Usefulness of MEG magnetometer for spike detection in patients with mesial temporal epileptic focus

www.elsevier.com/locate/ynimg NeuroImage 41 (2008) 1206 – 1219

Usefulness of MEG magnetometer for spike detection in patients with mesial temporal epileptic focus R. Enatsu,a N. Mikuni,a K. Usui,b J. Matsubayashi,b J. Taki,a T. Begum,b R. Matsumoto,c A. Ikeda,c T. Nagamine,b,⁎ H. Fukuyama,b and N. Hashimoto a a

Department of Neurosurgery, Kyoto University Graduate School of Medicine, Japan Human Brain Research Center, Kyoto University Graduate School of Medicine, Japan c Department of Neurology, Kyoto University Graduate School of Medicine, Japan b

Received 10 December 2007; revised 9 March 2008; accepted 26 March 2008 Available online 4 April 2008

The present study investigated the sensitivity of magnetoencephalography (MEG) for spikes depending on sensor type in patients with mesial temporal epileptic focus. We recorded MEG in 6 patients with mesial temporal epileptic focus using two sensor types (magnetometer and gradiometer) simultaneously. The number of spikes detected and the corresponding equivalent current dipole (ECD) parameters (distance from the coordinated head center (radius), and dipole moment) were evaluated with respect to sensor type. Among 426 MEG ‘consensus spikes’ determined by 3 reviewers, 378 spikes satisfied the predetermined criteria for source localization. Comparing ECD parameters, spikes detected by magnetometer alone displayed a smaller radius and larger dipole moment than those detected by gradiometer alone. Spikes estimated in the mesial temporal area were more frequently detected by magnetometer alone (38.5%) than by gradiometer alone (11.5%), whereas spikes in the lateral temporal area were detected less by magnetometer alone (3.7%) than by gradiometer alone (53.9%). The present results suggest that a magnetometer is advantageous for spike detection in patients with mesial temporal epileptic focus. This also implies the higher sensitivity of magnetometer for deep sources. © 2008 Elsevier Inc. All rights reserved. Keywords: Magnetometer; Mesial temporal epileptic focus; Magnetoencephalography

Introduction Magnetoencephalography (MEG) is an effective, non-invasive method for localizing epileptogenic zones, and has been increasingly used for presurgical evaluation. Many studies have suggested that presurgical MEG evaluation could avoid invasive intracranial electro⁎ Corresponding author. Human Brain Research Center, Kyoto University Graduate School of Medicine, 54 Kawaharacho, Syogoin, Sakyo-ku, Kyoto 606-8507, Japan. Fax: +81 75 751 3202. E-mail address: [email protected] (T. Nagamine). Available online on ScienceDirect (www.sciencedirect.com). 1053-8119/$ - see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2008.03.038

encephalography (EEG) recording for identifying epileptogenic zones (Knowlton et al., 2006; Fischer et al., 2005; Papanicolaou et al., 2005; Shiraishi et al., 2005; Burneo et al., 2004; Otsuki, 2004; Pataraia et al., 2004; Assaf et al., 2004; Baumgartner, 2004; Barkley, 2004; Knowlton and Shih, 2004; Sakamoto et al., 2003; Pataraia et al., 2002; Stefan et al., 1994; Wheless et al., 1999). Most of these studies have revealed the benefits of MEG in neocortical epilepsy, but little has been reported regarding the usefulness of MEG in patients with mesial temporal lobe epilepsy (MTLE) in epilepsy surgery. Whereas MEG appropriately detects epileptiform discharges arising from superficial areas such as lateral neocortical area (Bast et al., 2005; Knowlton and Shih, 2004; Park et al., 2004), MEG to specify epileptogenic regions sitting deep in the brain has been regarded as difficult. This may be due to reduction of the magnetic signal, which is detected at a long distance (Wennberg, 2006; Pataraia et al., 2005; Knowlton and Shih, 2004; Baumgartner et al., 2004, 2000a,b; Barkley, 2004; Quesney and Ortiz, 2004; Leijten et al., 2003; Barkley and Baumgartner, 2003; Oishi et al., 2002; Ebersole, 1999; Mikuni et al., 1997). Mikuni et al. (1997) also investigated this matter by simultaneously recording MEG and electrocorticography from implanted subdural electrodes. They reported that epileptiform discharges originating synchronously from the lateral temporal lobe might involve cortical areas of at least 4 cm2 when detected by MEG (Mikuni et al., 1997). This result contrasts with the findings of Oishi et al. (2002) that source estimation could not fulfill the predetermined criteria even when epileptiform discharges extended over 4 cm2 in patients with MTLE. Application of MEG has thus been considered to have limitations in the measurement of MTLE. Looking back at these previous studies, the fact that a gradiometer has been used as the MEG pick-up coil should not be overlooked (Pataraia et al., 2005; Knowlton and Shih, 2004; Baumgartner et al., 2004, 2000a,b; Barkley, 2004; Leijten et al., 2003; Barkley and Baumgartner, 2003; Oishi et al., 2002; Ebersole, 1999; Mikuni et al., 1997). A gradiometer is a two-loop gradient coil designed to cancel environmental magnetic noises that poorly detect

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

1207

Table 1 Patient characteristics Case

Age

Sex

Epileptogenic region

EEG

MRI

PET/SPECT

Histology

Surgical outcome

Test data 1 15

M

Right mesial temporal lobe

Right HS

I

33

F

–

–

3

23

M

Left mesial temporal lobe (post-right hippocampectomy) Right mesial temporal lobe

Right temporal hypometabolism No abnormality

HS

2

Bilateral fronto-temporal spike(+) (right N left) Left fronto-temporal spike(+) Right temporal spike

HS lateral temporal gliosis

I

4

15

F

Right mesial temporal lobe

Ganglioglioma

I

5

53

F

Right mesial temporal lobe

HS

I

6

22

M

Right mesial temporal lobe

Right temporal hypometabolism Right temporal hypometabolism Right temporal hypometabolism Right temporal hypometabolism

HS

I

Left temporal hypometabolism

–

–

Sample data 7 33 F

Left lateral temporal lobe

Left HS Right HS

Right fronto-temporal spike(+) Right fronto-temporal spike(+) Right fronto-temporal spike(+)

Right temporal tumor Right HS

Left temporal spike

No abnormality

Right HS

HS: hippocampal sclerosis.

