N30 and the effect of explorative finger movements: a model of the contribution of the motor cortex to early somatosensory potentials

Clinical Neurophysiology 110 (1999) 1589±1600

N30 and the effect of explorative ®nger movements: a model of the contribution of the motor cortex to early somatosensory potentials Till D. Waberski a,*, Helmut Buchner a, Michael Perkuhn a, Rene Gobbele a, Michael Wagner b, Wilhelm KuÈcker c, Jiri Silny d a

Department of Neurology, RWTH Aachen, University Hospital Aachen, Pauwelsstrasse 30, D-52057Aachen, Germany b Philips Research Hamburg, Hamburg, Germany c Department of Neuroradiology, RWTH Aachen, D-52057Aachen, Germany d Helmholz Institut for Biomedical Engineering, RWTH Aachen, D-52057Aachen, Germany Accepted 7 April 1999

Abstract Objectives: The source of the N30 potential in the median nerve somatosensory evoked potentials (SEP) has been previously attributed to a pre-central origin (motor cortex or the supplementary motor area, SMA) or a post-central located generator (somatosensory cortex). This attribution was made from results of lesion studies, the behavior of the potential under pathological conditions, and dipole source localization within spherical volume conductor models. Methods: The present study applied dipole source localization and current density reconstruction within individual realistically shaped head models to median nerve SEPs obtained during explorative ®nger movements. Results: The SEPs associated with movement of the stimulated hand showed a minor reduction of the N20 amplitude and a markedly reduced amplitude for the frontal N30 and parietal P27, exhibiting a residual frontal negativity around 25 ms. The brain-stem P14 remained unchanged. Mapping of the different SEPs (movement of the non-stimulated hand minus movement of the stimulated hand) showed a bipolar ®eld pattern with a maximum around 30 ms post-stimulus. In eight out of ten normal subjects, both the N30 and the gN30 (subtraction data) sources resided within the pre-central gyrus, more medially than the post-centrally located N20. Two subjects, in contrast, showed rather post-centrally localized sources in this time range. A model of the cortical SEP sources is introduced, explaining the data with respect to previously described ®ndings of dipole localization, and from lesion studies and the alterations seen in motor diseases. Conclusions: The results provide evidence for a pre-central N30 generator, predominantly tangentially oriented, located within the motor cortex, while no sources were detected elsewhere. It is suggested that the mechanisms underlying the `gating' effect during explorative ®nger movements in the 30 ms time range predominantly arise in the motor cortex. q 1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Median nerve SEP; N30; Gating; Source localization; Current density reconstruction; Boundary element method

1. Introduction A number of studies attempted to localize the generators of the early N20 and P22 somatosensory potentials evoked by median nerve stimulation (Mauguiere et al., 1993; Wood et al., 1985; Tiihonen et al., 1989; Baumgartner et al., 1990, 1991; Allison et al., 1991; Buchner et al., 1994a,b, 1996). There is wide agreement that the current generator of the N20±P20 ®eld arises at the posterior bank of the central sulcus, corresponding to area 3b, and increasing evidence that the radial P22 ®eld is generated at the crown of the postcentral gyrus in area 1 (Allison et al., 1989, 1991; Buchner * Corresponding author. Tel.: 1 49-241-808-9827; fax: 1 49-241-8888444. E-mail address: [email protected] (T.D. Waberski)

et al., 1996). On the other hand, the generators underlying the N30 ®eld remain the subject of controversy. Two hypotheses have been pronounced, namely the dual radial source thesis, consisting of two separate frontal and parietal radial oriented sources, and the single tangential central localized source thesis (Fig. 1). The dual radial hypothesis, among others discussed by Desmedt and Tomberg (1989) on the basis of the different behavior of the N30 and P27 potentials during cognitive tasks, has been supported by ®ndings in a variety of diseases. Patients with Parkinson's disease (Rossini et al., 1989; Cheron et al., 1994), a ®nding not reproduced by others (Huttunen and TeraÈvaÈinen, 1993; Mauguiere et al., 1993; Garcia et al., 1995), and Huntington's disease (Yamada et al., 1991; ToÈpper et al., 1993) demonstrated

1388-2457/99/$ - see front matter q 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S 1388-245 7(99)00092-9

CLINPH 98717

1590

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

Fig. 1. Illustration of the two hypothesis of the median nerve SEP generators with respect to their locations and orientations. Left: dual radial hypothesis, consisting of two independent radial oriented generators in areas 4 and 1, contributing to the frontal N30 and the parietal P27. Right: single tangential hypothesis, favoring a single tangential oriented generator within the central sulcus contributing to both the frontal N30 and the parietal P27.

changes in the amplitudes of either the frontal N30 or the parietal P27. The attenuated N30 in Parkinsonian patients could be enhanced to normal values by the application of dopaminergic drugs, while the parietal potentials remained unchanged (Rossini et al., 1993, 1995; Cheron et al., 1994). In addition, Rossini et al. (1989) reported a selective attenuation of N30 in a patient with falx meningeoma, compressing the supplementary motor area (SMA). Several studies have demonstrated an attenuation of the somatosensory potentials during voluntary movement, termed the `gating' effect (Rushton et al., 1981; GruÈnewald et al., 1984). While in the case of median nerve somatosensory evoked potentials (SEPs) the subcortically generated P14 and the early parietal cortical potential N20 remained almost unchanged during voluntary movement, the frontal N30 showed a distinct attenuation during active ®nger movement of the stimulated hand, as did the parietal P27 (Cohen and Starr, 1987; Cheron and Borenstein, 1987). These results are in accordance with a common generator of N30 and P27. On the other hand, a dissociated attenuation of N30 during mental movement simulation (Cheron and Borenstein, 1992; Rossini et al., 1996) seems to contradict a single source of these components. Essential support for the single central source thesis came from combined electric and magnetic recordings of the SEPs/SEFs, which demonstrated a clear bipolar ®eld pattern around the time range of the N30 potential maps (Wood et al., 1985). This ®eld was well modeled by a single, tangentially orientated, central located dipole (Buchner et al., 1994a,b). The results of lesion studies have been contradictory, providing evidence for both, the dual radial and the single tangential hypothesis. Several authors have demonstrated a dissociated loss of frontal or parietal SEP components, supporting independent predominantly radially oriented generators for N30 and P27 (Mauguiere et al., 1983; Mauguiere and Desmedt, 1991; Furlong et al., 1993). Others have proposed a post-central origin based on lesion studies

