Advertisement

Utility of magnetoencephalography combined with stereo-electroencephalography in resective epilepsy surgery: A 2-year follow-up

Open AccessPublished:March 17, 2022DOI:https://doi.org/10.1016/j.seizure.2022.03.013

      Highlights

      • Concordant localization between presurgical MEG and SEEG predicts better surgical outcome compared with inconsistent localization.
      • MEG dipole cluster helps SEEG implantation for completely sampling of MEG dipole cluster by SEEG electrodes would have a higher seizure-free rate.
      • MEG results characterized with a single, tight cluster, or stable orientation indicate a favorable resective surgery outcome.

      Abstract

      Purpose

      Precise and accurate implantation of stereo-electroencephalography (SEEG) electrodes is critical for the localization of the seizure onset zone (SOZ), which plays a leading role in the prognosis of resective epilepsy surgery. Magnetoencephalography (MEG) is a noninvasive technique which can delineate the epilepsy focus by visualizing interictal spikes into dipole clusters. MEG may provide supporting information for guiding SEEG electrode implantation and improve the long-term outcomes of epilepsy surgery. In this study, we evaluated the accuracy of MEG in determining the SOZ.

      Methods

      We retrospectively analyzed patients with refractory epilepsy who underwent MEG examination and SEEG implantation before resective epilepsy surgery in the Shanghai Ruijin Hospital. The SEEG plan was designed according to the dipole clusters and the resections were operated according to the SEEG recordings. We investigated the relationships of the pattern of MEG dipole clusters and SEEG sampling to the final resective surgery prognosis.

      Results

      We included 42 patients with a postoperative follow-up of at least 2 years (mean 34.1 months). Eighteen (42%) patients who showed concordant localization between MEG and SEEG evaluation had a higher probability of seizure-free outcome (p=0.046, χ2=4.835, odds ratio=5.00, 95% CI=1.12-22.30). Complete sampling of MEG dipole clusters by SEEG electrodes was found in 23 (54%) patients, who had higher probability of seizure-free outcome that those with incomplete sampling (p<0.001, odds ratio=16.67, 95% CI=3.11-89.28). MEG results showing a single, tight cluster or stable orientation were associated to better seizure outcomes after resective surgery.

      Conclusion

      MEG dipole cluster helps SEEG implantation in localizing the SOZ for better long-term epilepsy surgery outcome. The MEG results can play a role as prognostic predictors of epilepsy surgery.