signals from deeper sources. A different type of MEG pick-up coil, a magnetometer, comprises a single loop coil and is designed to detect signals arising from deeper regions. The magnetometer offers theoretical advantages in detecting deeper sources, but has rarely been used in recent years, due to susceptibility to interference from external magnetic noise (Knowlton and Shih, 2004; Tarkiainen et al., 2003; Fagaly, 1990). Recent advances in technology, such as the quality of magnetically shielded rooms and the coil itself, have allowed magnetometers to be used for clinical purposes. Despite theoretical superiority, almost no reports have demonstrated advantages in detecting epileptiform discharges, particularly those from deeper sources in a clinical study. MTLE is the most common medically intractable partial epilepsy. Several non-invasive methods have been established for presurgical work-up in MTLE (Mayanagi et al., 1996). Although we have used MEG for clinical evaluation in many types of epilepsy, we have not obtained favorable results for MTLE, possibly because of the poor sensitivity of the gradiometer for deeper sources. We are presently seeking to evaluate the merits of the magnetometer, which has recently been introduced to commercially available machines and is theoretically sensitive for deeper sources when assessing mesial temporal epileptic foci (Knowlton and Shih, 2004; Fagaly, 1990). To achieve this aim, we needed to solve two different unresolved problems: criteria for MEG epileptiform discharges; and variability of detection among EEG/MEG reviewers. Although several criteria have been defined for EEG epileptiform discharges, whether those criteria can be applied to MEG epileptiform discharge is uncertain, as the morphologies could differ (de Jongh et al., 2005; Fernandes et al., 2005). These differing signal characteristics between MEG and EEG can confuse readers when asked to detect epileptiform discharges independently. To establish a reasonable criterion for MEG epileptiform discharge, we undertook a twostep procedure: establishment of a tentative criterion for MEG epileptiform discharge; and evaluation using another data set differing from the test data. Another issue is the reviewer factor in detection of epileptiform discharges. Detecting epileptiform discharges not only depends upon discharge morphology and signal-to-noise ratio, but also upon

the training and experience of the reviewer (Zijlmans et al., 2002). Even when using the same criteria, the reviewer factor cannot be sufficiently excluded. We therefore let 3 experienced neurologists review data after blinding to sensor type and patient information. Among the ‘possible spikes’ detected by each reviewer based on the criterion of epileptiform discharge, spikes detected by 2–3 reviewers were accepted as ‘consensus spikes’, as described by Leijten et al. (2003) and Zijlmans et al. (2002). The present study used an MEG machine with two types of pickup coils to detect epileptiform discharges, and determined the type of sensor able to detect the larger number of spikes, depending on source location. Simultaneous recording of EEG was also adopted for reference.

Methods Patients Six patients suffering from medically refractory epilepsy with mesial temporal epileptic focus (Cases 1–6) and 1 patient with lateral temporal lobe epilepsy (Case 7) were recruited from among patients treated in the Departments of Neurosurgery and Neurology at Kyoto University Hospital, Japan, between July 2004 and May 2005 (Table 1). The 6 patients (3 men, 3 women) with mesial temporal epileptic focus were selected because of structural abnormalities in the mesial temporal lobe on magnetic resonance imaging (MRI) (Fig. 1) and frequent clear EEG spikes concordant with ictal semiology by history. Mean age of these patients was 26.8 years (range, 14–53 years). The patient with lateral TLE was a 33-year-old woman, who was arbitrarily selected from among patients showing frequent EEG and MEG spikes. All patients underwent MRI and interictal positron emission tomography (PET) and/or single photon emission computed tomography (SPECT). Continuous video-EEG monitoring with 21 electrodes (19 electrodes of 10-20 system added T1 and T2 electrodes of 10-10 system) was performed for 1 or 2 weeks in all patients except Case 4. At the time of MEG measurements,

1208

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

only Case 2 had a history of previous epilepsy surgery (because of frequent complex partial seizures). The right hippocampus had already been resected on the basis of congruent data (ictal scalp EEG, ictal EEG with bilateral depth EEG, and interictal hypoperfusion on SPECT and hypometabolism on FDG-PET). The patient had been seizure-free for more than 1 year following resection of the right hippocampus. EEG-wise, frequent spikes from the left temporal lobe were observed. Five cases (excluding Cases 2 and 7) underwent epilepsy surgery after MEG measurement, and became seizure-free for 1 to 3 years (Engel's class I) (Engel et al., 1993). Intraoperative electrocorticography was recorded in Cases 1, 3, 5 and 6. This procedure documented that interictal spikes arose from both the mesial and lateral temporal cortices in these 4 cases, although epileptogenic zones were located at the mesial temporal lobes (Fig. 1).

Table 2 Criteria for MEG epileptiform discharge

Data acquisition

2. Exclusion Remarkable artifact

All measurements were performed inside a magnetically shielded room using a helmet-shaped whole-head coverage MEG system comprising 102 identical units, each housing two orthogonally-placed planar figure-of-eight-shaped gradiometers and one magnetometer, resulting in 306 channels in total (VectorView; Elekta Neuromag, Helsinki, Finland). Each triple-sensor detector unit was a 28 mm × 28 mm square in shape, with a mean distance between sensor elements of 34.6 mm. EEG was recorded simultaneously during MEG measurement. Based on the International 10-20 or 10-10 Systems, 8–12 bilateral fronto-temporal EEG electrodes on which spikes were present during previous video-EEG monitoring or routine EEG were used for 6 patients with mesial temporal epileptic focus (Cases 1–6) and 19 electrodes for the lateral TLE patient (Case 7). The aim was to monitor the epileptiform discharges found on previous video-EEG monitoring or routine EEG. All EEG recordings were referred to the ear electrode ipsilateral to the epileptic side. Obtained EEGs were reformatted utilizing the calculated linked-ear reference during analysis. Electrooculography was also recorded simultaneously during MEG acquisition. Contamination by magnetic artifacts derived from the heart was monitored by electrocardiography in Cases 4–6. Continuous data were acquired at a sampling rate of 603 Hz for all signals of MEG and EEG using a band-pass filter of 0.1–200 Hz. Each recording session typically took 30–60 min and consisted of 10- to 30-min sessions with breaks in between. Patients were recorded in a supine position and instructed to rest or sleep. Four head position indicator coils were placed on the scalp, with the location determined based on anatomical fiducial points (nasion, bilateral preauricular points) using a three-dimensional digitizer (Polhemus, Colchester, VT, USA). The head shape of each patient was subsequently determined by checking evenly distributed points covering the whole scalp using the digitizer. To find the exact head position with respect to sensors, weak electrical current was led to the coils and the resulting magnetic signals were measured just before each measurement epoch. These sets of information were used for MEG-MRI co-registration.