in man and monkeys and on intracranial recordings (Allison et al., 1989, 1991a,b; Sonoo et al., 1991). Recent studies, which have applied MEG based dipole source analysis within spherically shaped head models to a large number of subjects, revealed a generator of the second peak, which seems to be the magnetic counterpart of the electrical N30, within area 4 (Kawamura et al., 1996). A better understanding of the physiology of the sensorimotor system requires further evaluation of the origin of the N30 and the generators underlying the `gating' effect. The aims of this study therefore were: (1) to evaluate the generator of the N30 component at rest and of the `gated' gN30; (2) to localize these generators in anatomical terms; and (3) to develop a model for the generators contributing to the median nerve evoked electric ®elds within the time range of the N30 potential. This model has to take into account the `gating' effect and previous results in lesion studies and in patients suffering from motor diseases. For this purpose an EEG-based study was performed; although MEG based inverse calculation techniques have the methodological advantage of avoiding the in¯uence of concentric inhomogeneities and conductivity's of the brain surrounding tissues (Cohen and Cuf®n, 1983), the EEG ®eld pattern is more sensitive to radial currents (Lopes da Silva et al., 1991). Since radially oriented source activity might contribute to the electric ®eld pattern around N30, MEG recordings might lead to a misinterpretation due to excluding these radial components. To prevent a suspected mislocation when using a single equivalent current dipole (ECD) model, the current density reconstruction (CDR) technique was applied, allowing the modeling of distributed source con®gurations. 2. Materials and methods 2.1. Subjects Median nerve SEPs and 3D-magnetic resonance (MR)tomography were obtained from 10 normal subjects, aged 23±32 years, 8 were males. All subjects gave their informed consent. 2.2. Stimulation The median nerve was stimulated at the right wrist using constant current square wave pulses of 0.2 ms duration with an intensity of twice the motor threshold of the opponens pollicis muscle. The trigger for the release of the stimulation pulses was derived from explorative movements of the hand by means of specially constructed equipment (Fig. 2). Three ring electrodes were placed on the thumb, indexand middle ®ngers, respectively. The subject was asked to explore a metal object, 2±3 cm in diameter and 65 g in weight. The stimulation pulse was applied whenever all electrodes had contact with the object (Fig. 2b). In this

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

1591

on the same day, using a 1.5 T superconducting magnet and a circular polarized head coil in a T1-weighted gradient echo pulse sequence. Technical acquisition values were: 50 ms repetition time, 5 ms echo time, 40 degrees ¯ip angle, one excitation, 30 cm ®eld of view, 256±256 image matrix (12 bits resolution). This resulted in 128 continuous slices with a thickness of 1.60 mm and a pixel size of 1.17 mm. 2.5. MR-pre-processing Fig. 2. Equipment used to couple the stimulus and explorative movements of the hand. Subjects were asked to explore a metal object with their eyes closed.

way, the SEP-stimulus was coupled to the most intensive ®nger movement, since Cohen and Starr (1987) showed a maximum `gating' effect at the maximum of EMG-activity. The stimulus rate was restricted to 2.5 Hz to avoid an attenuation of the SEPs at higher stimulus rates (Delberghe et al., 1990). Two different conditions were tested: 1. Stimulation of the right median nerve, triggered by explorative ®nger movements of the left hand. In a preliminary study, not presented in detail here, it was shown that SEPs triggered by movements of the hand contralateral to the stimulus did not differ from SEPs at a regular stimulus rate at 2.5 Hz. Hence, this condition was taken as the rest condition, since this advance ensures comparable conditions with respect to randomization, frequency and attention especially to the stimulus. 2. Stimulation of the right median nerve, triggered by explorative ®nger movement of the right hand. 2.3. Recordings SEPs were recorded from 64 electrodes with a reference at Cz, more densely spaced on the hemisphere contralateral to the stimulus (Fig. 3). The electric ®elds at the time points chosen for source reconstruction showed maxima fully covered by the dense array of electrodes above the left hemisphere (Fig. 3). So, the asymmetric electrode arrangement improves the spatial sampling and the accuracy of source localization (Wagner, 1998). After 12 bit A/D conversion, the SEP data was sampled at a rate of 3 kHz over a 40 ms pre- and 60 ms post-stimulus period using band-pass ®lter settings of 5±500 Hz (12 dB/ Oct.) on two 32-channel ampli®ers. Four replications of 1000 sweeps were recorded, resulting in 4000 averages for each condition. After SEP measurements, the position of each electrode was marked by replacing it with a small wooden disk. Disks had a 3 mm hole ®lled with fat to visualize it in the 3D-MR. 2.4. MR acquisition MR-tomography was performed after the SEP recording

The MR-slices were read into the CURRY-software package (Philips Research, Hamburg), reduced to 8 bit resolution and linearly interpolated to obtain an isotropic 3D data set. A coordinate system was de®ned with the origin (x ˆ y ˆ z ˆ 0) at the right front bottom corner of the 3DMR block. A surface reconstruction of the head was performed by radial search thresholding from outside for optimal visualization of the positions of the electrode markers. The X-Y-Zcoordinates of these markers were determined using a mouse controlled 3D-cursor. 2.6. Head modeling The surfaces of the brain, the inner and outer skull and of the skin were isolated from the isotropic 3D-MR as follows: (1) The brain surface was identi®ed by a three dimensional region growing, starting within the brain and stopping at the gray value, representing the border between the white and the gray matter. In critical regions, such as the temporal lobe and the orbita, where the automatic process was not able to ®nd the surface, `boundary markers' were inserted manually. The region growing stopped at these markers. (2) The surface of the inner skull was approximated by smoothing and dilating the brain surface. (3) The surface of the outer skull was found by three dimensional region growing starting from inside, after setting all gray values inside the inner surface of the skull to black and stopping at a gray value representing the border between the skull and the skin. (4) The surface of the skin was also found by region growing from inside after setting all gray values inside the outer limit of the skull to white and stopping at a gray value representing the skin surface (Wagner et al., 1995). The realistically shaped head model was constructed from these segmented surfaces. For the boundary-elementmethod, using a weighted isolated-problem-approach (IPA) (HaÈmaÈlainen and Sarvas, 1989; Fuchs et al., 1998), the individual surfaces were processed by triangulation of the inner skull (typically 1900 nodes, 7 mm distance between points), the outer skull (typically 1500 nodes, 9 mm distance between points), and the skin (typically 950 nodes, 14 mm distance between points). Surface elements were calculated together with the surface normals and intersections between the boundaries after thinning were avoided. A virtual 3Dre®nement of the triangulated surfaces using the vertex normal orientation and an optimized auto solid angle were

1592

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

Fig. 3. Grand average waveforms of 10 subjects under different stimulation procedures: rest, stimulation of the right median nerve triggered by movement of the left hand; gated, stimulation of the right median nerve triggered by movement of the right hand; and the subtraction data (rest 2 gated). Two frontal (F3 and Fz) and parietal (C3 and CP3) electrode positions were selected. The bold line represents the mean and the dashed line the standard deviation (reference: left ear lobe). Note the residual frontal negativity (maximum at 23.1 ms) in the gating condition, while the N30 is attenuated. The related spline interpolation maps of the time points at the potential maxima are shown on the right.