      Keywords

      1. Introduction

      Epilepsy is one of the most common neurological diseases, imposing a heavy burden on patients and society, with about 70 million people estimated to be affected worldwide [
      • Ngugi AK
      • Bottomley C
      • Kleinschmidt I
      • Sander JW
      • Newton CR.
      Estimation of the burden of active and life-time epilepsy: a meta-analytic approach.
      ]. More significantly, in 30% of the patients, epilepsy is drug-resistant and cannot be effectively controlled by antiepileptic drugs (AEDs) [
      • Spencer S
      • Huh L
      Outcomes of epilepsy surgery in adults and children.
      ]. For such patients, surgical intervention is the main therapeutic option, and about 60% of the patients with drug-resistant focal epilepsy become seizure-free after resective epilepsy surgery [
      • Jobst BC
      • Cascino GD.
      Resective epilepsy surgery for drug-resistant focal epilepsy: a review.
      ]. Despite such success, there is still much room for improvement, especially regarding the precise localization of the seizure onset zone (SOZ) [
      • Lamberink HJ
      • Otte WM
      • Blümcke I
      • Braun KPJ
      • Aichholzer M
      • Amorim I
      • et al.
      Seizure outcome and use of antiepileptic drugs after epilepsy surgery according to histopathological diagnosis: a retrospective multicentre cohort study.
      ].
      The outcome of resective epilepsy surgery mainly depends on the successful localization of the SOZ [
      • Burkholder DB
      • Sulc V
      • Hoffman EM
      • Cascino GD
      • Britton JW
      • So EL
      • et al.
      Interictal scalp electroencephalography and intraoperative electrocorticography in magnetic resonance imaging-negative temporal lobe epilepsy surgery.
      ,
      • Khoo A
      • de Tisi J
      • Mannan S
      • O’Keeffe AG
      • Sander JW
      • Duncan JS.
      Reasons for not having epilepsy surgery.
      ]. Although invasive electroencephalographic techniques like stereo-electroencephalography (SEEG) are considered the gold standard in seizure-focus localization [
      • Agirre-Arrizubieta Z
      • Huiskamp GJ
      • Ferrier CH
      • van Huffelen AC
      • Leijten FS.
      Interictal magnetoencephalography and the irritative zone in the electrocorticogram.
      ], they are limited by factors such as the limited coverage per electrode, the higher surgical risk caused by the higher number of electrodes implanted, and their high cost [
      • Spencer DD
      • Gerrard JL
      • Zaveri HP.
      The roles of surgery and technology in understanding focal epilepsy and its comorbidities.
      ]. Therefore, improving noninvasive methods is still important for the progress of presurgical procedures, to obtain additional information and to improve the placement of SEEG implantation [
      • Anderson CT
      • Carlson CE
      • Li Z
      • Raghavan M
      Magnetoencephalography in the preoperative evaluation for epilepsy surgery.
      ,
      • Uribe San Martin R
      • Di Giacomo R
      • Mai R
      • Gozzo F
      • Pelliccia V
      • Mariani V
      • et al.
      Forecasting Seizure Freedom After Epilepsy Surgery Assessing Concordance Between Noninvasive and StereoEEG Findings.
      ] .
      Magnetoencephalography (MEG) is an important noninvasive method of evaluation characterized by the advantage of high resolution, both temporal and spatial, and offers whole-brain epileptogenic network information without the attenuation by skull conductance compared with EEG [
      • Zijlmans M
      • Zweiphenning W
      • van Klink N.
      Changing concepts in presurgical assessment for epilepsy surgery.
      ,
      • Rampp S
      • Stefan H
      • Wu X
      • Kaltenhäuser M
      • Maess B
      • Schmitt FC
      • et al.
      Magnetoencephalography for epileptic focus localization in a series of 1000 cases.
      ]. Magnetic Source Imaging (MSI) of interictal epileptic activity recorded by MEG has shown promising findings in the delineation of the epileptogenic focus and is performed using the equivalent current dipole modeling (ECD) technique [
      • Ntolkeras G
      • Tamilia E
      • AlHilani M
      • Bolton J
      • Ellen Grant P
      • Prabhu SP
      • et al.
      Presurgical accuracy of dipole clustering in MRI-negative pediatric patients with epilepsy: Validation against intracranial EEG and resection.
      ]. In this way the epileptogenic focus can be visualized as dipole results, containing position and orientation information, on the patients’ MRI images [
      • Kharkar S
      • Knowlton R.
      Magnetoencephalography in the presurgical evaluation of epilepsy.
      ].
      Previous studies have shown that MEG plays a positive role in other intracranial invasive evaluations, such as electrocorticography (ECoG) or depth EEG, and can predict surgical outcomes [
      • Agirre-Arrizubieta Z
      • Huiskamp GJ
      • Ferrier CH
      • van Huffelen AC
      • Leijten FS.
      Interictal magnetoencephalography and the irritative zone in the electrocorticogram.
      ,
      • Oishi M
      • Kameyama S
      • Masuda H
      • Tohyama J
      • Kanazawa O
      • Sasagawa M
      • et al.
      Single and multiple clusters of magnetoencephalographic dipoles in neocortical epilepsy: significance in characterizing the epileptogenic zone.
      ,
      • Almubarak S
      • Alexopoulos A
      • Von-Podewils F
      • Wang ZI
      • Kakisaka Y
      • Mosher JC
      • et al.
      The correlation of magnetoencephalography to intracranial EEG in localizing the epileptogenic zone: a study of the surgical resection outcome.
      ,
      • Schneider F
      • Alexopoulos AV
      • Wang Z
      • Almubarak S
      • Kakisaka Y
      • Jin K
      • et al.
      Magnetic source imaging in non-lesional neocortical epilepsy: additional value and comparison with ICEEG.
      ,
      • Tenney JR
      • Fujiwara H
      • Horn PS
      • Rose DF.
      Comparison of magnetic source estimation to intracranial EEG, resection area, and seizure outcome.
      ,
      • Agirre-Arrizubieta Z
      • Thai NJ
      • Valentín A
      • Furlong PL
      • Seri S
      • Selway RP
      • et al.
      The value of Magnetoencephalography to guide electrode implantation in epilepsy.
      ]. However, as the use of SEEG has recently become more common and has been shown to be associated with a higher probability of a seizure-free outcome [
      • Jehi L
      • Morita-Sherman M
      • Love TE
      • Bartolomei F
      • Bingaman W
      • Braun K
      • et al.
      Comparative Effectiveness of Stereotactic Electroencephalography Versus Subdural Grids in Epilepsy Surgery.
      ], few dataset including MEG data together with SEEG evaluation have been available for assessment [
      • Yu T
      • Ni D
      • Zhang X
      • Wang X
      • Qiao L
      • Zhou X
      • et al.
      The role of magnetoencephalography in the presurgical evaluation of patients with MRI-negative operculo-insular epilepsy.
      ,
      • Jung J
      • Bouet R
      • Delpuech C
      • Ryvlin P
      • Isnard J
      • Guenot M
      • et al.
      The value of magnetoencephalography for seizure-onset zone localization in magnetic resonance imaging-negative partial epilepsy.
      ]. A longer follow-up is essential to determine surgical outcomes, since patients who underwent resective epilepsy surgery seem to have a higher recurrence rate after a longer follow-up duration, but most previous studies are limited by a short follow-up time [
      • Jung J
      • Bouet R
      • Delpuech C
      • Ryvlin P
      • Isnard J
      • Guenot M
      • et al.
      The value of magnetoencephalography for seizure-onset zone localization in magnetic resonance imaging-negative partial epilepsy.
      ].
      We herein report a retrospective cohort study of 42 patients with drug-resistance epilepsy who underwent MEG, SEEG, and resective surgery at our Center between 2017 and 2019, with a mean seizure outcome follow-up of 34 months (minimum 2 years). We retrospectively evaluated the application of MEG in guiding SEEG implantation and localizing the SOZ. We hypothesized that the MEG dipole information can be concluded to predict the outcome of resective surgery, and could be help of SEEG electrode sampling to have a more precise SOZ localization, therefore improving the prognosis after SEEG evaluation.

      2. Materials and methods

      2.1 Participants

      We retrospectively examined 140 patients who underwent SEEG implantation followed by resective surgery at Ruijin Hospital, Shanghai Jiaotong University School of Medicine (Shanghai, China) between April 2017 and June 2019. Among them, 42 patients had presurgical MEG evaluation and adequate post-surgical follow-up.
      Patients were included if they: 1) underwent preoperative MEG and at least 7 spikes were detected during one hour of MEG recording for epilepsy diagnosis; 2) were diagnosed with refractory epilepsy, with persistent seizures after treatment with at least two antiepileptic drugs, and finished a series of presurgical evaluations including a comprehensive neurological evaluation, scalp video-EEG recording, and MRI; 3) underwent SEEG evaluation which recorded at least one seizure attack; 4) underwent surgical resection; 5) had postoperative imaging examination confirming that the seizure focus was resected as per surgical plan; 6) had more than 24 months of post-surgical follow-up. All patients, or their parents, provided written informed consent, according to the Declaration of Helsinki, to participate in the MEG/EEG recordings as part of the clinical dataset and to use their anonymized data for scientific study.

      2.2 MEG data acquisition and analysis

      The MEG recordings were performed with a 306-channel, whole-head VectorView MEG system (Elekta Oy, Helsinki, Finland) in a magnetically shielded room (Euroshield, Eura, Finland). Head position within the MEG system was determined by digitizing the position of the bilateral pre-auricular and the nasion fiducial points. During recording, the patient rested in a supine position for 1 hour. The raw MEG data were bandpass filtered in 0.03–330 Hz range and digitized at a sampling rate of 1000 Hz. The magnetic artifacts and movement artifacts were removed by the temporal extension of the Signal Space Separation method (tSSS) implemented in the MaxFilter software (Neuromag 3.4, Elekta Oy, Helsinki, Finland). tSSS used a 10 s raw data buffer with a subspace correlation limit of 0.98.
      Continuous MEG data were band-pass filtered to 6–40 Hz and visually checked for interictal epileptiform discharges (IEDs). Sharp signals (duration < 200ms) surpassing 150% of the background signal variance, perceived on several neighboring channels and producing clear dipolar magnetic field patterns, were considered as potential epileptic spikes. Spikes related to physiological rhythms or artifacts were rejected.
      Source localizations of epileptic events were obtained by conventional dipole modeling tools (Elekta Oy) using spherical conductor models determined from patients’ individual MRIs. Equivalent current dipoles (ECD) were fitted at the peak of epileptic spikes using a selection of the whole channels. The spherical head model was systematically used in all patients. Dipole fits were recognized valid when the goodness-of-fit was >80% and the 95%-confidence volume was less than 20 mm. ECDs were then overlayed on the patients’ MRIs which have been registered.

