Magnetoencephalographic spikes not detected by conventional electroencephalography

Magnetoencephalographic spikes not detected by conventional electroencephalography

Clinical Neurophysiology 115 (2004) 2041–2047 www.elsevier.com/locate/clinph Magnetoencephalographic spikes not detected by conventional electroencep...

540KB Sizes 1 Downloads 107 Views

Clinical Neurophysiology 115 (2004) 2041–2047 www.elsevier.com/locate/clinph

Magnetoencephalographic spikes not detected by conventional electroencephalography E. Rodina,*, M. Funkeb, P. Bergc, F. Matsuoa b

a Department of Neurology, University of Utah, Salt Lake City, UT, USA Center for Advanced Medical Technology, University of Utah, Salt Lake City, UT, USA c Department of Psychology, University of Konstanz, Baden-Wuertemberg, Germany

Accepted 4 April 2004 Available online 4 May 2004

Abstract Objective: To investigate some of the reasons why magnetoencephalographic (MEG) spikes are at times not apparent in conventional electroencephalograms (EEG) when the data are co-registered, and to explore to what extent modern EEG analysis methods can improve the yield. Methods: Seventy seconds of MEG– EEG co-registration on a 122 channel Neuromag system were studied in a 10-year-old boy with Landau – Kleffner syndrome. Twenty-six EEG channels were originally recorded with a left ear reference. The EEG data were subsequently reformatted (BESA) to a variety of montages for the 10 – 20 and 10 – 10 electrode array. A 10 s data epoch was compared in detail for concordance between MEG and EEG spikes. To detect the characteristics of hidden low voltage EEG spikes, MEG spikes were averaged and compared with the concomitant averaged EEG spike. Results: While there was an abundance of EEG as well as MEG spikes on the left; definite right-sided spikes were not visible in the EEG. Right hemispheric MEG spikes were, however, plentiful with an average strength of 757 fT. When the individual MEG spikes from the right hemisphere were compared with the corresponding EEG events their amplitude ranged between 24 and 31 mV and were, therefore, indistinguishable from background activity. The majority of them became visible, however, with further sophisticated data analysis. Conclusions: When the relative merits of MEG versus EEG recordings for the detection of epileptogenic spike are investigated the 10 – 20 system of electrode placement and conventional methods of EEG analysis do not provide optimal data assessment. The use of the 10 – 10 electrode array combined with modern methods of digital data analysis can provide better concordance with MEG data. q 2004 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. Keywords: Magnetoencephalographic spikes; Electroencephalograph detection; Landau–Kleffner; Epilepsy

1. Introduction Ever since the studies by Cohen and Cuffin (1983) it has been generally agreed that magnetoencephalography (MEG) is more sensitive to neuronal activity which produces tangential dipoles, while electroencephalography (EEG) can detect radial as well as tangentially oriented sources. The disadvantage of EEG is, however, that the various compartments which surround the brain, i.e. CSF, bone, scalp have different conductivities. They attenuate cortical activity and a blurring or smearing effect takes place (Gevins et al., 1994, 1999). These factors are less operative * Corresponding author. Address: 3 Mountainwood Lane, 84092 Sandy, UT, USA. Tel.: þ 1-801-527-5140; fax: þ1-801-576-9746. E-mail address: [email protected] (E. Rodin).

in MEG, and since tangentially oriented dipole sources originate in sulci and sulcal activity vastly outnumbers that of surface cortex it should not be surprising that in certain instances the EEG appears less sensitive than MEG. Furthermore, since modern MEG instruments record neuronal activity with up to 306 sensors an inherent bias against conventional EEG is unavoidable. Apart from these methodological problems there is also the question how an ‘epileptogenic spike’ is defined. Zijlmans et al. (2002) published a study recently in which they pointed out that magnetic spikes may differ in morphology from electrical ones and that criteria as to what defines a magnetic spike are currently lacking. These considerations are important when one evaluates the literature which compares the detection of EEG spikes with those recorded by MEG. When simultaneous

1388-2457/$30.00 q 2004 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.clinph.2004.04.002

2042

E. Rodin et al. / Clinical Neurophysiology 115 (2004) 2041–2047

MEG – EEG evaluations were performed with modern instrumentation, MEG sensors outnumbered EEG channels with a ratio of at least 2:1 in favor of MEG (Assaf et al., 2003; Knowlton et al., 1997; Lantz et al., 2003; Leijten et al., 2003; Lin et al., 2003; Shiraishi et al., 2001; Wheless et al., 1999; Yoshinaga et al., 2002). Furthermore, only conventional methods of EEG analysis have been presented so far. These aspects preclude at present a fair comparison of the relative capabilities of the two recording modalities. The following case presents an example of an apparent superiority of MEG over EEG and points out some of the reasons for the discrepancy.