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

applied. This procedure was shown to be precise for locating super®cial sources (Fuchs et al., 1998). Thus, an additional local mesh re®nement to improve localization accuracy (Yvert et al., 1996) could be avoided (Fuchs et al., 1998). The conductivity's of the skin, brain and liquor compartments were set to 0.33 [1/V m] and of the skull to 0.0042 [1/ V m] (Gedes and Baker, 1963). 2.7. EEG-processing SEPs were read into the CURRY-software and digitally ®ltered (highpass: 10 Hz, 6 dB/Oct, forward ®lter and 250 Hz, 24 db/Oct, zero phase shift) to enhance the signal-tonoise ratio and to reduce the overlap of low frequency EEG components. Baseline correction was applied by subtracting the mean signal from 240 to 25 ms and the stimulus artifact was removed within 25 to 5 ms. Source reconstruction signals were referenced to the common average reference. 2.8. Source reconstruction Two different approaches for source reconstruction were used: 1. a single moving dipole and, 2. current density reconstruction, which solves the inverse problem by adding the constraint that the sum of the magnitudes of all sources has to be minimal, the minimum-norm condition. The regularization parameter lambda is the weighting between the minimum-norm condition and the goodness of ®t of the data. The nonlinear L1-norm was used for regularization, minimizing the absolute value (Buchner et al., 1997; Wagner et al., 1999). Lambda was iteratively adjusted until a desired ®t error of 1/SNR was achieved (x 2-criterion) (Morozov, 1984). The amount of noise in the data was estimated from a pre-trigger interval (245 to 25 ms). To prevent insensivity to minor distributed source activity, in addition a L2-norm was calculated minimizing the sums of squares in the same manner described above. This resulted in a more distributed source con®guration. The inverse procedure was spatially constrained by restricting the source space to the individual cortical surface segmented and triangulated in the manner described above, resulting in typically 16 000 support points at a spacing of 1.17 mm. Current sources were reconstructed at these discrete locations; normal to it, thereby modeling the cortical generator orientations, which are known to be perpendicular to the gray matter surface. This assumption is based on a variety of anatomical and physiological data (Lorente de No, 1938; Mitzdorf, 1985; Nunez, 1990). Its effect on the inverse procedure has been described previously (Akhtari et al., 1994; Fuchs et al., 1994; Buchner et al., 1996).

1593

3. Results 3.1. Effect of explorative ®nger movements on the SEPs De®ning the absolute amplitude of SEP components is crucial. The look on single waveforms and de®ning the amplitude by measurements of peak to peak, covers the risk of measuring remaining overlapping activity from different generators, which, as will be shown later, is obvious in the N20±P27 peak to peak amplitude. In addition, waveforms depend on the reference (Lehmann et al., 1987). The maxima of global ®eld power (GFP), which is reference independent, was therefore used for statistical evaluation, given that the GFP is unable to differentiate contributions from different sources. Comparing the amplitude of the maximal GFP, the subcortically generated P14 revealed no signi®cant changes in amplitude and latency relative to the resting condition. A minor, barely signi®cant (P ˆ 0:039, Wilcoxon signed paired rank test) reduction of 20% of the N20 was observed in the `gating' condition relative to the rest condition. However, in the time range of both the N30 and the P27 components a distinct decrease in amplitude was seen, in some cases extending even to a reversal of the ®eld polarity of N30. In the `gating' condition a negativity at the frontal electrodes in the time range 22±26 ms (N25) and a positivity in the parietal leads were preserved, which we labeled gN25/gP25 for clearness. The comparison of the GFP at the maxima of gN25 in the `gating' condition to the GFP at the same latency in the rest condition revealed no signi®cant differences in amplitude. Fig. 3 presents the grand average waveforms of the 10 subjects under different stimulation procedures and subtraction data. For clari®cation, waveforms at two frontal (F3 and Fz) and two parietal (C3 and CP3) electrodes were selected. To depict comparable results to earlier waveform based studies, signals were recalculated to the earlobe reference contralateral to the stimulus. The related maps at the time points of the potential maxima (e.g. GFP) are shown on the right. 3.2. Source localization For source analysis, time points with high activity based on the maximum of GFP were de®ned: (1) at the maximum of the N20±P20 around 20 ms in the rest condition; (2) at the maximum of the gN25/gP25 around 22±26 ms in the `gated' condition; (3) at the maximum of the N30±P30 around 30 ms in the rest condition; and (4) at the maximum of the `gated' gN30 around 30 ms in the subtraction-data (rest minus explorative ®nger movement). This was done for the following reasons: while the parietal N20 is known to be barely or unaffected by the explorative ®nger movement, both, the frontal N30 and the parietal P27, are known to be drastically depressed (Cohen and Starr, 1987; Cheron and Borenstein, 1991). Hence, subtracting the SEPs

1594

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

Fig. 4. Location of the ECDs (yellow rectangles) and of the CDR results (red dots) with respect to the individual anatomy, derived from 3D-MR. Left: visualization in a 3D-view, rotated 458 towards the viewer. Note, that the brain surface is plotted in a transparent mode. Right: horizontal slices demonstrating the relationship to the central sulcus. Note, that in the case of CDR only the location of the maximum source activity is shown for clari®cation.