1209

Samples 1. Inclusion Sharp peak Duration ranging from 20 ms to 200 ms Involvement of more than 2 MEG channels Outstanding from High amplitude the background Independent ongoing activity from background rhythm Followed by after-slow wave

Criteria for MEG epileptiform discharge We adopted 4 criteria in detecting ‘possible spikes’ on MEG, according to the principles recommended by the International Federation of Clinical Neurophysiology (IFCN) for EEG epileptiform discharge (Cobb, 1983) (Table 2): #1 #2 #3 #4

sharp peak duration 20–200 ms outstanding from ongoing background activity involvement of more than two MEG channels.

The term ‘outstanding’ used in criterion #3 was applied if signals had at least one of the following 3 features: high amplitude; independence from background rhythm; or being followed by a slow wave. Criterion #4 was employed to exclude sharp noise caused by unstable signals in noisy channels. To confirm the validity of these criteria in detecting epileptiform discharges in advance, two other experienced neurophysiologists (R.E, J.M) reviewed 5-min magnetometer, gradiometer and EEG datasets obtained from the patient with left lateral temporal lobe epilepsy (Case 7). A 2–40-Hz band-pass filter was applied. For the sake of simple detection of MEG spikes, appropriate numbers and locations of channels were selected depending on the type of sensor in the following way, taking into consideration the signal distribution. The signal distribution detected by the magnetometer forms two peaks of influx and outflux away from the location of the generator source, whereas that detected by the planar gradiometer shows maximum amplitude just above the generator source (Hämäläinen et al., 1993). As a result, 13 pairs at 13 locations over the left temporal area were selected for planar gradiometer traces, with the intention of choosing the maximum signal at the center of selection. In contrast, 26 magnetometer channels taken from 26 locations over the left fronto-temporal area were chosen to include both influx and outflux, which were

Fig. 1. MRI images of Cases 1–6. MRI images of Case 1, 3, 5, 6 showed right hippocampal atrophy with increased intensity on fluid-attenuated inversion recovery (FLAIR) image (FLAIR image of Case 6 showing this high intensity scanned in the other institute was not available.). Preoperative FLAIR image of Case 2 showed right hippocampal atrophy and hyperintensity of the left hippocampus. T1 and T2 images of Case 4 showed cystic tumor in the right temporal lobe. White circles of Cases 1, 3, 5, 6 are the areas where spikes of intraoperative electrocorticogram were detected.

1210

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

Fig. 2. Schematic depiction of the preparation of datasets and examples of display during 5 s of simultaneous EEG, gradiometer and magnetometer recordings. EEG electrodes were selected based on the results of video-EEG monitoring or routine EEG. MEG sensor locations are shown schematically on the lower side (filled circle). For gradiometers, sensors in the temporal area were selected, while sensors in the fronto-temporal area were selected for magnetometers.

observed away from just above the generator source. For EEG data, all 19 electrodes were selected in this patient (Case 7). The reviewers were given 26 traces of either sensor type and instructed

to detect epileptiform discharges as ‘possible spikes’ by applying the MEG epileptiform discharge criteria for MEG data. All reviewers were blinded to the type of sensor. The reviewers

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

1211

Fig. 3. Schematic depiction of the study protocol. ‘Possible spikes’ were identified from simultaneous EEG, gradiometer and magnetometer recordings, and were reviewed independently by 3 reviewers. ‘Possible spikes’ detected by two or more reviewers were accepted as ‘consensus spikes’ for further analysis. We compared the following: 1-A) the number of spikes with respect to the combination of detection modality (EEG and MEG); 1-B) the number of spikes for each sensor type with respect to the combination of detected sensor types; 2-A) ECD parameters (radius, dipole moment) between MEG + EEG spikes and MEG-only spikes; 2-B) ECD parameters (radius, dipole moment) between MEG sensor types; and 3) the number of spikes with respect to the location (mesial and lateral spikes). EEG spike: spike detected by EEG; MEG spike: spike detected by MEG; EEG-only spike: spike detected by EEG only; MEG-only spike: spike detected by MEG only; MEG + EEG spike: spike detected by both MEG and EEG; M-only spike: spike detected by magnetometer only; MG spike: spike detected by both magnetometer and gradiometer; G-only spike: spike detected by gradiometer only.

checked 19 traces of EEG data, following the recommendations of the IFCN for EEG data (Cobb, 1983). The number of ‘possible spikes’ detected by magnetometer, gradiometer and EEG were 228, 134, and 157 for the first reviewer and 247, 164, and 206 for the second reviewer. The coincidence rate of spike detection between the two reviewers was evaluated based on the method of Zijlmans et al. (2002). Datasets were divided into virtual epochs of 500-ms duration to determine whether peaks of ‘possible spikes’ detected by the two reviewers occurred in the same epoch. Inter-reviewer kappa values were 0.83

for magnetometer, 0.87 for the gradiometer and 0.80 for EEG. Landis and Koch (1977) suggested that inter-reviewer kappa values exceeding 0.80 can be regarded as “almost perfect”, validating this tentative criterion for application in detecting MEG epileptiform discharges. Spike identification To delineate spike configuration, a 2–40-Hz band-pass filter was first applied to all traces taken from the 6 patients. A typical

1212

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

Table 3 Number of spikes detected by EEG and MEG as is shown in comparison (1-A) in Fig. 3 Case

EEG-only

MEG + EEG

MEG-only

Total

1 2 3 4 5 6 Total

19 1 13 6 1 0 40

19 0 95 1 0 5 120

13 19 186 54 27 7 306

51 20 294 61 28 12 466

EEG-only spikes; spikes detected by EEG only. MEG + EEG spikes; spikes detected by both MEG and EEG MEG-only spikes; spikes detected by MEG only.

