Arezzo JC, Vaughan Jr HG, Legatt AD. Topography and intracranial sources of somatosensory evoked potentials in the monkey. II. Cortical components. Electroencephalogr Clin Neurophysiol 1981;51:1±18. Baumgartner C, Barth DS, Levesque MF, Sutherling WW. Human hand and lip sensorimotor cortex as studied on electrocorticography. Electroencephalogr Clin Neurophysiol 1992;84:115±126. Cracco RG, Cracco JB. Somatosensory evoked potential in man: far ®eld potentials. Electroencephalogr Clin Neurophysiol 1976;41:460±466. Curio G, Mackert B-M, Burghoff M, Koetitz R, Abraham-Fuchs K, Harer W. Localization of evoked neuromagnetic 600 Hz activity in the cerebral somatosensory system. Electroencephalogr Clin Neurophysiol 1994;91:483±487. Eisen A, Roberts K, Low M, Hoirch M, Lawrence P. Questions regarding the sequential neural generator theory of the somatosensory evoked potential raised by digital ®ltering. Electroencephalogr Clin Neurophysiol 1984;59:388±395. Emerson RG, Sgro JA, Pedley TA, Hauser WA. State-dependent changes in the N20 component of the median nerve somatosensory evoked potential. Neurology 1988;38:64±68. Emori T, Yamada T, Seki Y, Yasuhara A, Ando K, Honda Y, Leis AA, Vachatimanont P. Recovery functions of fast frequency potentials in the initial negative wave of median SEP. Electroencephalogr Clin Neurophysiol 1991;78:116±123. Green JB, Nelson AV, Michael D. Digital zero-phase-shift ®ltering of short-latency somatosensory evoked potentials. Electroencephalogr Clin Neurophysiol 1986;63:384±388. Gobbele R, Buchner H, Curio G. High-frequency (600 Hz) SEP activities originating in the subcortical and cortical human somatosensory system. Electroencephalogr Clin Neurophysiol 1998;108:182±189. Hashimoto I, Mashiko T, Imada T. Somatic evoked high-frequency magnetic oscillations re¯ect activity of inhibitory interneurons in the human somatosensory cortex. Electroencephalogr Clin Neurophysiol 1996;100:189±203. Ikeda A, LuÈders HO, Burgess RC, Sakamoto A, Klem GH, Morris HH, Shibasaki H. Generator locations of movement-related potentials with tongue protrusions and vocalizations: subdural recording in human. Electroencephalogr Clin Neurophysiol 1995;96:310±328. Katayama Y, Tsubokawa T. Somatosensory evoked potentials from the thalamus sensory relay nucleus (VPL) in humans: correlations with short latency somatosensory evoked potentials recorded at the scalp. Electroencephalogr Clin Neurophysiol 1987;68:187±201. LuÈders H, Lesser RP, Hahn J, Dinner DS, Klem G. Cortical somatosensory

evoked potentials in response to hand stimulation. J. Neurosurg. 1983;58:885±894. LuÈders H, Dinner DS, Lesser RP, Morris HH. Evoked potentials in cortical localization. J Clin Neurophysiol 1986;3:75±84. LuÈders H, Lesser RP, Dinner DS, Morris HH, Hahn JF, Friedman L, Skipper G, Wyllie E, Friedman D. Commentary: chronic intracranial recording and stimulation with subdural electrodes. In: Engel Jr J, editor. Surgical treatment of the epilepsies, New York: Raven, 1987. pp. 297±321. Maccabee PJ, Pinkhasov EI, Cracco RQ. Short latency somatosensory evoked potentials to median nerve stimulation: effect of low frequency ®lter. Electroencephalogr Clin Neurophysiol 1983;55:34±44. McCarthy G, Allison T, Spencer DD. Localisation of the face area of human sensorimotor cortex by intracranial recording of somatosensory evoked potentials. J Neurosurg 1993;79:874±884. Mochizuki H, Ugawa Y, Machii K, Terao Y, Hanajima R, Furubayashi T, Uesugi H, Kanazawa I. Somotosensory evoked high-frequency oscillation in Parkinson's disease and myoclonic epilepsy. Clin Neurophysiol 1999;110:185±191. Morioka T, Shima F, Kato M, Fukui M. Origin and distribution of thalamic somatosensory evoked potentials in humans. Electroencephalogr Clin Neurophysiol 1989;74:186±193. Morris HH. Protocols for surgery of epilepsy in different centers. In: LuÈders H, editor. Epilepsy surgery, New York: Raven, 1992. pp. 786±788. Ozaki I, Suzuki C, Yaegashi Y, Baba M, Matsunaga M, Hashimoto I. High frequency oscillations in early cortical somatosensory evoked potentials. Electroencephalogr Clin Neurophysiol 1998;108:536±542. Parent A. Cerebral cortex. Carpenter's human neuroanatomy, 9. Baltimore, MD: Williams & Wilkins, 1996. pp. 864±944. Peterson NN, Schroeder CE, Arezzo JC. Neural generators of early cortical somatosensory evoked potentials in the awake monkey. Electroencephalogr Clin Neurophysiol 1995;96:248±260. Rasmusson DD. Changes in the response properties of neurons in the ventroposterior lateral thalamic nucleus of the raccoon after peripheral deafferentation. J Neurophysiol 1996;75:2441±2450. Rossini PM, Cracco RQ, Cracco JB, House WJ. Short latency somatosensory evoked potential to peroneal nerve stimulation: scalp topography and the effect of different frequency ®lters. Electroencephalogr Clin Neurophysiol 1981;52:540±552. Sonoo M, Genba-Shimizu K, Mannen T, Shimizu T. Detailed analysis of the latencies of median nerve somatosensory evoked components, 2: analysis of subcomponents of the P13/P14 and N20 potentials. Electroencephalogr Clin Neurophysiol 1997;104:296±311. Yamada T, Kameyama S, Fuchigami Y, Nakazumi Y, Dickins QS, Kimura J. Changes of short latency somatosensory evoked potential in sleep. Electroencephalogr Clin Neurophysiol 1988;70:126±136.