recorded during explorative ®nger movement from those obtained during rest minimizes the contribution of the N20 to the later potentials, while the N30 ®eld, respectively its `gated' counterpart is enhanced. In the `gating' condition, on the other hand, gN25/gP25 is preserved and less in¯uenced by the attenuated N30. 3.3. Single moving dipole N20 was located in the region of the post-central gyrus, except in one subject who showed a pre-central location (yellow rectangle in Fig. 4, mean goodness of ®t (GoF) 90.2%). The ECD modeling of the gN25/gP25 revealed an inconstant location around the central ®ssure (GoF 93.7%). This might be due to the distribution of the remaining N20 source activity as well as to the in¯uence of the occasional reversal of activity in the time range of 30 ms post-stimulus under the `gating' condition. The ECD modeling of the N30 (GoF 93.5%) and the gN30 ®eld (GoF 94.6%) localized in front of the central ®ssure in 8 subjects. Compared to the location of N20, there was a signi®cantly (P , 0:01) more medial location for the `gated' gN30, and

a more frontal (P , 0:05) and medial (P , 0:05) location of the N30. 3.4. Current density reconstruction (L1-norm) The maximum of source activity in the N20 interval was localized post-centrally in all but one subject (red dots in Fig. 4, GoF 90.9%). The solution for the gN25/gP25 was localized behind the central ®ssure in all but one subject (GoF 92.3%). For the N30 and gN30 intervals the maximum of source activity was localized pre-centrally, near the central ®ssure, in 8 subjects. In two subjects the location was behind the central ®ssure (GoF 95.2% for N30 and 94.9% for gN30). When comparing the locations of the maximum of source activity of the N30 and gN30 component, both located signi®cantly (P , 0:05) more medially than the N20. No signi®cant difference was found for N20 with respect to gN25/gP25. Statistical comparisons were performed for the maxima of source activity of CDR, and are therefore likely to be an

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600 Table 1 Locations (pre- vs. post-central) of the ECD and CDR for N20, P25, N30 and gN30 Potential

Method

Post-central

Pre-central

N20

ECD CDR±L1 ECD CDR±L1 ECD CDR±L1 ECD CDR±L1

9 9 5 9 2 2 2 2

1 1 5 1 8 8 8 8

P25 N30 gN30

oversimpli®cation arising from the distributed source model. In Fig. 4, the locations of the ECDs (yellow rectangles) and CDR (red dots) are shown with respect to the individual anatomy. The horizontal slices demonstrates the relationship to the central ®ssure. It should be noted, that in the case of CDR the slices only show the location of the maximum of source activity for clari®cation. As CDR tends to enhance the amplitude of super®cial sources, in some cases the maximum source activity of N20 tends to be rather super®cial (JE, AH, RG). However, this error is in the radial projection and not in the critical anterior-posterior or lateral-medial direction. In some subjects (JE, AH, TM, SD) N20 seems to be composed of more than one maximum. This might be due to the complex and variable course of the central sulcus represented by the restricted source space in the applied method. This truly causes a more complex source con®guration in comparison to a non restricted source space. This effect is most prominant in L1-norm regularization, because of its ability to reduce the number of sources. Locations (pre-central vs. post-central) of the ECD and CDR results are summarized in Table 1.

1595

revealed an independent frontal negativity, N25, which was attributed to a separate tangential generator arising in the sensory cortex. The location of the `gated' gN30 provides evidence that the mechanism underlying the `gating' effect in this time range predominantly arises within the motor cortex. 4.1. Applicability of current source reconstruction Our source localization regarding the location of N30 is supported by a recent MEG based study using single ECD localization within spherical shaped head models (Kawamura et al., 1996). In this study, a signi®cantly more medial and superior location of the second peak (P35m of Huttunen, 1997), which we assume is the magnetic counterpart of the electrical N30, relative to the N20 was found. Taking into account the anatomical shape of the central sulcus, this result suggests a generator of N30 (P35m) within area 4. However, the single dipole model

3.5. Current density reconstruction (L2-norm) To avoid excluding minor source activity, L2-norm regularization, minimizing the sums of squares, was used as well. Source activity localized and was more distributed around the central sulcus for the N20, gN25/gP25, N30 and gN30 peaks. The main focus of the N20 and gN25/ pP25 was located post-central and of the N30 and gN30 pre-central. No source activity outside the central region was found. Focusing on the missing activity within the SMA Fig. 5 shows the source reconstruction (L1-norm vs. L2-norm) of a representative subject. 4. Discussion The presented results obtained from 10 normal subjects provide evidence for a generator of the N30 situated in the pre-central motor cortex. There was no hint of an involvement of the SMA. `Gating' by explorative ®nger movements

Fig. 5. Comparison of the L1-norm versus L2-norm regularization in a representative subject. Note, that in the L2-norm as well as the L1-norm solution no source activity within the SMA could be detected (currents contributing by more than 50% explained variance of the data are shown).

1596

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

can cause mislocalizations, because of the curved course of the central sulcus within the somatosensory hand area (omega shape). A mislocalization of the center of gravity of the more widespread N30-potential-®eld towards the frontal gyrus can be expected in this case (Huttunen, 1997). Therefore, although a more medial location of the N30-ECD in comparison to the N20-ECD was previously demonstrated (Kawamura et al., 1996), and reproduced in our study, this might be an artifact arising from modeling a distributed electrical ®eld such as the N30 with a single equivalent dipole model. To prevent this, current density reconstruction (CDR) allowing distributed source con®gurations was applied in our study. Thus, using regularization within a realistically shaped individual volume conductor model and a restricted source space to the surface of the brain, solutions are not only matched to the individual anatomy, but implemented in the inverse calculation. This advance considers a particular view on the results with respect to the individual anatomy. The chosen methods reveal no systematic error, since the source of N20 was correctly localized post-centrally close to the central ®ssure in 9 of 10 subjects. We do not claim that source modeling is able to explain the entire electrical activation of the cortex after median nerve stimulation. One can assume further more dif®cult source con®gurations, which probably lead to canceling of at least parts of the electric activity when measuring in the distance. In this view a source model is clearly a simpli®cation. 4.2. SEP and `gating' We employed the `gating' effect of explorative ®nger movement. This was done for the following reason: while the N20 ®eld is not markedly attenuated by `gating,' the N30 ®eld is clearly suppressed (Cheron and Borenstein, 1987). The N20 ®eld, with its distribution to the later potentials, is reduced or even abolished by subtracting the SEPs triggered by movements of the stimulated hand from those obtained in the rest condition. This results in a ®eld distribution of the N30 less in¯uenced by the earlier generators and leads to an enhanced signal-tonoise-ratio in the subtraction data. Furthermore, this advance reveals the possibility for source localization on data emphasizing a frontal negativity, gN25, distinguished from N30. 4.3. Location and orientation of the generator underlying N30 Earlier studies have interpreted the attenuation of N30 by active movements of the stimulated hand or ®ngers or even imagined movements of ®ngers, the so called `gating' effect (Cheron and Borenstein, 1991, 1992), as evidence for the origin of N30 in the SMA. These ®ndings were supported by the attenuation of N30 in patients with Parkinson's (Rossini et al., 1989; Cheron et al., 1994) and Huntington's disease