2. Methods The patient was a 10-year-old boy who had suffered from language regression since 5 years of age. By the time of recording he was nonverbal and showed mild apraxia, but there was no history of epileptic seizures. Co-registration of MEG and EEG was performed on a Neuromag system which utilized 122 planar gradiometers. Simultaneous EEG data were recorded from a specially designed electrode cap with 30 contacts arrayed in an augmented 10 –20 system montage. Forty minutes of continuous MEG – EEG data were obtained using a bandpass filter of 1– 100 Hz at a sampling rate of 300 Hz. Offline EEG and MEG analyses, as reported here, were performed on a 70 s epoch with a commercial software package (BESA).

3. Results The results for conventional EEG analysis are displayed where the same 10 s epoch is presented in various montages. In Fig. 1 the top half demonstrates the data as acquired prior to reformatting for other montages. Due to electrode artifacts only 26 channels are displayed and the left ear had been used as reference. The data show two major spikes with maximum amplitude at FC5, C3 (105 vs. 95 mV), one eye-movement artifact (single asterisk), and a lower voltage somewhat sharp appearing transient at F8 (60 mV, double asterisk). In view of the concomitant positivity at Fp1 and F7 its significance is somewhat ambiguous since it might represent lateral eye movement. The bottom half of the figure shows the data reformatted for an average reference montage. The leftsided spikes and eye-movement artifact are unchanged and the F8 sharp transient now suggests a spike. The data for additional channels were interpolated by the program, using spherical splines (Perrin et al., 1987), and the labels correspond to the 10– 10 terminology (e.g. T7 ¼ T3, P7 ¼ T5, T8 ¼ T4, P8 ¼ T6). Fig. 2 shows the same epoch on bipolar montages with ‘double banana’ on top and transverse on the bottom. On the double banana montage the left-sided spikes show

phase reversal at C3 and the right-sided sharp transient shows phase reversal between F8 and T8 with relative isoelectricity at F8 –T8. The transverse montage on the bottom shows the major phase reversal between T7 and C3. The right-sided transient shows suggestive phase reversal at F8 and to a lesser extent at T8. Fig. 3 displays the MEG for the same epoch. The single asterisk denotes the EEG eye-blink and the double asterisk the right-sided EEG transient. The relevant left-sided sensors are on top and the relevant right-sided sensors on the bottom. The top figure reveals 2 major spikes as well as 3 and possibly 4 additional ones of lower strength. The bottom half shows considerable spiking during nearly every second of the data which could not have been anticipated from the EEG recordings. When spikes were averaged by placing the cursor on the peak of MEG spikes at sensor position M010 (approximate 10– 10 location: FC6), a concomitant EEG spike did become apparent. Fig. 4 provides a montage of the findings. The average of 43 spikes is shown for the EEG on the left with an FC6 spike amplitude of 17 mV. Next to it is the EEG current source density map (CSD). CSD mapping, also referred to as Laplacian montage, was chosen over voltage display because by suppressing activity from deeper structures it provides more discrete localization of cortical phenomena, as has been previously demonstrated (Hjorth, 1975; Perrin et al., 1987; Rodin, 1990, 1999). To the right is the corresponding MEG average with its spike peak at M010 (757 fT) and this is followed by the magnetic flux field. EEG as well as MEG fields show a tangential orientation. When the actual individual EEG events, which had been invisible on the conventional EEG display (top of Fig. 1) were identified by MEG and measured they ranged between 24 and 31 mV and were likewise tangentially oriented. Since it was clear that conventional EEG analysis failed to reveal the numerous right-sided spikes in this sample the question arose whether or not they could be extracted from the EEG by additional electrodes, special filtering (1.6 –15 Hz, 24, 48 db cutoff), CSD display, virtual source modeling, and radical increase in amplifications. Fig. 5 shows the result. The 9 channels represent 5 groups of the most relevant data. The first 2 channels show the CSD display of the data from the 10 –20 system. The next 2 channels reflect the 10 –10 electrode array; the third group shows a virtual source montage model; the 4th the MEG and the ECG is shown for comparison on the bottom. The single asterisk denotes the eye-blink and the double asterisk the originally seen right-sided transient. In the top two channels there is next to the transient, which is marked by the double asterisk a small spike with phase reversal between F4 and C4. The high amplitude eye-movement artifact (asterisk) noted earlier might be regarded to contain a spike because it also shows a phase reversal between F4 and C4. Whether or not other spikes can be discerned in this data set is open to individual interpretation.