      2.3 Definition of cluster

      We defined a cluster as 5 or more dipoles located contiguously within neighboring gyri. Clusters were classified as tight or loose [
      • Murakami H
      • Wang ZI
      • Marashly A
      • Krishnan B
      • Prayson RA
      • Kakisaka Y
      • et al.
      Correlating magnetoencephalography to stereo-electroencephalography in patients undergoing epilepsy surgery.
      ]. A tight cluster is focal, that is, occurs within two neighboring gyri; while a loose cluster is distributed continuously over three or more gyri. We describe several clusters separated by a distance of more than one gyrus as multiple clusters, distributed across several brain sublobes of the same hemisphere or both hemispheres. If the dipoles were scattered unconsciously or the number of dipoles in neighboring gyri was less than 5, no cluster would be recorded. Clusters were classified depending on the orientation of the dipoles. When at least 80% of the dipoles in a cluster had uniform and similar orientation, the cluster would be classified as stable. Otherwise, if any cluster has a variable or multiform orientation it would be regarded as a variable cluster. Fig. 1 shows examples of different types of clusters based on these definitions.
      Fig. 1
      Fig. 1Example of different types of MEG cluster.
      (A) patient with a single tight cluster with stable orientation; (B) patients with a single tight cluster with stable orientation (the scattered dipoles in the left temporal lobe doesn't meet the criteria of a cluster that contains at least 5 dipoles in neighboring gyri); (C) patients with a single loose cluster with variable orientation; (D) patient with multiple tight clusters with variable orientation; (E) patient with a single loose cluster with stable orientation; (F) patient with a single tight cluster with variable orientation.

      2.4 Concordance analysis

      Spatial concordance between MEG clusters and SEEG findings was assessed at the sublobar region level. If the MEG cluster and SEEG contacts indicating the SOZ were completely located in the same sublobar region, they would be regarded as completely concordant. If the delineated locations almost matched, but MEG or SEEG showed extended brain areas, or only areas of overlap existed, they would be regarded as partially concordant. An illustration of concordance is shown in Supplementary Figure S1. If the MEG occupied a larger (smaller) brain area compared to SEEG, it would be classified into the partial-plus (partial-minus) subgroup. In case of no overlap between MEG clusters and SEEG results, the results would be regarded as inconsistent. Spatial concordance analysis between SEEG and other presurgical evaluations, such as video-EEG, MRI, and positron emission tomography/magnetic resonance (PET/MR) was also conducted. The concordance between SEEG and video-EEG was evaluated at the lobar level. The concordance classification between SEEG and PET/MR was the same used in our previous study [
      • Zhang M
      • Liu W
      • Huang P
      • Lin X
      • Huang X
      • Meng H
      • et al.
      Utility of hybrid PET/MRI multiparametric imaging in navigating SEEG placement in refractory epilepsy.
      ].

      2.5 Presurgical evaluation

      The SEEG surgery plan was different for every patient, depending on MEG results, symptomatic manifestations, neurological evaluation, scalp video-EEG, diagnostic MR, and PET/MR.
      The symptomatic manifestations, neurological evaluation concluded details about the patients’ epilepsy history, seizure semiology, and any other potentially useful type of clinical information, including gender, age, cognitive level, surgical history, and medication history. PET scanning was performed on a 3T Siemens PET/MR scanner and FDG PET/MR acquisition was started 40 min after 18F-FDG was injected; PET and MRI images were fused after simultaneous PET/MRI scanning. Structural MR scans were performed on a 3T GE or PHILIPS scanner, and three dimensional T1-MPRAGE and FLAIR sequence followed by two dimensional T2, T2 Flair, and SWI were performed. Long-term scalp video-EEG was performed with the monitoring time ranging from 3 to 7 days to capture habitual ictal events. The electrode placement of scalp video-EEG obeyed the international 10-20 system, with sphenoidal electrodes placed if appropriate. The exact SEEG numbers and the position to implant SEEG electrode were finally based on the conclusions of a multi-disciplinary team (MDT) including epileptologists, neurosurgeons, neuroradiologists, and other practitioners.

      2.6 SEEG recordings

      After the presurgical evaluation by an MDT led to the indication of the possible position of the epileptic focus, the SEEG electrodes were implanted. The surgery plans were performed in the Leksell Surgiplan software (Elekta Inc., Stockholm, Sweden) using co-registered CT and MRI. The CT scan was performed after a stereotactic frame (Elekta Inc., Stockholm, Sweden) was placed, and MRI was required to contain MEG information. All patients underwent a post-surgery CT scan to confirm the localization of each contact of the SEEG electrodes. The mean number of SEEG electrodes implanted ranged from 2 to 10, each electrode containing 8 or 16 contacts. (SDE-08: S8 and S16, Beijing Sinovation Medical Technology CO., LTD, Beijing, China). Intracranial ictal and interictal SEEG data were recorded for 72–240 h (3–10 days), requiring the recording of a minimum of 3 seizures in each patient. A positive SEEG result indicated that sufficient evidence was available to define the SOZ and a radiofrequency lesion ablation may be performed if appropriate before the intracranial electrodes were removed. For patients diagnosed with temporal lobe epilepsy (TLE), resection surgery was performed using a standardized protocol of anterior temporal and hippocampus lobectomy immediately after the intracranial electrodes were removed. Fig. 2 shows a representative case included in our study.
      Fig. 2
      Fig. 2MEG navigate SEEG implanting and postsurgical follow-up review
      A female patient, 34-year-old, with a history of seizure onset at 14 A and D show the results of MEG. According to our classification of the cluster, It's a single tight cluster with variable orientation B shows the SEEG implantation surgery plan. With the guidance of MEG dipoles (white dote) and other presurgical evaluation methods, 6 SEEG electrodes were implanted monitoring several possible SOZ including the right hippocampus, amygdala, and lateral temporal lobe. Temporal lobectomy was conducted by the results of SEEG recording C shows the postoperative brain MRI at the 6-month follow-up visit E was postoperative MR image co-registered with presurgical SEEG implantation plan to confirm the resected seizure focus. In F postoperative MR co-registered with CT scan contains SEEG location information in case electrode deviated from planned location. This patient had a concordant result between SEEG and MEG results, and SEEG completely sampled MEG the dipole cluster. At the 29-months follow-up visit, this patient achieved a favorable clinical outcome (Engel Class I).