(Yamada et al., 1991; ToÈpper et al., 1993) and by the selective attenuation of N30 in a patient with falx meningeoma compressing the SMA (Rossini et al., 1989). Lesion studies, which have reported a selective loss of frontal and parietal SEP components, have also found evidence for an early activation of the pre-central motor cortex within the time range of 30 ms post-stimulus (MauguieÁre et al. 1983; MauguieÁre and Desmedt, 1991). A predominantly radial orientation of the generator has been proposed. This has led to the hypothesis of two independent radially orientated generators, one in the pre-central motor cortex, generating N30 and a second located in area 1, generating N20/P27 (dual radial hypothesis, Fig. 1a). On the other hand, combined electric and magnetic recordings of the SEPs/SEFs both demonstrated a clear bipolar ®eld pattern around the time range of the N30 potential maps (Wood et al., 1985). This ®eld could be suf®ciently modeled by a single tangentially orientated dipole (Buchner et al., 1994), which supports the hypothesis of a single tangentially orientated generator (Fig. 1b), as proposed by Allison et al. (1980) on the basis of intracranial recordings. The associated reduction in amplitude of N30 and P27 under `gating' conditions (Cohen and Starr, 1987; Cheron and Borenstein, 1987), as well as the harmonic attenuation of N30 and P27 with increasing stimulus rates (Delberghe et al., 1986) can also be interpreted as supporting a common origin of these ®elds, under the assumption of a common, predominantly tangential orientated generator. In the present study the ECD modeling of the N30 and the gN30 ®eld localized in front of the central ®ssure in 8 subjects, signi®cantly (P , 0:01) more medial for the `gated' gN30, and more frontal (P , 0:05) and medial (P , 0:05) for the location of the N30 compared to N20. The location by regularization of the N30 and gN30 reveals a maximum of source activity pre-centrally in 8 subjects, signi®cantly (P , 0:05) more medially located with respect to the N20. In consideration of the above mentioned course of the central sulcus, a more medial location is in accordance with an origin in the pre-central cortex, despite the in part missing signi®cance in the anterior-posterior direction. The early activation of portions of area 4 by median nerve stimulation and the consequent contribution of the motor cortex to the surface SEP is strongly supported by intracortical recordings in monkeys (Nicholson Peterson et al., 1995). This study provides evidence of an activation of area 4 within the anterior wall of the central sulcus. The onset of this activation was seen only 1.5 ms after the activation of area 3b and is most likely to be mediated by direct thalamocortical input, as previously suggested by Desmedt and Cheron, 1980, 1981; Arezzo et al. (1981). The ®nding of an early activation within the anterior wall causing a tangential orientated source supports our suggestion of an early pre-central tangential SEP generator. Note, that source analysis can not distinguish between a direct thalamocortical vs. cortico-cortical activation of the primary motor cortex.

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

4.4. Location and orientation of the generator underlying gN25/gP25 The persistence of a negative potential (gN25) prior to the attenuated N30 under the `gating' condition points to the existence of two independent generators contributing to the frontal electric ®eld in the time range 30 ms post-stimulus. Both are predominantly tangential in orientation, and contribute to a frontal compound negative gN25/N30 peak, and a parietal positive gP25/P27 potential, respectively. The `unmasking' of gN25/gP25 under the `gating' condition is in agreement with the results obtained under pathological conditions in patients suffering from Huntington's disease (ToÈpper et al., 1993) and in a patient with a pre-frontal localized meningeoma (see Fig. 5 in Rossini et al., 1989). Further evidence for a separate generator contributing to this frontally recordable negativity has been provided by Garcia Larrea et al. (1992), demonstrating a remaining N25/P25 complex with increasing stimulus rates, while N30 was attenuated. Source reconstruction localized the generator of N30 in the pre-central area 4, while the generator of the persisting gN25/gP25 (in the `gated' data) localized near the posterior wall of the central sulcus, close to the location of N20. Hence, the latter seems to depict neuronal activity in area 3b, while N30 represents independent activation of the motor cortex. 4.5. Model of the generators contributing to the frontally recordable negativity The thesis of one tangential oriented generator seems to contradict lesion studies, which have reported a dissociated loss of frontal and parietal SEP components in patients with restricted brain lesions (MauquieÁre et al., 1983; MauguieÁre and Desmedt, 1991). Assuming a predominantly tangential generator, the absence of N30 should lead to a marked decrease of the parietal N20±P27 peak to peak amplitude. However, this data can be interpreted on the basis of a new model, which is introduced by the following simulation using BESA (MEGIS, Munich). The model focuses on the generators, which from our results, predominantly contribute to the frontal and parietal components in the time range of 30 ms post-stimulus. A model of the early median nerve SEP, including the subcortical sources and the radial P22 is not shown, because effects of `gating' on these components do not substantially in¯uence the dominant effects of the frontal recordable potentials within this time range. Two independent sources in the time range 30 ms post-stimulus were assumed, one representing N20/P25 and the other the N30. Relative locations of the sources were selected from their mean localization in single moving dipole modeling: N30 was taken 10 mm medial and 6 mm frontal with respect to N20/P25. Sources were ®xed at these positions and dipole orientation calculated independently for each subject. Median dipole orientations were chosen as a model for simulation (Fig. 6b). The N20/P25 source