15-min section of continuous recording that contained frequent epileptiform discharges was then selected from the 30–60-min recording for each patient. For convenience of detection, representative channels were selected as employed in the test data. We therefore selected 13 pairs of orthogonally arranged planar gradiometers at 13 locations in the temporal areas and 26 magnetometer channels in 26 locations in the fronto-temporal areas. For EEG, all of the 8–12 electrodes in bilateral fronto-temporal areas were selected (Fig. 2). For each patient, the section of 15-min recordings was divided into 90 contiguous segments of 10 s each. One segment for each set of channel selections (gradiometer, magnetometer, or EEG) was printed on a sheet. These 90 sheets within the same set of channel selections were arranged and sorted randomly. For gradiometer and magnetometer, a 10-s recording of background activity taken from 26 channels at the occipital area was also printed on a piece of paper at the same scale as a reference. Therefore, a bundle of 90 data sheets and 1 reference sheet were prepared for the gradiometer and magnetometer, while a bundle of 90 data sheets was prepared for EEG. As epileptiform discharges were detected in bilateral temporal areas independently in Cases 1 and 2, channels for the magnetometer and gradiometer were selected for both sides in these cases. For the remaining cases (Cases 3–6), channels were selected only from the right temporal area where EEG spikes had been previously detected. As a result, 5 sets were prepared in total for each of Cases 1 and 2: 1 set of EEG channels; 2 sets of magnetometer channels selected from the right and left fronto-temporal areas; and 2 sets of gradiometer channels selected from the right and left temporal areas. Three sets were prepared for each of Cases 3–6: 1 set of EEG channels; 1 set of magnetometer channels from the right fronto-temporal area; and 1 set of gradiometer channels from the right temporal area. In total, 22 channel selection sets were prepared for Cases 1–6. A total of 22 bundles were also rearranged randomly irrespective of subject and channel selection set. These preparations were performed by the first author (R.E). The 22 bundles (5 sets for Cases 1 and 2; 3 sets for Cases 3–6) of either 90 or 91 sheets were inspected independently by 3 reviewers who were blinded to channel selections and patient information. The 3 reviewers were medical doctors (J.T, T.B, K.U) with more than 2 years of experience in reading EEG who were recruited from among the authors. The reviewers were instructed to mark ‘possible spikes’ following the above-mentioned criteria for MEG, and IFCN criteria for EEG. A ‘possible spike’ detected by at least two of the 3 reviewers in each sensor type of MEG or EEG was defined as a ‘consensus spike’ in that category (Leijten et al., 2003; Zijlmans et al., 2002).

Estimation of generator sources for spikes We estimated generator sources for ‘consensus spikes’ detected by MEG assuming equivalent current dipoles (ECDs) in a realistic head model (boundary-element model). Topographic magnetic field maps were estimated from the gradiometer and magnetometer signals at the peak latency of the spikes. An isocontour map was then calculated and constructed by determining the minimum-norm current estimate (MNE) using the measured signals of both magnetometer and gradiometer. When the contour map showed a dipolar pattern at the peak latency of the spikes, source estimation was conducted from at least 9 sets of channels that had a maximum local signal at the center of selected gradiometers. ECDs that best explained the most dominant signals were determined using the MNE method (Hämäläinen et al., 1993). We accepted calculated ECDs if goodness-of-fit (g%) was N 80%. Finally, estimated ECDs were superimposed on the MRI of the subject after co-registration of MRI and MEG coordinates by matching the positions of anatomical fiducial points (nasion and preauricular points). To delineate the characteristics of deep and superficial sources, we selected spikes derived from the mesial and lateral temporal areas using the following procedures. On the base of the temporal lobe on MRI in the coronal plane that included the hippocampus, we applied two landmarks as imaginary vertical lines at the levels of the collateral and occipitotemporal sulci. Among the estimated ECDs from accepted ‘consensus spikes’ located in the temporal lobe, spikes with ECDs located mesial to the vertical line at the collateral sulcus were classified as mesial temporal spikes. Spikes with ECDs located lateral to the vertical line at the occipitotemporal sulcus were regarded as lateral temporal spikes. Statistical analysis The number of detected spikes and ECD parameters were evaluated depending on the combination of EEG and MEG, and on the sensor types (Fig. 3). We compared two parameters of the estimated ECD: radius; and dipole moment. Radius indicates the distance from the coordinated head center. Dipole moment describes the current moment of the estimated ECD (Hämäläinen et al., 1993). Wilcoxon's signed-rank test was used to evaluate differences between two sets of paired data, while the Mann–Whitney U test was used to evaluate differences between two sets of unpaired data. The Kruskal–Wallis test with Bonferroni correction was used to evaluate differences among three sets of unpaired data, and the Friedman test with Bonferroni correction was used to evaluate the differences among three sets of paired data.

Table 4 Number of spikes among the detection by combination of MEG sensor types as is shown in comparison (1-B) in Fig. 3 Case

G-only

MG

M-only

Total

1 2 3 4 5 6

11 2 115 34 11 7

10 3 144 7 5 4

11 14 22 14 11 1

32 19 281 55 27 12

M-only spike: spike detected by magnetometer only. MG spike: spike detected by both magnetometer and gradiometer. G-only spike: spike detected by gradiometer only.

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

Results Number of detected spikes Comparison between EEG and MEG A total of 160 EEG and 426 MEG ‘consensus spikes’ were accepted from the 6 patients with a mesial temporal epileptic focus.

1213

Among these spikes, 40 were detected by EEG only (EEG-only spikes; range, 0–19 in each case), 120 by both MEG and EEG (MEG + EEG spikes; range, 0–95), and 306 by MEG only (MEG-only spikes; range, 7–186) (Table 3). Percentages of each spike type in each patient were 0–37.3% for EEG-only spikes, 0–41.7% for MEG + EEG spikes, and 25.5–96.4% for MEG-only spikes. The number of MEG spikes (MEG-only and MEG + EEG spikes) was significantly greater than

Fig. 4. a)Representative examples of M-only, MG and G-only spikes from two time segments. Upper, middle and lower groups of channels show tracings taken from magnetometers, gradiometers and EEG, respectively. These representative spikes shown here were accepted as ‘consensus spikes’ by MEG only, but not by EEG. b) Isocontour maps of each presented spikes. The black line shows magnetic flux coming out of the head and the gray line shows that going into the head. These maps show no remarkable difference of inter-flux peak distance among these spikes.