      2.7 Postsurgical outcome

      Postsurgical seizure outcome was rated according to Engel's classification, evaluated by a neurologist [
      • Engel J
      • Van Ness P
      • Rasmussen T.
      ]. We regarded a seizure-free outcome as Engel's class I and not seizure-free as classes II-IV. The follow-up time ranged 24 to 53 months, with a median follow up time of 34.07 months.

      2.8 Statistical analysis

      Statistical analyses were performed using Statistical Package for Social Sciences (version 25.0; SPSS, IBM, Chicago, IL, USA). Chi-square test were conducted to identify the relationship between parameters and prognosis; Fisher's exact test or Continuous correction were applied when needed; Kolmogorov-Smirnov test for normality was conducted for continuous variables, an independent sample T-test or Mann-Whitney U-test was performed to assess the association with outcomes. Continuous variables were expressed as mean ± standard deviation (SD), and categorical data were shown as frequency and percentages. A p-value of less than 0.05 was considered statistically significant and Bonferroni Correction was applied when needed.

      2.9 Data availability

      Anonymized evaluation data are available upon request.

      3. Results

      3.1 Demographic and clinical characteristics

      The main demographic and clinical data of the 42 patients are shown in Table 1. Twenty-six patients (63%) were seizure-free at a follow-up of more than 24 months. Age, epilepsy duration, epilepsy onset age, seizure frequency, lateral of SOZ, MRI findings (positive or negative), pathology results, and follow-up duration had no significant association with epilepsy surgery outcomes. The mean age at the time of surgery was 28 years (median 27, standard deviation [SD] 11.35, range 9-53), and the mean duration of epilepsy was 12 years (median 10, SD 9.17, range 1-34). Twenty-two patients (51%) were female. There were 23 (54%) patients with TLE, 7 (16.7%) with frontal lobe epilepsy, 1 with parietal lobe epilepsy, 2 with insula lobe epilepsy, and nine (21.4%) with multi-lobe resections.
      Table 1Demographic data of the total epilepsy patients.
      Patients’ demographicsTotal patientsEngel Class IEngel Class II-IVp-value
      Number of subjects, n4226(61.90)16(38.10)NA
      Gender0.010
      Male, n (%)208(40)12(60)NA
      Female, n (%)2218(82)4(18)NA
      Age at surgery, y, mean (SD)27.54(11.35)28.12(11.66)26.63(10.76)0.688
      Age at epilepsy onset, mean (SD)15.67(11.17)17.08(11.90)14.43(8.75)0.305
      Duration of epilepsy, y, mean (SD)12.02(9.17)11.52(8.98)13.39(9.00)0.357
      Seizure frequency (SD, per year)126.63(187.41)109.96(173.43)155.21(206.09)0.124
      MRI findings0.485
      MRI negative, n (%)11(26.19)8(72.72)3(27.27)NA
      Others, n (%)31(73.80)18(58.06)13(41.94)NA
      Side of epilepsy SOZ0.597
      Left-sided, n (%)17(40.48)9(52.94)8(47.05)NA
      Right sided, n (%)24(57.14)16(66.67)8(33.33)NA
      Location of epilepsy SOZ0.582
      TLE, n (%)23(54.76)14(60.87)9(39.13)NA
      ETLE, n (%)19(45.24)12(63.16)7(36.84)NA
      Number of implanted SEEG, n(SD)6.42(1.93)6.35(1.88)6.56(2.00)0.829
      Follow up duration, mom, mean (SD)34.07(8.41)33.07(8.03)35.68(8.74)0.452
      Pathology/FCD, n (%)14(40)10(71.43)4(28.57)0.284
      (Abbreviations: SOZ = seizure onset zone; TLE = temporal lobe epilepsy; ETLE = extratemporal epilepsy; FCD = focal cortical dysplasia. The age of the subjects, age at epilepsy onset, duration of epilepsy, seizure frequency, the number of implanted SEEG and follow-up duration were reported as the mean ± standard deviation (SD). For continuous variables of the age at surgery and age at seizure onset between the Engel Class I and Engel Class II-IV groups, independent-sample t tests analysis was conducted. For the seizure frequency, epilepsy duration, number of implanted SEEG and follow up duration, Mann-Whitney U-tests were conducted. For categorical variables, χ2 tests were conducted).
      The mean follow-up period was 34.07 months (median 32, SD 8.41, range 24–53). Twenty-six patients (including 14 with TLE) who met the criteria of Engel class I were classified as being seizure-free. Sixteen patients (including 9 TLE) who met the criteria for Engel Class II, III, or IV were classified as not being seizure-free. The majority of MEG studies showed one or two clusters. The mean implanted SEEG electrodes number was 6.42 (median 6, SD 1.93, range 3-10). As shown in Supplementary Table S1, 11 patients (26.2%) had no significant abnormality on visual inspection of the presurgical MRI. Nine patients (23%) had hippocampal sclerosis and abnormally high hippocampus signal on the T2 FLAIR sequence was found in 3 patients. Encephalomalacia foci could be found in 5 patients (12%) and 16 patients (38.1%) had other MRI abnormalities, regardless of their association with epilepsy. Twenty of 35 patients (57.14%) had pathological abnormalities. Fourteen patients (40%) were identified by pathology to be focal cortical dysplasia (FCD; 4 patients with International League Against Epilepsy [ILAE] type Ⅰa, 2 with type b, 4 with type Ⅱa, 1 with type Ⅱb and 3 with Ⅲa)), and hippocampus sclerosis was identified in 5 patients. Fifteen patients (42.86%) had negative pathological result, including 10 with gliosis and 5 without any pathological abnormality. There was no statistically significant difference between FCD-positive and -negative patients in terms of surgical outcome (p=0.284).