1597

was simulated as a negative de¯ection (max 21.5 nAm) and a consecutive positivity (max 1.2 nAm), N30 as a broader negative de¯ection (max 3 nAm). The resulting waveforms showed a widened N30 potential at the Fz-position, with a slight de¯ection around 27 ms, while at the CP- or P3-position the N20-potential was followed by a broader positive de¯ection around 27 ms (P27) (earlobe-reference, Fig. 6c). When simulating an attenuation of 66% of the N30 source but maintaining the N20/P25 activity (Fig. 6a), a marked reduction in the N30-potential could be seen in the frontal lead (Fz), while at 25 ms a frontal negativity was preserved (note the slightly shortened latency with respect to the de¯ection in the original data). At the parietal leads (CP3, P3), a smaller reduction of the compound P27 can be seen, as a consequence of the reduced portion of the mirror image of N30. Note that the latency of the positivity following N20 is also shortened slightly as a consequence of the more dominant contribution of P25. The effect at the parietal leads could also be seen using an Fz-reference, with an even sharper reduction of the N20±P27 peak to peak amplitude (Fig. 6d). Thus, a reduced activity of the N30 generator leads to a minor reduction in the peak to peak amplitude of N20±P27 at parietal electrodes, but to a major one of the P22±N30 at frontal locations when using extracephalic or earlobe-references, on the assumption of a predominantly tangential orientated generator of N30. This reduction is thought to be insigni®cant with respect to the contralateral unaffected side and can also be seen in the original data of the mentioned lesion studies (Figs. 1±4 in MauguieÁre and Desmedt, 1991 and Figs. 3±5 in MauguieÁre et al., 1983). 4.6. SEP under pathological conditions The modi®cation of the early median nerve SEPs under pathological conditions remains a subject of controversy. Various authors demonstrated a selective loss of the frontal N30 in patients with Parkinson's disease (Rossini et al. 1989; Cheron et al., 1994), a result not reproduced by others (MauguieÁre et al., 1993, Garcia et al., 1995). We do not want to speculate about the reasons, but a dissociated attenuation of the frontal N30 in spite of persisting parietal components seems to contradict the presence of a unique generator source in the region of the central sulcus that is responsible for both the parietal and frontal components and favors separate generators in parietal and frontal lobes simultaneously activated by the sensory input (Rossini et al., 1996). In our view the assumption of a post-centrally located tangential orientated generator, gN25/gP25, predominantly contributing to the parietal P27 underlying a precentral N30/P30 generator is consistent with this observation. In particular, it has to be pointed out that de®ning the absolute value of the parietal P27 is crucial. Assuming two tangential orientated generators contributing to a compound P27 potential the amplitude depends on several factors. (1) A variable oblique orientation in medio-lateral direction

1598

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

Fig. 6. Model (A) Upper trace: simulation of two distinguished sources in the time range 20±40 ms post-stimulus. The N20/P25 source was simulated as a negative de¯ection (max 2 1.5 nAm) and a consecutive positivity (max 2 1.2 nAm), and N30 as a broader negative de¯ection (max. 3.0 nAm). Lower trace: simulation of an attenuation of 66% of the N30 (shaded area) source, maintaining the N20/P25 activity. (B) Relative locations of the sources were assumed from their median localization in single moving dipole model of the 10 subjects: N30, 10 mm medial and 6 mm frontal with respect to N20/P25. Sources were ®xed at these positions and dipole orientations were calculated for each subject, independently. Median dipole orientations were selected for the model. Note the almost tangential orientation of both dipoles. (C) and (D) Waveforms of the normal and attenuated N30 were superimposed, with the shaded areas indicating the effect of the attenuated N30 activity in the simulation. (C) resulting waveforms using an earlobe-reference, or (D) using an Fz-reference.

should be expected when considering the inconstant course of the central sulcus. (2) The interindividual excess of the N30/P30 vs. the gN25/gP25 contribution to the compound frontal N25/N30 and parietal P27 component, resulted in a variability of the P27 latency and a sometimes w-shaped curve of the positivity following N20. This might be an essential aspect in ®nding the dissociated effect of dopaminergic drugs on the N30 (Cheron et al., 1994; Rossini et al., 1993, 1995). As our simulation showed, the effect of the attenuation of the N30 on the parietal P27 is exiguous and might be missed in recordings restricted to a limited number of electrodes. So, a densely spaced multichannel electrode array is necessary to take into account the suspected high spatial variability of this component. The extent to which the demonstrated effects of an attenuated N30 on the parietal recordable components is responsible for the impairment of the parietal potentials in patients suffering from Huntington's disease (ToÈpper et al., 1993) remains uncertain, but it should be taken into account when analyzing SEPs in patients with predominantly motor de®cits. 4.7. Locus of `gating' Most previous studies have favored a locus of the `gating' effect on SEP in explorative ®nger movements superior to

the thalamus, due to the unaffected amplitude of the subcortical generated P14. This is supported by our results. Since parietal and frontally recordable potentials are altered during `gating' the extent to which pre- and post-central cortical structures or their thalamocortical input are involved remains uncertain. Only a minor reduction in the GFP of the N20 potential at this time point was observed in our data. This might be in part due to the in¯uence of P22 in this time range, as this component is known to be altered by `gating' (Cohen and Starr, 1987). Source reconstruction in the time range of the maximum effect of `gating' induced attenuation indicated a predominantly pre-central localization of the underlying source (8 out of 10 subjects). No signi®cant shift in location with respect to N30 could be seen and both, the electric ®eld distribution and the orientation of the calculated dipoles revealed a predominantly tangential orientation of the underlying generator. Thus, these results suggest a predominantly pre-central location for the mechanisms underlying the `gating' effect during explorative ®nger movements. Of course the presented data of source reconstruction under explorative ®nger movement is not valid to elaborate a hypothesis of the much more complex physiology and pathophysiology underlying the `gating' effect, including centrifugal and centripetal mechanisms (Jones et al., 1989) thalamocortical loops (McCormick and Bal, 1994) and cortico-cortical connections (Yamaguchi and Knight,