1214

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

Fig. 4 (continued ).

the number of EEG spikes (EEG-only and MEG + EEG spikes) (n = 6, Wilcoxon's signed-rank test; p = 0.046). Comparison among the combinations of MEG sensor types Among the 426 spikes detected by MEG sensors, 180 were detected by gradiometer only (G-only spikes; range, 2–115 spikes in each case), 173 were detected by both sensor types (MG spikes; range, 3–144), and 73 were detected by magnetometer only (M-only spikes; range, 1–22) (Table 4). No significant differences were seen in the number of detected spikes between magnetometer (M-only + MG) and gradiometer (G-only + MG) (n = 6, Wilcoxon's signed-rank test; p = 0.273). Typical examples of these 3 kinds of spikes and the contour maps with simultaneous EEG recordings are shown in Fig. 4. On the visual inspection of the contour maps, there was no remarkable difference of inter-flux peak distance among these spikes. ECD parameters Comparison between EEG and MEG Of 426 MEG spikes, 378 satisfied the criteria for source localization. These 378 spikes included 275 MEG-only spikes (Cases 1–6) and 103 MEG + EEG spikes (Cases 1, 3 and 6). We compared the radius and moments of the estimated ECD between MEG-only spikes and MEG + EEG spikes in Cases 1, 3 and 6. With respect to ECD radius, the range of median values in each patient was 49–94 mm for MEG-only spikes and 46–59 mm for MEG + EEG spikes (Fig. 5a). Among those cases, Cases 3 and 6

showed significantly larger radii for MEG-only spikes than for MEG + EEG spikes (Mann–Whitney U test; Case 3, p b 0.001; Case 6, p = 0.047). Ranges of median ECD moments in each case were 93–317 nAm for MEG-only spikes and 417–676 nAm for MEG + EEG spikes (Fig. 5a). Evaluation of Cases 3 and 6 showed that dipole moments of MEG + EEG spikes were significantly larger than those of MEG-only spikes (Mann–Whitney U test; Case 3, p b 0.001; Case 6, p = 0.028). Regarding Case 1, both the radius and ECD moment showed no significant difference between the two spike groups. With regard to correlation between radius and dipole moment, MEG-only spikes tended to have a larger radius and smaller dipole moment and MEG + EEG spikes had a smaller radius and larger dipole moment. These correlations were significant (Spearman's Rank Correlation Coefficient between radius and dipole moment: MEG-only spike, − 0.323; p b 0.01, MEG + EEG spike, − 0.567; p b 0.01) (Fig. 5b). Comparison among the combination of MEG sensor types Thereafter, these parameters of the estimated ECD were compared with respect to the combination of sensor types by which the spikes had been detected. With respect to radius, the range of median values was 43–59 mm for M-only spikes, 47–92 mm for MG spikes and 56–92 mm for G-only spikes (Fig. 6a). Cases 1–5 had a common tendency in that the radius of MG spikes was larger than that of M-only spikes, but smaller than that of G-only spikes (Case 6 had no M-only spike). Comparison on an individual basis revealed that G-only spikes showed a significantly larger radius than M-only spikes in Cases 3 and 4 (Kruskal–Wallis test; Case 3, p b 0.001 with Bonferroni correction;

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

1215

Fig. 5. a) Plots of ECD parameters for MEG-only spikes and MEG + EEG spikes. b) Relationship between radius and dipole moment. MEG-only spikes tended to display a larger radius and smaller dipole moment, while MEG + EEG spikes showed a smaller radius and larger dipole moment.

Case 4, p = 0.006 with Bonferroni correction). The remaining cases (Cases 1, 2 and 5) displayed no significant differences, possibly due to the small number of spikes. The range of median dipole moment values for each case was 130–435 nAm for M-only spikes, 138–689 nAm for MG spikes and 94–340 nAm for G-only spikes (Fig. 6a). Although Cases 1–5 showed a common tendency in that the dipole moment of M-only spikes was larger than that of G-only spikes (Case 6 was excluded from this evaluation due to a lack of M-only spikes), only Case 4 demonstrated a significantly larger dipole moment for M-only spikes than for G-only spikes (Kruskal–Wallis test; p b 0.001 with Bonferroni correction). With regard to the correlation between radius and dipole moment, M-only spikes tended to have a smaller radius with relatively larger dipole moment, while G-only spikes had a larger radius and smaller dipole moment. These correlations were significant (Spearman's Rank Correlation Coefficient between radius and dipole moment: M-only spike, − 0.273, p b 0.05; G-only spike, − 0.383, p b 0.01 (Fig. 6b). Superimposition of ECDs onto MRI demonstrated the tendency that ECDs of M-only spikes were mainly located in the mesial temporal area, while those of G-only spikes were mainly in the lateral temporal area (Fig. 7).

magnetometer only (range, 0–11), and 49 spikes were detected by both magnetometer and gradiometer (range, 0–35). The number of spikes detected by magnetometer (M-only + MG spikes) was significantly larger than the number detected by gradiometer (G-only + MG spikes) (n = 6; Wilcoxon signed-rank test, p = 0.041). In contrast, among lateral spikes, 128 spikes were detected by gradiometer only (range, 1–103 for each case), 8 spikes were detected by magnetometer only (range, 0–4), and 99 spikes were detected by both magnetometer and gradiometer (range, 0–95). The number of spikes detected by gradiometer (G-only + MG spikes) was significantly larger than the number detected by magnetometer (M-only + MG spikes) (n = 6; Wilcoxon signed-rank test, p = 0.042). Among mesial spikes, 0–85.7% in each case were M-only spikes and 0–28.6% were G-only spikes, while for lateral spikes, 0–50% were M-only spikes and 44.4–100% were G-only spikes. In summary, spikes in the mesial temporal area were more often detected by magnetometer alone (38.5%) than by gradiometer alone (11.5%), whereas spikes in the lateral temporal area were less often detected by magnetometer alone (3.7%) than by gradiometer alone (53.9%) in total.

Comparison between mesial and lateral spikes

We investigated whether the magnetometer or gradiometer is more advantageous in signal detection for deep sources. For this purpose, we chose spikes arising from the mesial temporal structure in patients with epilepsy as samples of deep sources. Simultaneous gradiometer, magnetometer and EEG recordings were performed in these patients, and the degree of spike detection was rated for each sensor by blinded reviewers. Our results indicate that among the

Among the 378 spikes for which ECDs were available, 93 and 235 spikes were selected as mesial and lateral spikes, respectively, according to anatomical locations on MRI (Figs. 7, 8). Concerning mesial spikes, 9 spikes were detected by gradiometer only (range, 0–3 for each case), 35 spikes were detected by

Discussion

1216

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

Fig. 6. a) Plots of ECD parameters of spikes detected by each sensor type. b) Relationship between radius and dipole moment. Spikes detected solely by magnetometer showed a smaller radius with relatively large dipole moment, while those detected solely by gradiometer showed a larger radius and smaller dipole moment.

sensors, the magnetometer is the most useful for detecting mesial spikes. Comparing EEG and MEG for spike detection (Table 3), MEG detected a larger number of spikes than EEG in all except Case 1, in whom the number of spikes detected by EEG was larger than that detected by MEG. This result indicates that MEG is superior to EEG. This may be due to the lack of reduction of magnetic field by the skull, which is totally different from the detection of current Furthermore, the orientation and distribution of spikes may affect the difference of sensitivity. However these matters were not investigated at the present study. This finding is congruent with earlier studies that have conducted similar experiments in patients with focal epilepsy (Knake et al., 2006; Rodin et al., 2004; Ramantani et al., 2006; Pataraia et al., 2005; Iwasaki et al., 2005; Zijlmans et al., 2002). Among 3 cases in which ECDs were obtained for both MEG-only spikes and MEG + EEG spikes, 2 cases demonstrated a tendency for MEG-only spikes to have a larger radius and smaller dipole moment than MEG + EEG spikes (Fig. 5b). This result suggests that MEG is superior to EEG for superficial spikes even if the dipole moment is small. This finding agrees with previous reports of the advantages of MEG over EEG in spike detection, particularly for patients with lateral neocortical epilepsy (Baumgartner et al., 2000a,b; Knake et al., 2006; Lin et al., 2003). If spikes were located in the superficial area, MEG could detect spike activity spreading over an area as small as 4 cm2 (Mikuni et al., 1997), whereas EEG picks up an areas as small as 10 cm2 (Tao et al., 2005).