      3.2 Concordance between MEG findings and SEEG results

      Forty-two patients were analyzed to estimate the accuracy and reliability of MEG results. Fig. 3 shows an example of a partially concordant result. As shown in Table 2, after surgery 26 out of 42 patients (62%) were seizure-free (Engel Class I). Among patients for whom SEEG findings indicated the SOZ to be in complete concordance with the MEG results, 15 out of 18 (83%) became seizure-free. In contrast, when the two findings were only partially concordant, seizure-free status was achieved in 11 out of 22 patients (50%). Two patients had completely inconsistent MEG and SEEG findings, and neither of them had a seizure-free outcome. These results suggest that patients with complete concordance between MEG and SEEG results have a much higher probability of a seizure-free outcome compared with those with only partial concordance. (P=0.046, χ2=4.835, odds ratio=5.00, 95% confidence interval [CI]=1.12-22.30). As shown in Supplementary Table S2, the volume of the SOZ delineated by MEG and SEEG results was compared when the concordance result was classified as partially concordant, but no significant difference was found between the partial-plus (M>S) group and the overlap-only subgroup (p=0.057).
      Fig. 3
      Fig. 3Example of a partially concordant result.
      A male juvenile patient, 14-year old, with a history of seizure onset at 2 and 8 SEEG electrodes were implanted, Fig A and B show the result of MEG and the planned SEEG electrode pathway which would completely sample all MEG clusters. Fig C shows the ictal SEEG monitoring data. This patient had a partially concordant result for the MEG demonstrated the SOZ located in the left temporal and frontal lobe, but Ictal SEEG data shows seizures originating from the left temporal lobe.
      Table 2The results of SEEG and MEG concordance and different cluster classifications.
      Engel Class
      Concordance of SEEG and MEGTotalClass IClass II-IVSeizure-free rate, %p-valueχ2
      Completely concordant1815383.330.0464.835
      Partially concordant22111150Reference
      Inconsistent2020
      Single/multiple clusterEngel Class
      TotalClass IClass II-IVSeizure-free rate, %p-valueχ2
      Single2421387.50.00110.904
      Multiple1861233.33Reference
      Tight/loose clusterEngel Class
      TotalClass IClass II-IVSeizure-free rate, %p-valueχ2
      Tight1715288.230.0088.396
      Loose25111444Reference
      Stable/variable clusterEngel Class
      TotalClass IClass II-IVSeizure-free rate, %p-value
      Stable2419579.170.007
      Variable113827.27Reference
      2 tests were conducted and p-value were generated by Fisher's exact tests when needed. In the part of concordance of MEG and SEEG, statistical analysis was performed between complete concordant subgroup and partial concordant subgroup)
      As shown in Supplementary Table S3, 24 patients were diagnosed with TLE. All the 8 patients with completely consistent results had a favorable outcome, compared with 7 out of 15 patients with partially consistent results (p=0.019). However, there was no significant difference when compared within the extra-temporal lobe epilepsy subgroup (p=0.644). The spatial concordance analysis results between SEEG and video-EEG, MRI, and PET/MR are shown in supplementary Table S4. Comparisons associated with surgical outcomes were conducted separately between the concordant group and the partially concordant group, but no statistically significant differences were found.

      3.3 SEEG sampling of MEG clusters and outcome

      Sampling analysis was performed on 37 patients, as five patients were excluded because the measured SEEG electrode's location assessed by post-surgery CT deviated from planned location, or due to MEG results not meeting the criteria defining a cluster, namely 5 or more dipoles located contiguously within neighboring gyri. Supplementary Figure S1 shows examples of patients whose SEEG sampling results were classified into different classes. As shown in Table 3, all MEG clusters were completely sampled in 23 patients, 20 of whom (86.96%) were free of disabling seizures (Engel I). For comparison, in 14 patients with incomplete MEG cluster sampling, only four became seizure-free. Thus, patients for whom a complete MEG cluster sampling could be conducted had significantly higher probability of a seizure-free outcome than those with MEG clusters that were only incompletely sampled by SEEG (p=0.0008, odds ratio=16.67, 95% CI=3.11-89.28). The mean number of SEEG electrodes did not significantly differ between these two groups (6.40 and 6.23, respectively).
      Table 3The results of surgery outcome in different cluster sample condition.
      Engel Class
      TotalClass IClass II-IVp-valueMean SEEG electrodes
      Complete sample23203<0.0016.40
      Incomplete sample14410Reference6.23
      2 tests were conducted and p-value were generated by Fisher's exact tests.)

      3.4 Predictive value of MEG clusters in surgically treated patients

      The characteristics of enrolled patients are shown in Table S1. As shown in Table 2, these patients were divided into two groups, namely the single cluster group (24 patients) and the multiple clusters group (18 patients). Twenty-one patients (87.5%) in the single cluster group reached the criteria of Engel I Class, while only 6 patients (33.33%) with multiple clusters became seizure-free. Thus, patients with a single cluster had greater odds of becoming seizure-free (p=0.0014, χ2=10.904, odds ratio=10.00, 95% CI=2.34-42.78). One or more loose clusters were found in 25 patients, while in 17 patients only showed tight clusters. Out of 25 patients who had one or more loose clusters, 11 (44%) patients became seizure-free, while out of 17 patients who had only tight clusters, 15 (88.23%) became seizure-free. Thus, patients with only tight clusters were significantly more likely to become seizure-free than those with loose clusters (p=0.008, χ2=8.396, odds ratio=9.55, 95% CI=1.79-50.88)
      The clusters of 24 patients showed stable orientation of the dipole source. Of these patients, 19 (79.17%) had seizure-free outcome. Eleven patients were found to have variable orientation clusters, and only 3 of them (27.27%) had a seizure-free outcome. Thus, patients with variable cluster orientation were less likely to become seizure-free than those with only stable orientation clusters (p=0.0069, odds ratio=10.13, 95% CI=1.94-52.9).
      Taken together, these results demonstrate that patients for whom MEG results showed a single cluster, tight clusters, or a stable dipole orientation had significantly better seizure outcomes (Bonferroni correction, p<0.0167).

      4. Discussion

      In this study we report a retrospective investigation of the second largest cohort of patients to date having received MEG, SEEG, and surgery, with a longer follow-up than in previous studies. We compared the MEG dipole mapping results with SEEG results and their association with long-term seizure outcome in 42 patients who underwent MEG and SEEG surgery prior to resective epilepsy surgery. Our study demonstrated three main findings: 1) concordant localization between presurgical MEG and SEEG predicts better surgical outcome compared with inconsistent localization; 2) patients for whom SEEG electrodes completely sampled the MEG clusters had a more favorable outcome compared with those with incomplete sampling; 3) MEG cluster results with specific features, such as a single cluster, tight clusters, and stable orientation were associated with better prognosis after SEEG implantation and resective surgery.