E. Rodin et al. / Clinical Neurophysiology 115 (2004) 2041–2047

2043

Fig. 1. Top half shows 10 s of EEG data as initially recorded with left ear reference. Two left-sided spikes and an eye-blink (single asterisk) dominate the tracings. There is also a lower voltage F8 sharp transient (double asterisk). The simultaneous Fp1, F7 positivity may make its interpretation, as a possible lateral eye-movement, uncertain. Bottom half shows same data reformatted to average reference montage. The F8, T8 event can now be regarded as a spike.

When the 10– 10 system electrode array was interpolated from the existing 26 EEG channels, the number of discernable spikes is somewhat increased over the conventional electrode display. This is displayed in channels 3 and 4, and it is apparent that what might have been regarded as a lateral eye movement on the original recording did indeed mask a spike. But it is also noteworthy that the electrical fields are quite discrete and the last C6 spike, especially, is not clearly represented at

neighboring electrodes. Channels 5 and 6 show a virtual source montage model recording and considerably better spike resolution is achieved than with the previous methods. Channels 7 and 8 represent the MEG from sensor positions M010 and M011. It is, therefore, apparent that the virtual source montage model shows much better correspondence to the MEG data than other methods of EEG analysis. The ECG on the bottom shows that the spikes are not artifactual in nature.

2044

E. Rodin et al. / Clinical Neurophysiology 115 (2004) 2041–2047

Fig. 2. Same 10 s epoch. Top half, bipolar double banana montage. Right-sided spike (double asterisk) shows relative isoelectricity at F8 –T8 and phase reversals between F8 and T8. Bottom half, horizontal bipolar montage also shows phase reversal at F8 and less so at T8.

4. Discussion When one reviews the literature which compares the relative spike detection capabilities of EEG versus MEG, it becomes apparent that sophisticated digital EEG analysis methods have not yet been fully utilized in many instances. Our study demonstrated not only the value of adding electrodes to the 10– 20 system but showed also that spike detection could be considerably improved when modern methods of digital EEG analysis were employed.

The limitations of the 10 – 20 system of electrode placements for dipole source analysis have been pointed out repeatedly and again most recently by Lantz et al. (2003). Van der Meij et al. (2001), who compared dipole source localization obtained from 32 EEG channels with those from 84-channel high resolution EEG, in a patient with Rolandic spikes, also found that high resolution EEG clearly demonstrated two sources, one in the precentral and the other in postcentral cortex. This result was similar to the one obtained by MEG. Our case, presented here, demonstrates that more

E. Rodin et al. / Clinical Neurophysiology 115 (2004) 2041–2047

2045

Fig. 3. Same 10 s epoch but MEG. Top half shows the relevant left-sided sensors. In addition to the large two spikes which were obvious in the EEG, several smaller spikes are discernable. Bottom half shows the relevant right-sided sensors and an abundance of smaller spikes, especially at position M010. The eye-blink (single asterisk) concealed a magnetic spike and the double asterisk denotes the position of the EEG detected transient.

advanced data analysis is valuable not only for equivalent dipole source localizations but also for spike detection. There are two factors which currently limit conclusions drawn from MEG – EEG co-registration. Some MEG helmets may not adequately cover the lower temporal areas (Leijten et al., 2003) and may thereby miss information from those regions. The most important aspect is, however, that the available software for MEG systems which co-register the EEG is not optimal for EEG data analysis. The programs were designed primarily for MEG analysis and modern methods of digital EEG analysis are

not incorporated. This disadvantage can be overcome by off-line data analysis with programs which accommodate both EEG and MEG data as shown here. Even after one was alerted to the presence of spikes by MEG in the right hemisphere, rather than the left where they were abundant, only some could be found in the EEG even when CSD analysis was used. Detection improved with the interpolated 10 – 10 electrode array. But a considerably better result was achieved by virtual source modeling as has been described recently by Scherg et al. (2002). These authors have also demonstrated the value of the method for detecting source

2046

E. Rodin et al. / Clinical Neurophysiology 115 (2004) 2041–2047

Fig. 4. Forty-three spikes were averaged from MEG sensor M010. The average of the concomitant EEG spikes is shown on the left with the corresponding CSD map. Spike amplitude at FC6 is only 17 mV. The right side of the figure shows the MEG spike (757 fT at M010) and its flux map.

Fig. 5. This collage of the previously shown 10 s epoch allows for comparison of CSD tracings utilizing the 10 –20 and interpolated 10–10 system, as well as a source montage with MEG. Only the most relevant two channels are shown for each modality. Channel 9 reflects the ECG.