Comparing ECD parameters between each type of MEG sensor, the magnetometer is advantageous for detecting spikes that have a smaller radius and larger dipole moment, while the gradiometer is advantageous for spikes with a larger radius and smaller dipole moment (Fig. 6). This result suggests that the magnetometer is superior for deeper sources and the gradiometer is superior for superficial sources. This proposal is in agreement with an experimental study by Fagaly (1990), which found that sensitivity of the gradiometer decays more rapidly over distance than sensitivity of the magnetometer. The gradiometer has the characteristic of canceling out signals that arise from distant areas, offering an advantage for noise cancellation. However, this characteristic of the gradiometer has been considered as a disadvantage for deeper sources. Conversely, the magnetometer appears to display better sensitivity for deep sources and less sensitive for superficial sources compared with the gradiometer. As magnetometer recordings are susceptible to interference from environmental noise, spike detection using a magnetometer requires relatively large dipole moment to overcome background noise even for superficial spikes located in the lateral area. In addition, since the dipolar pattern generated by superficial spikes is condensed in space, the size of the sensor unit and the distance between sensor elements of magnetometer might not be small enough to delineate a dipolar pattern. We consider that these features led to the higher sensitivity of the magnetometer for signals arising from deep sources and that produced

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

1217

Fig. 7. Anatomical ECD locations of M-only and G-only spikes. ECDs were classified in terms of mesial and lateral groups using imaginary vertical lines from the collateral sulcus and occipitotemporal sulcus as landmarks. Mesial temporal spikes: spikes with ECDs located mesial to the collateral sulcus; Lateral temporal spikes: spikes with ECDs located lateral to the occipitotemporal sulcus.

a large dipole moment, and the lower sensitivity of signals located in the superficial area. To compare sensitivity between sensor types with respect to ECD locations, spikes were classified as lateral or mesial spikes on the basis of the location of ECDs on each the MRI of each patient. Among mesial spikes, the range of M-only spikes (except for Case 6) was 13.6–85.7%, greater than the range for G-only spikes (Fig. 8). In contrast, among lateral spikes, the range of percentages for G-only spikes was 44.4–100%, indicating a significantly larger

number of spikes for the gradiometer than for the magnetometer (Fig. 8). This suggests that the magnetometer is more advantageous for mesial spikes and that the gradiometer is preferable for lateral spikes. Related to this issue, dipole moments of the lateral spikes are smaller than those of mesial spikes (Figs. 5b, 6b). This tendency might be derived solely from the matter of selection bias, given the small number of patients recruited. Another possibility is that patients with mesial temporal epileptic foci might have smaller epileptiform

Fig. 8. Populations of spikes that were detected by gradiometer only, by magnetometer only, and by both gradiometer and magnetometer for mesial and lateral spikes, as shown on a percentage scale for 6 patients. Among the 93 spikes located in the mesial temporal area, magnetometer detected a significantly larger number of spikes than gradiometer (n = 6, Wilcoxon signed-rank test; p = 0.041). Among the 235 spikes located in the lateral temporal area, gradiometer detected a significantly larger number of spikes than magnetometer (n = 6, Wilcoxon signed-rank test; p = 0.042).

1218

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219

discharges arising from the lateral neocortex compared with those arising from the mesial temporal structure. We also have to consider the other possibility that this trend is affected by the properties of least square fit which tend to estimate a source at deeper location as a source with larger strength at deeper location. However, considering the larger variation of source strength in medial spikes compared to that in lateral spikes, this effect must not be large. In this respect, whether the superiority of gradiometer over magnetometer in detecting lateral spikes is also true for spikes with a large dipole moment in that area is uncertain. Further studies with an increased number of patients and other types of epilepsy are thus needed, focusing on lateral spikes that can have large dipole moments. Nevertheless, the results of the present study clearly indicate the advantage of magnetometers for mesial spikes, even considering the limitations of patient number and epilepsy type. With respect to the relation of orientation and location, Ebersole (1997) presented that anterior temporal horizontal dipole correlated with mesial temporal lobe onset, the anterior temporal vertical dipole with anterior and perhaps mesial temporal lobe onset, and the posterior temporal vertical dipole with lateral or nonlocalized onset (Ebersole, 1999, 1997a,b). However, there was not remarkable difference of the orientation between mesial and lateral spikes when we investigated. In the present study, spikes with ECDs located mesial to the vertical line at the collateral sulcus were classified as mesial temporal spike and lateral to the vertical line at the occipitotemporal sulcus as lateral temporal spikes. This classification was not correlated to Ebersole's classification strictly. Furthermore, it is important that ECD is just a representation of a certain extent of the activated area. This may be the reason why our result differed from the previous study. The number of channels showing spikes can be another important factor in visual inspection. The present study employed visual inspection, which is still regarded as the criteria most frequently used for spike detection. Waveform configuration is important in visual inspection, and gradiometers tended to show clearer waveforms than magnetometer in G-only spikes, while magnetometers showed clearer waveforms in M-only spikes (Fig. 4a). M-only spikes showed a larger number of channels and broader distribution, which may strengthen the impression of spikes. The advantage of magnetometers for spike detection can also be explained by this broad distribution, in addition to the signal-to-noise ratio. Concerning this point, as the number of channels in the present study was confined to 26 for both magnetometer and gradiometer, some spikes might not have been detected using this small coverage of magnetometers, and the advantage of magnetometers might have been underestimated in some situations. As a result, some spikes were detected by magnetometer only and other spikes were detected by gradiometer only. Therefore, complementary use of both sensors would be ideal in detecting mesial spikes. A couple of untouched issues remain in the present study. One is the type of gradiometer. From a theoretical perspective, axial gradiometers would seem likely to perform worse than a magnetometer, but better than a planar gradiometer for deep sources (Hämäläinen et al., 1993). Since the current study employed planar-type gradiometers, we cannot refer to the nature of axial-type gradiometers. Another important point is that the accuracy of source estimation was not evaluated in the present study. This factor can only be determined by simultaneous recording of invasive electrocorticography and MEG. In conclusion, the present study confirmed the clinical usefulness of magnetometers in detecting epileptiform discharges arising from

deep sources. Application of the magnetometer will strengthen the usefulness of MEG for spike detection. Complementary use of both sensors would be ideal in detecting medial spikes. These results suggest another possibility for investigating deep sources other than mesial spikes of epilepsy (e.g., activities of the hippocampus).