      4.1 Advantage of MEG in guiding SEEG implantation

      As the “gold standard” in epilepsy surgery, precise and accurate implantation of SEEG electrodes is critical to locate the SOZ. However, the current presurgical technologies widely used in clinical practice to guide SEEG implantation when planning surgery can only provide offered an ambiguous localization [
      • Minotti L
      • Montavont A
      • Scholly J
      • Tyvaert L
      • Taussig D.
      Indications and limits of stereoelectroencephalography (SEEG).
      ]. Previous studies have indicated the value of MEG to locate the SOZ with higher sensitivity and spatial resolution compared with other common presurgical evaluation methods [
      • Burgess RC.
      MEG for Greater Sensitivity and More Precise Localization in Epilepsy.
      ]. In our study, we imported presurgical MEG/MRI images into the SEEG positioning system and fused the images with postsurgical CT images containing SEEG electrodes’ location information. In this way, the judgment of the actual SEEG electrodes sampling position could be judged more accurately. Among 23 patients with complete sampling of MEG clusters with SEEG electrodes, 20 were free of disabling seizures after surgery, while patients with incomplete SEEG electrode sampling had a significantly smaller probability of being seizure-free. Few previous studies have considered how MEG dipole information can be used in guiding SEEG implantation at the cluster level, since it is unrealistic to cover all of the MEG dipole abnormal information, given that many patients have scattered dipoles; moreover, the use of SEEG electrodes, whose number is limited, is also influenced by other presurgical evaluations, such as PET. On the other hand, the definition of a dipole cluster is inconvenient in clinical practice and is not uniform across previous studies. Our study adopted a realistic and clear definition, with favorable surgery outcomes, that seems suitable for clinical practice. Previous studies have proved the value of MEG in guiding other intracranial electrodes, such as ECoG, and demonstrated the superior performance of MEG in guiding electrode placement compared with other noninvasive examinations. In this study, we demonstrated that MEG has important value also in guiding SEEG surgery, and can provide unique information during presurgical evaluation.

      4.2 The concordance of MEG and SEEG localization predicts the resection outcomes

      The fact that patients tend to have more favorable epilepsy surgery outcomes if they have concordant results between MEG and other invasive intracranial EEG is widely accepted and has been confirmed by many previous studies [
      • Tenney JR
      • Fujiwara H
      • Horn PS
      • Rose DF.
      Comparison of magnetic source estimation to intracranial EEG, resection area, and seizure outcome.
      ,
      • He X
      • Zhou J
      • Teng P
      • Wang X
      • Guan Y
      • Zhai F
      • et al.
      The impact of MEG results on surgical outcomes in patients with drug-resistant epilepsy associated with focal encephalomalacia: a single-center experience.
      ]. Our study further demonstrated the same phenomenon concerning MEG and SEEG, and provided further insight about their relationships. All the enrolled patients underwent postoperative imaging to make sure the resective surgery had been performed as per the surgery plan. This aspect could have introduced an additional source of variance to the results of some studies that did not verify the resected area, given that, in clinical practice, intervention surgery cannot always be performed as planned.
      In practice, the propagation of spikes may affect the concordance, and the principle of resection of the effective brain areas may limit the outcome. For example, in cases #5 and #29, shown in Table S1, MEG detected 2 clusters in different regions, but the SEEG recordings demonstrated that, one region was confirmed as the ictal zone while the other region was the propagated zone. Thus, the ictal zone was resected and the patients get a release from the attack. Indeed, MEG records both the SOZ and the irritative zone. With the SEEG ictal recordings, the SOZ can be distinguished from the other interictal spikes [
      • Kim D
      • Joo EY
      • Seo DW
      • Kim MY
      • Lee YH
      • Kwon HC
      • et al.
      Accuracy of MEG in localizing irritative zone and seizure onset zone: Quantitative comparison between MEG and intracranial EEG.
      ]. In case#7, the SEEG detected two ictal zones in two hemispheres, and the neurosurgeons chose to operate one side, which lead to the remain of attack. In case24#, MEG detected the dipole cluster in the right temporal, but from the SEEG ictal recordings showed that the seizure initiated from the left temporal, and the left temporal was removed according to the SEEG results. The attack remains even though the frequency is much less than the preoperative period. We speculate that this was a case of bilateral temporal epilepsy. Due to the sclerosis, the left temporal has fewer neurons to generate enough oscillation to be detected by MEG, in contrast with the right temporal. But with the precise SEEG recording with a high temporal resolution, the ictal spike could be caught before it transports to the right temporal. On the other hand, we might not record long enough to catch the seizure from the right temporal. In this case, the neuromodulation might be a better choice.