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

1990). In our view these questions exceed the possibilities of source localization techniques. But, assuming that our hypothesis is also valid for the `gating' effect, a mechanism restricted to the motor cortex, respectively, its thalamocortical or cortico-cortical input is suf®cient to explain the attenuation of both the frontal N30 and the parietal P27 during explorative ®nger movements. The `gating' effect on parietal potentials may therefore be due to a mirror imaging of the mechanism occurring in the primary motor cortex. Acknowledgements This work was supported by a grant from the Deutsche Forschungsgemeinschaft (Bu 651/5-1). References Akhtari M, McNay D, Mandelkern M, Teeter B, Cline HE, Mallick J, Clark G, Tatar R, Lufkin R, Chan K, Rogers RL, Sutherling WW. Somatosensory evoked response source localization using actual cortical surface as the spatial constraint. Brain Topogr 1994;7:63±69. Allison T, McCarthy G, Wood CC, Darcey TM, Spencer DD, Williamson PD. Human cortical potentials evoked by stimulation of the median nerve. I. Cytoarchitectonic areas generating short-latency activity, J Neurophysiol 1989;62:694±710. Allison T, McCarthy G, Wood CC, Jones SJ. Potentials evoked in human and monkey cerebral cortex by stimulation of the median nerve. Brain 1991;114:2465±2503. Allison T, Wood CC, McCarthy G, Spencer D. Cortical somatosensory evoked potentials. II. Effects of excision of somatosensory or motor cortex in humans and monkeys, J Neurophysiol 1991;66:64±82. Arezzo J, Vaughan Jr, HG, Legatt AD. Topography and intracranial sources of somatosensory evoked potentials in the monkey. II. Cortical components, Electroenceph clin Neurophysiol 1981;51:1±18. Baumgartner C, Sutherling WW, Barth DS. Spatiotemporal modeling of somatosensory evoked magnetic ®elds. In: Williamson SJ, editor. Advances in biomagnetism, New York: Plenum Press, 1990. Baumgartner C, Barth DS, Levesque MF, Sutherling WW. Functional anatomy of human hand sensorimotor cortex from spatiotemporal analysis of electrocorticograhy. Electroenceph clin Neurophysiol 1991;78:56±65. Buchner H, Adams L, Knepper A, RuÈger R, Laborde G, Ludwig I, Reul J, Scherg M. Pre-invasive determination of the central sulcus by dipole source analysis of early somatosensory evoked potentials and 3DNMR-tomography. J Neurochir 1994;80:849±856. Buchner H, Fuchs M, Wischmann HA, DoÈssel O, Ludwig I, Knepper A, Berg P. Source analysis of median nerve and ®nger stimulated somatosensory evoked potentials: multichannel simultaneous recording of electric and magnetic ®elds combined with 3D-MR tomography. Brain Topogr 1994;6:299±310. Buchner H, Waberski TD, Fuchs M, Drenckhahn R, Wagner M, Wischmann HA. Post-central origin of P22: evidence from source reconstruction in a realistically shaped head model and from a patient with a postcentral lesion. Electroenceph clin Neurophysiol 1996;100:332±342. Buchner H, Knoll G, Fuchs M, RienaÈcker A, Beckmann R, Wagner M, Silny J, Pesch J. Inverse localization of electric dipole current sources in ®nite element models of the human head. Electroenceph clin Neurophysiol 1997;102:267±278. Cheron G, Borenstein S. Speci®c gating of the early somatosensory evoked potentials during active movement. Electroenceph clin Neurophysiol 1987;67:537±548. Cheron G, Borenstein S. Gating of the early components of the frontal and

1599

parietal somatosensory evoked potentials in different sensory-motor interference modalities. Electroenceph clin Neurophysiol 1991;80: 522±530. Cheron G, Borenstein S. Mental movement simulation affects the N30 frontal component of the somatosensory evoked potential. Electroenceph clin Neurophysiol 1992;84:288±292. Cheron G, Piette T, Thiriaux A, Jacquy J, Godaux E. Somatosensory evoked potentials at rest and during movement in Parkinsonian's disease: evidence for a speci®c apomorphin effect on the frontal N30 wave. Electroenceph clin Neurophysiol 1994;92:491±501. Cohen D, Cuf®n BN. Demonstration of useful differences between magnetoencephalogram and electroencephalogram. Electroenceph clin Neurophysiol 1983;56:38±51. Cohen LG, Starr A. Localization, timing and speci®city of gating of somatosensory evoked potentials during active movement in man. Brain 1987;110:451±467. Delberghe X, Mavroudakis N, Zegers de Bey D, Brunko E. The effect of stimulus frequency on post- and pre-central short-latency somatosensory evoked potentials (SEPs). Electroenceph clin Neurophysiol 1990;77:86±92. Desmedt JD, Cheron G. Somatosensory evoked potentials to ®nger stimulation in healthy octogenarians and in young adults: wave forms, scalp topography and transit times of parietal and frontal components. Electroenceph clin Neurophysiol 1980;50:404±423. Desmedt JD, Cheron G. Non-cephalic reference recording of early somatosensory potentials to ®nger stimulation in adult or aging normal man: differentiation of widespread N18 and contralateral N20 from the prerolandic P22 and N30 components. Electroenceph clin Neurophysiol 1981;52:553±570. Desmedt JD, Tomberg C. Mapping early somatosensory evoked potentials in selective attention: critical evaluation of control conditions used for titrating by difference the cognitive 30. P40, P100 and N140. Electroenceph clin Neurophysiol 1989;74:321±346. Fuchs M, Drenkhahn R, Wischmann H-A, Wagner M. An improved boundary element method for realistic volume-conductor modeling. IEEE Trans Biomed Eng 1998;45:980±997. Fuchs M, Wagner M, Wischmann H-A, Ottenberg K, DoÈssel O. Possibilities of functional brain imaging using a combination of MEG and MRT. In: Pantev C, editor. Oscillatory event-related brain dynamics, New York: Plenum Press, 1994. pp. 435±457. Furlong PL, Wimalaratna S, Harding GF. Augmented P22±N31 SEP component in a patient with a unilateral space occupying lesion. Electroenceph clin Neurophysiol 1993;88:72±76. Garcia Larrea L, Bastuji H, MauguieÁre F. Unmasking of cortical SEP components by changes in stimulus rate: a topographic study. Electroenceph clin Neurophysiol 1992;84:71±83. Garcia PA, Aminoff MJ, Goodin DS. The frontal N30 component of the median-derived SEP in patients with predominantly unilateral Parkinson's disease. Neurology 1995;45:989±992. Gedes LA, Baker LE. The speci®c resistance of biological material ± a compendium of data for the biomedical engineer and physiologist. Med Biol Eng 1963;5:271±293. GruÈnewald G, GruÈnewald-Zuberbier E, Schumacher H, Mewaldedes J, Noth J. Somatosensory evoked potentials to mechanical disturbances of positioning movements in man: gating of middle-range components. Electroenceph clin Neurophysiol 1984;58:525±536. HaÈmaÈlainen MS, Sarvas J. Realistic conductivity geometry model of the human head for interpretation of neuromagnetic data. IEEE Trans Biomed Eng 1989;36(2):165±171. Huttunen J, TeraÈvaÈinen H. Pre- and post-central cortical somatosensory evoked potentials in hemiparkinsonism. Movement Disord 1993;8:430±436. Huttunen J. Does the P35m SEF de¯ection really come from the motor cortex? Electroenceph clin Neurophysiol 1997;104:101±102. Jones SJ, Halonen JP, Shawkat F. Centrifugal and centripetal mechanisms involved in the `gating' of cortical SEPs during movement. Electroenceph clin Neurophysiol 1989;74:36±45.