E. Rodin et al. / Clinical Neurophysiology 115 (2004) 2041–2047

activity from deep structures, especially the baso-temporal area. Thus, for a fair assessment of the relative merits of MEG versus EEG, to detect interictal spikes or ictal seizure onset, sophisticated EEG analysis software programs which have similar capabilities as those used for MEG analysis should be employed.

References Assaf BA, Karkar KM, Laxer KD, Garcia PA, Austin EJ, Barbaro NM, Aminoff MJ. Ictal magnetoenecephalography in temporal and extratemporal lobe epilepsy. Epilepsia 2003;44:1320 –7. Cohen D, Cuffin BN. Demonstration of useful differences between magnetoencephalogram and electroencephalogram. Electroencephalogr Clin Neurophysiol 1983;56:38–51. Gevins A, Le J, Martin NK, Brickett P, Desmond J, Reutter B. High resolution EEG: 124-channel recording, spatial deblurring and MRI integration methods. Electroencephalogr Clin Neurophysiol 1994;90:337–58. Gevins A, Le J, Leong H, McEvoy LK, Smith ME. Deblurring. J Clin Neurophysiol 1999;16:204– 13. Hjorth B. An on-line transformation of EEG scalp potentials into orthogonal source derivations. Electroencephalogr Clin Neurophysiol 1975;39:526 –30. Knowlton RC, Laxer KD, Aminoff MJ, Roberts PL, Wong STC, Rowley HA. Magnetoencephalography in partial epilepsy: clinical yield and localization accuracy. Ann Neurol 1997;42:622 –31. Lantz G, Grave de Peralta R, Spinelli L, Seeck M, Michel CM. Epileptic source localization with high density EEG: how many electrodes are needed? Clin Neurophysiol 2003;114:63 –9. Leijten FS, Huiskamp GJ, Hilgersom I, Van Huffelen AC. High-resolution source imaging in mesiotemporal lobe epilepsy: a comparison between MEG and simultaneous EEG. J Clin Neurophysiol 2003;20:227–38.

2047

Lin YY, Shih YH, Hsieh JC, Yu HY, Yiu CH, Wong TT, Yeh TC, Kwan SY, Ho LT, Yen DJ, Wu ZA, Chang MS. Magnetoencephalographic yield of interictal spikes in temporal lobe epilepsy. Comparison with scalp EEG recordings. NeuroImage 2003;19:1115–26. Perrin F, Bertrand O, Pernier J. Scalp current density mapping: value and estimation from potential data. IEEE Trans Biomed Eng 1987;34: 283– 8. Rodin E. Source derivation recordings for cerebral electrical topography. Am J EEG Technol 1990;30:127–38. Rodin E. Decomposition and mapping of generalized spike-wave complexes. Clin Neurophysiol 1999;110:1868–75. Scherg M, Ille N, Bornfleth H, Berg P. Advanced tools for digital EEG review: virtual source montages, whole head mapping, correlation and phase analysis. J Clin Neurophysiol 2002;19:91–112. Shiraishi H, Watanabe Y, Watanabe M, Inoue Y, Fujiwara T, Yagi K. Interictal and ictal magnetoencephalographic study in patients with medial frontal lobe epilepsy. Epilepsia 2001;42:875– 82. Van der Meij W, Huiskamp GJ, Rutten GJ, Wieneke GH, van Huffelen AC, van Nieuwenhuizen O. The existence of two sources in rolandic epilepsy: confirmation with high resolution EEG, MEG and fMRI. Brain Topogr 2001;13:275–82. Wheless JW, Willmore LJ, Breier JI, Kataki M, Smith JR, King DW, Meador KJ, Park YD, Loring DW, Clifton GL, Baumgartner J, Thomas AB, Constantinou JEC, Papanicolaou AC. A comparison of magnetoencephalography, MRI, and V-EEG in patients evaluated for epilepsy surgery. Epilepsia 1999;40:931–41. Yoshinaga H, Nakahori T, Ohtsuka Y, Oka E, Kitamura Y, Kiriyama H, Kinugasa K, Miyamoto K, Hoshida T. Benefit of simultaneous recording of EEG and MEG in dipole localization. Epilepsia 2002;43: 924– 8. Zijlmans M, Huiskamp GM, Leijten FSS, van der Meij WM, Wieneke G, van Huffelen AC. Modality-specific spike identification in simultaneous magnetoencephalography/electroencephalography. A methodological approach. J Clin Neurophysiol 2002;19:183–91.