      4.3 The classification of MEG clusters reveals affected brain areas

      Many previous studies have reported the associations between different MEG clusters characteristics and surgery outcomes. Each study interpreted this phenomenon in their own way ([
      • Oishi M
      • Kameyama S
      • Masuda H
      • Tohyama J
      • Kanazawa O
      • Sasagawa M
      • et al.
      Single and multiple clusters of magnetoencephalographic dipoles in neocortical epilepsy: significance in characterizing the epileptogenic zone.
      ,
      • RamachandranNair R
      • Otsubo H
      • Shroff MM
      • Ochi A
      • Weiss SK
      • Rutka JT
      • et al.
      MEG predicts outcome following surgery for intractable epilepsy in children with normal or nonfocal MRI findings.
      ,
      • Iida K
      • Otsubo H
      • Matsumoto Y
      • Ochi A
      • Oishi M
      • Holowka S
      • et al.
      Characterizing magnetic spike sources by using magnetoencephalography-guided neuronavigation in epilepsy surgery in pediatric patients.
      ]) [
      • Shirozu H
      • Hashizume A
      • Masuda H
      • Kakita A
      • Otsubo H
      • Kameyama S.
      Surgical strategy for focal cortical dysplasia based on the analysis of the spike onset and peak zones on magnetoencephalography.
      ,
      • Wang S
      • Tang Y
      • Aung T
      • Chen C
      • Katagiri M
      • Jones SE
      • et al.
      Multimodal noninvasive evaluation in MRI-negative operculoinsular epilepsy.
      ,
      • Vadera S
      • Jehi L
      • Burgess RC
      • Shea K
      • Alexopoulos AV
      • Mosher J
      • et al.
      Correlation between magnetoencephalography-based "clusterectomy" and postoperative seizure freedom.
      ], and the definition of a MEG cluster varied between studies and was difficult to generalize to clinical practice, until Hiroshima Murakami proposed a practical definition [
      • Murakami H
      • Wang ZI
      • Marashly A
      • Krishnan B
      • Prayson RA
      • Kakisaka Y
      • et al.
      Correlating magnetoencephalography to stereo-electroencephalography in patients undergoing epilepsy surgery.
      ], which appears likely to become a consensus definition MEG clusters thanks to its scientific classification meaning and clinical applicability. In our study, we followed this definition and slightly modified adjusting it to practical situations.
      In our study, multiple, loose MEG dipole clusters or their variable orientation may predict poor outcome and indicate a more comprehensive presurgical evaluation and a larger resection area. The affected brain areas generating the seizures determine the characteristics of the MEG clusters. A single, tight MEG cluster normally indicates that the epileptogenic focus is a focal zone which prone to be removed entirely [
      • Oishi M
      • Kameyama S
      • Masuda H
      • Tohyama J
      • Kanazawa O
      • Sasagawa M
      • et al.
      Single and multiple clusters of magnetoencephalographic dipoles in neocortical epilepsy: significance in characterizing the epileptogenic zone.
      ,
      • Stefan H
      • Wu X
      • Buchfelder M
      • Rampp S
      • Kasper B
      • Hopfengartner R
      • et al.
      MEG in frontal lobe epilepsies: localization and postoperative outcome.
      ]. A single loose cluster in MEG implies that the epileptogenic focus is expanded to the peripheral areas of the focal zone. Multiple tight clusters could represent two or more focal epileptogenic zones, and their resection may need two or more operations. The etiology of the patients with multiple loose clusters in MEG is general encephalitis. The orientation of dipole implies the arrangement of the affected neural tissue. Theoretically, a focal zone located in a single sulcus should present stable orientation.
      The accuracy of the position of the dipole in MRI is related to the depth and the power of the seizure source. A dipole from the superficial source with sharp power can be localized to its precise position in the brain area on MRI. On the contrary, a depth source from the medial temporal is generally localized to the lateral temporal, a few millimeters from the hippocampus and the amygdala.

      4.4 Limitations

      The most notable limitation of the present study is its retrospective nature. The final surgery plan and the SEEG implant area are not only determined by MEG results, but are also influenced by other examinations like PET, video-EEG, and MRI. We could not directly access the effect that MEG or other examinations had on the overall surgical process. For some of patients in our study, the epileptogenic relevance may need to be discussed as the limited SEEG electrodes (less than 6) implanted. As a retrospective study, confounding variables associated with surgical selection bias could not be eliminated.
      Moreover, the sample of this study was relatively limited (n=42) and the patients were enrolled from a single medical center, which may potentially cause bias in the statistical analysis.

      5. Conclusion

      MEG results can help improve SEEG electrode placement, and a MEG abnormal area completely sampled by SEEG predicts better outcome after resective surgery. Patients with concordant SOZ localization on MEG and SEEG had more favorable outcomes. A single, tight cluster or with a stable orientation on MEG were also predictive of better outcomes after resective surgery.

      Author contributions

      The article was written by Chao Zhang, Chunyan Cao and Shikun Zhan have provided guidance to the manuscript preparation. Wei Liu, Xiaoxiao Zhang, and Peng Huang performed SEEG and resection surgeries. Jing Zhang performed the MEG evaluation and follow-up. All authors have approved the final version of the editorial.

      Ethical publication statement

      We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

      Conflicts of Interest

      All authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as being potential conflicts of interest.

      Acknowledgments

      The study was supported by the National Natural Science Foundation of China (NSFC) No. 82071547 .