1600

T.D. Waberski et al. / Clinical Neurophysiology 110 (1999) 1589±1600

Kawamura T, Nakasato N, Seki K, Kanno A, Fujita S, Fujiwara S, Yoshimoto T. Neuromagnetic evidence of pre- and post-central cortical sources of somatosensory evoked responses. Electroenceph clin Neurophysiol 1996;100:44±50. Lehmann D. Principles of spatial analysis. In: Gevins AS, Remond A, editors. EEG handbook, methods of analysis of brain electrical and magnetic signals, Amsterdam: Elsevier, 1987. pp. 309±354. Lopes da Silva FH, Wieringa HJ, Peters MJ. Source localization of EEG versus MEG: empirical comparison using visually evoked responses and theoretical considerations. Brain Topogr 1991;4:133±142. Lorente de No R. Cerebral cortex: architecture, intracortical connections, motor projections. In: Fulton JF, editor. Physiology of the nervous system, Oxford: University Press, 1938 Ch. 15. Mauguiere F, Desmedt JE. Focal capsular vascular lesions can selectively deafferent the prerolandic or the parietal cortex: somatosensory evoked potentials evidence. Ann Neurol 1991;30:71±75. Mauguiere F, Desmedt J, Courjon J. Astereognosis and dissociated loss of frontal or parietal components of somatosensory evoked potentials in hemispheric lesions ± detailed correlations with clinical signs and computerized tomographic scanning. Brain 1983;106:271±311. Mauguiere F, Broussolle E, Isnard J. Apomorphine-induced relief of the akinetic-rigid syndrome and early median nerve somatosensory potentials (SEPs) in Parkinson's disease. Electroenceph clin Neurophysiol 1993;88(4):243±254. McCormick DA, Bal T. Sensory gating mechanisms of the thalamus. Curr Opin Neurobiol 1994;4:550±556. Mitzdorf U. Current source-density method and application in cat cerebral cortex: investigation of evoked potentials and EEG phenomena. Physiol Rev 1985;65:37±100. Morozov VA. Methods for solving incorrectly posed problems. New York: Springer, 1984. Nicholson Peterson N, Schroeder CE, Arezzo JC. Neural generators of early cortical somatosensory evoked potentials in the awake monkey. Electroenceph clin Neurophysiol 1995;96:248±260. Nunez PL. Localization of brain activity with electroencephalography. In: Sato S, editor. Magnetoencephalography, New York: Raven Press, 1990. pp. 39±65. Rossini PM, Babiloni F, Bernardi G, Cecchi L, Johnson PB, Malentacca A, Stanzione P, Urbano A. Abnormalities of short-latency somatosensory evoked potentials in parkinsonian patients. Electroenceph clin Neurophysiol 1989;74:277±289. Rossini PM, Traversa R, Boccasena P, Martino G, Passarelli F, Paci®ci L, Bernardi G, Stanzione P. Parkinson's disease and somatosensory

evoked potentials: apomorphin-induced transient potentation of frontal components. Neurology 1993;43:2495±2500. Rossini PM, Bassetti MA, Pasqualetti P. Median nerve somatosensory evoked potentials. Apomorphine-induced transient potentiation of frontal components in Parkinson's disease and in parkinsonism. Electroenceph clin Neurophysiol 1995;96:236±247. Rossini PM, Caramia D, Bassetti MA, Pasqualetti P, Techio F, Bernardi G. Somatosensory evoked potentials during the ideation and execution of individual ®nger movements. Muscle Nerve 1996;19:191±202. Rushton DN, Rothwell JC, Craggs MD. Gating of somatosensory evoked potentials during different kinds of movement in man. Brain 1981;104:465±491. Sonoo M, Shimpo T, Takeda K, Genba K, Nakano I, Manne T. SEPs in two patients with localized lesions of the post-central gyrus. Electroenceph clin Neurophysiol 1991;80:536±546. Tiihonen J, Hari R, Haemaelaeinen M. Early de¯ections of cerebral magnetic responses to median nerve stimulation. Electroenceph clin Neurophysiol 1989;74:290±296. ToÈpper R, Schwarz M, Podoll K, DoÈmges F, Noth J. Absence of frontal somatosensory evoked potentials in Huntington's disease. Brain 1993;116:87±101. Wagner M, Fuchs M, Wischmann H-A, Ottenberg K, DoÈssel O. Cortex segmentation from 3D MR images for MEG reconstructions. In: Williamson SJ, Stroink G, Deecke L, Baumgartner C, editors. Biomagnetism: fundamental research and clinical applications, Amsterdam: IOS Press, 1995. pp. 433±438. Wagner M, Wischmann H-A, Fuchs M, KoÈhler T, Drenckhahn R. Current density reconstructions using the L1 norm. In: Aine C, Okada Y, Stroink G, Swithenby S, Wood v, editors. Advances in biomagnetism research, New York: Springer, 1999 BIOMAG 96. Wagner, M., Rekonstruktion neuronaler StroÈme aus bioelektrischen und biomagnetischen Messungen auf der aus MR-Bildern segmentierten Hirnrinde, Aachen: Shaker, 1998. Wood CC, Cohen D, Cuf®n BN, Yarita M, Allison T. Electrical sources in human somatosensory cortex: identi®cation by combined magnetic and potential recordings. Science 1985;227:1051±1053. Yamada T, Rodnitzky RL, Kameyama S, Matsuoka H, Kimura J. Alteration of SEP in Huntington's patients and their relatives at risk. Electroenceph clin Neurophysiol 1991;80:251±261. Yamaguchi S, Knight RT. Gating of somatosensory input by human prefrontal cortex. Brain Res 1990;521:281±288. Yvert B, Bertrand O, Echallier J, Pernier J. Improved dipole localization using local mesh re®nement of realistic head geometries: an EEG simulation study. Electroenceph clin Neurophysiol 1996;99:79±89.