Acknowledgments Part of this study was presented at the 15th International Conference on Biomagnetism in 2006 (Biomag 2006). This study was supported by Grants-in-Aid for Scientific Research on Priority Areas (18020014) for HF from the Japan Ministry of Education, Culture, Sports, Science and Technology (MEXT) and a Scientific Research Grant (C2) 1859158 for NM from the Japan Society for the Promotion of Sciences.

References Assaf, B.A., Karkar, K.M., Laxer, K.D., Garcia, P.A., Austin, E.J., Barbaro, N.M., Aminoff, M.J., 2004. Magnetoencephalography source localization and surgical outcome in temporal lobe epilepsy. Clin. Neurophysiol. 115 (9), 2066–2076. Barkley, G.L., 2004. Controversies in neurophysiology MEG is superior to EEG in localization of interictal epileptiform activity. Pro. Clin. Neurophysiol 115 (5), 1001–1009. Barkley, G.L., Baumgartner, C., 2003. MEG and EEG in epilepsy. J. Clin. Neurophysiol 20 (3), 163–178. Bast, T., Oezkan, O., Rona, S., Stippich, C., Seitz, A., Rupp, A., Fauser, S., Zentner, J., Rating, D., Scherg, M., 2004. EEG and MEG source analysis of single and averaged interictal spikes reveals intrinsic epileptogenicity in focal cortical dysplasia. Epilepsia 45 (6), 621–631. Baumgartner, C., 2004. Controversies in clinical neurophysiology MEG is superior to EEG in the localization of interictal epileptiform activity. Con. Clin. Neurophysiol. 115 (5), 1010–1020. Baumgartner, C., Pataraia, E., Lindinger, G., Deecke, L., 2000a. Magnetoencephalography in focal epilepsy. Epilepsia 41 (Suppl 3), S39–S47. Baumgartner, C., Pataraia, E., Lindinger, G., Deecke, L., 2000b. Neuromagnetic recordings in temporal lobe epilepsy. J. Clin. Neurophysiol. 17 (2), 177–189. Burneo, J.G., Bebin, M., Kuzniecky, R.I., Knowlton, R.C., 2004. Electroclinical and magnetoencephalographic studies in epilepsy patients with polymicrogyria. Epilepsy Res. 62 (2–3), 125–133. Cobb, W.A. (Ed.), 1983. IFCN Recommendations for the Practice of Clinical Neurophysiology. Elsevier, Amsterdam. de Jongh, A., de Munck, J.C., Goncalves, S.I., Ossenblok, P., 2005. Differences in MEG/EEG epileptic spike yields explained by regional differences in signal-to-noise ratios. J. Clin. Neurophysiol. 22 (2), 153–158. Ebersole, J.S., 1999. Non-invasive pre-surgical evaluation with EEG/MEG source analysis. Electroencephalogr. Clin. Neurophysiol. Suppl. 50, 167–174. Ebersole, J.S., 1997a. EEG and MEG dipole source modeling. In: Engel, J., J., Pedley, T.A (Eds.), Epilepsy: A Comprehensive Textbook. LippincottRaven, Philadelphia, pp. 919–935. Ebersole, J.S., 1997b. Magnetoencephalography/magnetic source imaging in the assessment of patients with epilepsy. Epilepsia 38 (Suppl 4), S1–S5. Engel Jr., J., Van Ness, P., Rasmussen, T., Ojemann, L., 1993. Outcome with respect to epileptic seizures. In: Engel Jr., J. (Ed.), Surgical Treatment of the Epilepsies. Raven Press, New York, pp. 609–621. Fagaly, R.L., 1990. Neuromagnetic instrumentation. In: Sato, S. (Ed.), Advances in Neurology. Raven Press, New York, pp. 11–32. Fernandes, J.M., da Silva, A.M., Huiskamp, G., Velis, D.N., Manshanden, I., de Munck, J.C., da Silva, F.L., Cunha, J.P., 2005. What does an epileptiform spike look like in MEG? Comparison between coincident EEG and MEG spikes. J. Clin. Neurophysiol. 22 (1), 68–73.