      Appendix. Supplementary materials

      References

        • Ngugi AK
        • Bottomley C
        • Kleinschmidt I
        • Sander JW
        • Newton CR.
        Estimation of the burden of active and life-time epilepsy: a meta-analytic approach.
        Epilepsia. 2010; 51: 883-890
        • Spencer S
        • Huh L
        Outcomes of epilepsy surgery in adults and children.
        Lancet Neurol. 2008; 7: 525-537
        • Jobst BC
        • Cascino GD.
        Resective epilepsy surgery for drug-resistant focal epilepsy: a review.
        JAMA. 2015; 313: 285-293
        • Lamberink HJ
        • Otte WM
        • Blümcke I
        • Braun KPJ
        • Aichholzer M
        • Amorim I
        • et al.
        Seizure outcome and use of antiepileptic drugs after epilepsy surgery according to histopathological diagnosis: a retrospective multicentre cohort study.
        Lancet Neurol. 2020; 19: 748-757
        • Burkholder DB
        • Sulc V
        • Hoffman EM
        • Cascino GD
        • Britton JW
        • So EL
        • et al.
        Interictal scalp electroencephalography and intraoperative electrocorticography in magnetic resonance imaging-negative temporal lobe epilepsy surgery.
        JAMA Neurol. 2014; 71: 702-709
        • Khoo A
        • de Tisi J
        • Mannan S
        • O’Keeffe AG
        • Sander JW
        • Duncan JS.
        Reasons for not having epilepsy surgery.
        Epilepsia. 2021;
        • Agirre-Arrizubieta Z
        • Huiskamp GJ
        • Ferrier CH
        • van Huffelen AC
        • Leijten FS.
        Interictal magnetoencephalography and the irritative zone in the electrocorticogram.
        Brain. 2009; 132: 3060-3071
        • Spencer DD
        • Gerrard JL
        • Zaveri HP.
        The roles of surgery and technology in understanding focal epilepsy and its comorbidities.
        Lancet Neurol. 2018; 17: 373-382
        • Anderson CT
        • Carlson CE
        • Li Z
        • Raghavan M
        Magnetoencephalography in the preoperative evaluation for epilepsy surgery.
        Curr Neurol Neurosci Rep. 2014; 14: 446
        • Uribe San Martin R
        • Di Giacomo R
        • Mai R
        • Gozzo F
        • Pelliccia V
        • Mariani V
        • et al.
        Forecasting Seizure Freedom After Epilepsy Surgery Assessing Concordance Between Noninvasive and StereoEEG Findings.
        Neurosurgery. 2020; 88: 113-121
        • Zijlmans M
        • Zweiphenning W
        • van Klink N.
        Changing concepts in presurgical assessment for epilepsy surgery.
        Nat Rev Neurol. 2019; 15: 594-606
        • Rampp S
        • Stefan H
        • Wu X
        • Kaltenhäuser M
        • Maess B
        • Schmitt FC
        • et al.
        Magnetoencephalography for epileptic focus localization in a series of 1000 cases.
        Brain. 2019; 142: 3059-3071
        • Ntolkeras G
        • Tamilia E
        • AlHilani M
        • Bolton J
        • Ellen Grant P
        • Prabhu SP
        • et al.
        Presurgical accuracy of dipole clustering in MRI-negative pediatric patients with epilepsy: Validation against intracranial EEG and resection.
        Clin Neurophysiol. 2021;
        • Kharkar S
        • Knowlton R.
        Magnetoencephalography in the presurgical evaluation of epilepsy.
        Epilepsy Behav. 2015; 46: 19-26
        • Oishi M
        • Kameyama S
        • Masuda H
        • Tohyama J
        • Kanazawa O
        • Sasagawa M
        • et al.
        Single and multiple clusters of magnetoencephalographic dipoles in neocortical epilepsy: significance in characterizing the epileptogenic zone.
        Epilepsia. 2006; 47: 355-364
        • Almubarak S
        • Alexopoulos A
        • Von-Podewils F
        • Wang ZI
        • Kakisaka Y
        • Mosher JC
        • et al.
        The correlation of magnetoencephalography to intracranial EEG in localizing the epileptogenic zone: a study of the surgical resection outcome.
        Epilepsy Res. 2014; 108: 1581-1590
        • Schneider F
        • Alexopoulos AV
        • Wang Z
        • Almubarak S
        • Kakisaka Y
        • Jin K
        • et al.
        Magnetic source imaging in non-lesional neocortical epilepsy: additional value and comparison with ICEEG.
        Epilepsy Behav. 2012; 24: 234-240
        • Tenney JR
        • Fujiwara H
        • Horn PS
        • Rose DF.
        Comparison of magnetic source estimation to intracranial EEG, resection area, and seizure outcome.
        Epilepsia. 2014; 55: 1854-1863
        • Agirre-Arrizubieta Z
        • Thai NJ
        • Valentín A
        • Furlong PL
        • Seri S
        • Selway RP
        • et al.
        The value of Magnetoencephalography to guide electrode implantation in epilepsy.
        Brain Topogr. 2014; 27: 197-207
        • Jehi L
        • Morita-Sherman M
        • Love TE
        • Bartolomei F
        • Bingaman W
        • Braun K
        • et al.
        Comparative Effectiveness of Stereotactic Electroencephalography Versus Subdural Grids in Epilepsy Surgery.
        Ann Neurol. 2021;
        • Yu T
        • Ni D
        • Zhang X
        • Wang X
        • Qiao L
        • Zhou X
        • et al.
        The role of magnetoencephalography in the presurgical evaluation of patients with MRI-negative operculo-insular epilepsy.
        Seizure. 2018; 61: 104-110
        • Jung J
        • Bouet R
        • Delpuech C
        • Ryvlin P
        • Isnard J
        • Guenot M
        • et al.
        The value of magnetoencephalography for seizure-onset zone localization in magnetic resonance imaging-negative partial epilepsy.
        Brain. 2013; 136: 3176-3186
        • Murakami H
        • Wang ZI
        • Marashly A
        • Krishnan B
        • Prayson RA
        • Kakisaka Y
        • et al.
        Correlating magnetoencephalography to stereo-electroencephalography in patients undergoing epilepsy surgery.
        Brain. 2016; 139: 2935-2947
        • Zhang M
        • Liu W
        • Huang P
        • Lin X
        • Huang X
        • Meng H
        • et al.
        Utility of hybrid PET/MRI multiparametric imaging in navigating SEEG placement in refractory epilepsy.
        Seizure. 2020; 81: 295-303
        • Engel J
        • Van Ness P
        • Rasmussen T.
        Outcome with respect to epileptic seizures. 1993; (p): 609-621
        • Minotti L
        • Montavont A
        • Scholly J
        • Tyvaert L
        • Taussig D.
        Indications and limits of stereoelectroencephalography (SEEG).
        Neurophysiol Clin. 2018; 48: 15-24
        • Burgess RC.
        MEG for Greater Sensitivity and More Precise Localization in Epilepsy.
        Neuroimaging Clin N Am. 2020; 30: 145-158
        • He X
        • Zhou J
        • Teng P
        • Wang X
        • Guan Y
        • Zhai F
        • et al.
        The impact of MEG results on surgical outcomes in patients with drug-resistant epilepsy associated with focal encephalomalacia: a single-center experience.
        J Neurol. 2020; 267: 812-822
        • Kim D
        • Joo EY
        • Seo DW
        • Kim MY
        • Lee YH
        • Kwon HC
        • et al.
        Accuracy of MEG in localizing irritative zone and seizure onset zone: Quantitative comparison between MEG and intracranial EEG.
        Epilepsy Res. 2016; 127: 291-301
        • RamachandranNair R
        • Otsubo H
        • Shroff MM
        • Ochi A
        • Weiss SK
        • Rutka JT
        • et al.
        MEG predicts outcome following surgery for intractable epilepsy in children with normal or nonfocal MRI findings.
        Epilepsia. 2007; 48: 149-157
        • Iida K
        • Otsubo H
        • Matsumoto Y
        • Ochi A
        • Oishi M
        • Holowka S
        • et al.
        Characterizing magnetic spike sources by using magnetoencephalography-guided neuronavigation in epilepsy surgery in pediatric patients.
        J Neurosurg. 2005; 102 (Suppl): 187-196
        • Shirozu H
        • Hashizume A
        • Masuda H
        • Kakita A
        • Otsubo H
        • Kameyama S.
        Surgical strategy for focal cortical dysplasia based on the analysis of the spike onset and peak zones on magnetoencephalography.
        J Neurosurg. 2019; : 1-13
        • Wang S
        • Tang Y
        • Aung T
        • Chen C
        • Katagiri M
        • Jones SE
        • et al.
        Multimodal noninvasive evaluation in MRI-negative operculoinsular epilepsy.
        J Neurosurg. 2019; 132: 1334-1344
        • Vadera S
        • Jehi L
        • Burgess RC
        • Shea K
        • Alexopoulos AV
        • Mosher J
        • et al.
        Correlation between magnetoencephalography-based "clusterectomy" and postoperative seizure freedom.
        Neurosurg Focus. 2013; 34: E9
        • Stefan H
        • Wu X
        • Buchfelder M
        • Rampp S
        • Kasper B
        • Hopfengartner R
        • et al.
        MEG in frontal lobe epilepsies: localization and postoperative outcome.
        Epilepsia. 2011; 52: 2233-2238