R. Enatsu et al. / NeuroImage 41 (2008) 1206–1219 Fischer, M.J., Scheler, G., Stefan, H., 2005. Utilization of magnetoencephalography results to obtain favourable outcomes in epilepsy surgery. Brain 128 (Pt 1), 153–157. Iwasaki, M., Pestana, E., Burgess, R.C., Luders, H.O., Shamoto, H., Nakasato, N., 2005. Detection of epileptiform activity by human interpreters: blinded comparison between electroencephalography and magnetoencephalography. Epilepsia 46 (1), 59–68. Knake, S., Halgren, E., Shiraishi, H., Hara, K., Hamer, H.M., Grant, P.E., Carr, V.A., Foxe, D., Camposano, S., Busa, E., et al., 2006. The value of multichannel MEG and EEG in the presurgical evaluation of 70 epilepsy patients. Epilepsy Res 69 (1), 80–86. Knowlton, R.C., Elgavish, R., Howell, J., Blount, J., Burneo, J.G., Faught, E., Kankirawatana, P., Riley, K., Morawetz, R., Worthington, J., Kuzniecky, R.I., 2006. Magnetic source imaging versus intracranial electroencephalogram in epilepsy surgery: a prospective study. Ann. Neurol. 59 (5), 835–842. Knowlton, R.C., Shih, J., 2004. Magnetoencephalography in epilepsy. Epilepsia 45 (Suppl 4), 61–71. Landis, J.R., Koch, G.G., 1977. The measurement of observer agreement for categorical data. Biometrics 33 (1), 159–174. Leijten, F.S., Huiskamp, G.J., Hilgersom, I., Van Huffelen, A.C., 2003. High-resolution source imaging in mesiotemporal lobe epilepsy: a comparison between MEG and simultaneous EEG. J. Clin. Neurophysiol. 20 (4), 227–238. Lin, Y.Y., Shih, Y.H., Hsieh, J.C., Yu, H.Y., Yiu, C.H., Wong, T.T., Yeh, T.C., Kwan, S.Y., Ho, L.T., Yen, D.J., Wu, Z.A., Chang, M.S., 2003. Magnetoencephalographic yield of interictal spikes in temporal lobe epilepsy. Comparison with scalp EEG recordings. NeuroImage 19 (3), 1115–1126. Hamalainen, M., Hari, R., Iimoniemi, R., Knuutila, J., Lounasmaa, O., 1993. Magnetoencephalography-theory, instrumentation, and applications to noninvasive studies of the working human brain. Rev. Mod. Phys. 65, 413–497. Mayanagi, Y., Watanabe, E., Kaneko, Y., 1996. Mesial temporal lobe epilepsy: clinical features and seizure mechanism. Epilepsia 37 (Suppl 3), 57–60. Mikuni, N., Nagamine, T., Ikeda, A., Terada, K., Taki, W., Kimura, J., Kikuchi, H., Shibasaki, H., 1997. Simultaneous recording of epileptiform discharges by MEG and subdural electrodes in temporal lobe epilepsy. NeuroImage 5 (4 Pt 1), 298–306. Oishi, M., Otsubo, H., Kameyama, S., Morota, N., Masuda, H., Kitayama, M., Tanaka, R., 2002. Epileptic spikes: magnetoencephalography versus simultaneous electrocorticography. Epilepsia 43 (11), 1390–1395. Otsuki, T., 2004. Neuroimaging and presurgical evaluation of symptomatic epilepsies. Psychiatry Clin. Neurosci. 58 (3), S13–S15. Papanicolaou, A.C., Pataraia, E., Billingsley-Marshall, R., Castillo, E.M., Wheless, J.W., Swank, P., Breier, J.I., Sarkari, S., Simos, P.G., 2005. Toward the substitution of invasive electroencephalography in epilepsy surgery. J. Clin. Neurophysiol. 22 (4), 231–237. Park, H.M., Nakasato, N., Iwasaki, M., Shamoto, H., Tominaga, T., Yoshimoto, T., 2004. Comparison of magnetoencephalographic spikes with and without concurrent electroencephalographic spikes in extratemporal epilepsy. Tohoku J. Exp. Med. 203 (3), 165–174.

1219

Pataraia, E., Baumgartner, C., Lindinger, G., Deecke, L., 2002. Magnetoencephalography in presurgical epilepsy evaluation. Neurosurg. Rev. 25 (3), 141–159 discussion, 160-1. Pataraia, E., Simos, P.G., Castillo, E.M., Billingsley, R.L., Sarkari, S., Wheless, J.W., Maggio, V., Maggio, W., Baumgartner, J.E., Swank, P.R., Breier, J.I., Papanicolaou, A.C., 2004. Does magnetoencephalography add to scalp video-EEG as a diagnostic tool in epilepsy surgery? Neurology 62 (6), 943–948. Pataraia, E., Lindinger, G., Deecke, L., Mayer, D., Baumgartner, C., 2005. Combined MEG/EEG analysis of the interictal spike complex in mesial temporal lobe epilepsy. NeuroImage 24, 607–614. Quesney, L.F., Ortiz, T., 2004. MEG in the presurgical investigation of temporal lobe epilepsy. J Clin Neurophysiol 21 (2), 132. Ramantani, G., Boor, R., Paetau, R., Ille, N., Feneberg, R., Rupp, A., Boppel, T., Scherg, M., Rating, D., Bast, T., 2006. MEG versus EEG: influence of background activity on interictal spike detection. J. Clin. Neurophysiol. 23 (6), 498–508. Rodin, E., Funke, M., Berg, P., Matsuo, F., 2004. Magnetoencephalographic spikes not detected by conventional electroencephalography. Clin. Neurophysiol. 115 (9), 2041–2047. Sakamoto, S., Tsuyuguchi, N., Takami, T., Morino, M., Goto, T., Hattori, H., Tsutada, T., Haque, M., Sunada, I., Shimogawara, M., Hara, M., 2003. Interictal patterns of cerebral glucose metabolism, perfusion, and magnetic field in mesial temporal lobe epilepsy. Epilepsia 44 (9), 1196–1206. Shiraishi, H., Ahlfors, S.P., Stufflebeam, S.M., Takano, K., Okajima, M., Knake, S., Hatanaka, K., Kohsaka, S., Saitoh, S., Dale, A.M., Halgren, E., 2005. Application of magnetoencephalography in epilepsy patients with widespread spike or slow-wave activity. Epilepsia 46 (8), 1264–1272. Stefan, H., Schuler, P., Abraham-Fuchs, K., Schneider, S., Gebhardt, M., Neubauer, U., Hummel, C., Huk, W.J., Thierauf, P., 1994. Magnetic source localization and morphological changes in temporal lobe epilepsy: comparison of MEG/EEG, ECoG and volumetric MRI in presurgical evaluation of operated patients. Acta Neurol. Scand., Suppl. 152, 83–88. Tao, J.X., Ray, A., Hawes-Ebersole, S., Ebersole, J.S., 2005. Intracranial EEG substrates of scalp EEG interictal spikes. Epilepsia 46 (5), 669–676. Tarkiainen, A., Liljestrom, M., Seppa, M., Salmelin, R., 2003. The 3D topography of MEG source localization accuracy: effects of conductor model and noise. Clin. Neurophysiol. 114 (10), 1977–1992. Wennberg, R., 2006. Magnetic source imaging versus intracranial electroencephalogram: neocortical versus temporolimbic epilepsy surgery. Ann. Neurol. 60 (2), 271. Wheless, J.W., Willmore, L.J., Breier, J.I., Kataki, M., Smith, J.R., King, D.W., Meador, K.J., Park, Y.D., Loring, D.W., Clifton, G.L., Baumgartner, J., Thomas, A.B., Constantinou, J.E., Papanicolaou, A.C., 1999. A comparison of magnetoencephalography, MRI, and V-EEG in patients evaluated for epilepsy surgery. Epilepsia 40 (7), 931–941. Zijlmans, M., Huiskamp, G.M., Leijten, F.S., Van Der Meij, W.M., Wieneke, G., Van Huffelen, A.C., 2002. Modality-specific spike identification in simultaneous magnetoencephalography/electroencephalography: a methodological approach. J. Clin. Neurophysiol. 19 (3), 183–191.