Impact of 18 Months of Magnetoencephalography-Magnetic Source Imaging on Presurgical Decision-Making for Epilepsy Surgery in South Korea: A Single Pediatric Epilepsy Center Study
Article information
Abstract
Purpose
This study evaluated the impact of magnetoencephalography-magnetic source imaging (MEG-MSI) on the presurgical evaluation of patients with epilepsy.
Methods
We retrospectively analyzed presurgical evaluation reports for 64 patients considered for epilepsy surgery at a single pediatric epilepsy center.
Results
Overall, 52 of 64 patients (81%) were considered candidates for resective surgery before MEG-MSI. In 37 of these 52 patients (71%), MEG-MSI findings were concordant with those of other modalities, supporting the clinical decision to proceed with resective surgery. However, in 11 of the 52 cases (21%), resective surgery was no longer considered beneficial after MEG-MSI. Among these 11 patients, five underwent either corpus callosotomy or vagus nerve stimulator insertion.
Conclusion
Since MEG was incorporated into routine presurgical evaluation for patients with epilepsy, MEG-MSI findings have influenced clinical decision-making in 11 patients.
Introduction
Magnetoencephalography (MEG) is a neurophysiological test that records cerebral activity by detecting magnetic fields generated by primary currents traversing the cortical columns of the gray matter [1]. The primary current is a conceptual current that represents the aggregate of intracellular sources, sinks, induced currents, and transmembrane currents generated by neuronal electrical activity [2]. Based on measurements of induced magnetic fields, the corresponding primary current is modeled as a single equivalent current dipole (SECD), also known as a MEG dipole. Each MEG dipole is then localized to the patient’s co-registered brain magnetic resonance imaging (MRI) scan for source localization of cerebral activity, a process termed magnetic source imaging (MSI).
Owing to its high spatial and temporal resolution, as well as its sensitivity to tangential sources of cortical activity compared with electroencephalography (EEG) [3], MEG-MSI has been considered an essential component of presurgical evaluation and a complement to other modalities, including video-EEG, positron emission tomography-computed tomography (PET-CT), single-photon emission computed tomography (SPECT) [4], and functional MRI. Unlike electrical activity, magnetic fields undergo minimal attenuation by tissues such as the scalp and skull [1]. EEG is sensitive to radial sources of cerebral activity, whereas MEG is more sensitive to tangential sources. A quantitative assessment of MEG sensitivity to cerebral activity in the adult human cortex showed that MEG can reliably detect magnetic fields down to a 15° angle between the cerebral source and the magnetometer. In contrast, the detection probability for nearly radial sources, defined as sources with a 0° to 15° angle between the cerebral source and the magnetometer, is low; however, such sources account for less than 5% of the cortex [3]. Furthermore, MEG has inherently higher resolution and can detect activity from a smaller area of synchronously discharging cortex than EEG: approximately 4 cm² for MEG versus approximately 6 cm² for EEG. Although EEG and MEG are complementary, MEG can detect more sources of cerebral activity with higher resolution than EEG [5].
Several studies have demonstrated the benefits of MEG-MSI in epilepsy surgery. For example, Sutherling et al. [6] reported that MEG-MSI provided nonredundant information in 33% of patients who underwent presurgical evaluation and benefited 21% of patients who underwent epilepsy surgery. More recently, concordance between stereoelectroencephalography (SEEG) and MEG-MSI, followed by surgical removal of concordant regions, has shown potential as a positive predictor of favorable seizure outcomes after epilepsy surgery [7]. In addition, postoperative seizure freedom at 12 months was associated with both complete intracranial sampling using SEEG, with seizure freedom rates of 62% versus 25%, and complete resection of MEG-MSI clusters, with seizure freedom rates of 88% versus 52% [8].
The first successful detection of electrical activity in the brain using a superconducting magnetometer was demonstrated by Cohen [9] in 1972. The availability of a superconducting magnetometer was essential to this achievement because magnetic-field noise sources with amplitudes greater than those of the brain are ubiquitous, ranging from home appliances (1×10−5 T) and urban environmental noise (1×10−6 T) to the Earth’s magnetic field (5×10−5 T). By contrast, magnetic fields generated by the brain are several orders of magnitude weaker, ranging from 1×10−11 T for alpha rhythm to 1×10−14T for evoked cortical fields [10]. Therefore, the implementation of superconducting magnetometers, most commonly in the form of superconducting quantum interference devices, and the instrumentation required to maintain temperatures near absolute zero, specifically 4.2 K, equivalent to −269°C, using liquid helium as a cryogenic refrigerant, have hindered the widespread use of MEG-MSI worldwide [2,11].
Even in the United States, MEG-MSI has been underutilized in routine presurgical evaluation of patients with epilepsy. A recent survey-based appraisal of MEG-MSI utilization in the United States in 2016 showed that the average number of MEG-MSI orders was approximately 11 studies per year, with 717 studies ordered by 64 hospitals; however, the distribution of MEG-MSI orders was highly skewed toward a select group of hospitals. Of the 717 MEG-MSI studies performed for patients with epilepsy, 53% were ordered by the top five centers, and nearly 90% were ordered by the top 16 centers nationwide [12]. When the analysis was limited to MEG-MSI studies for presurgical evaluation of patients with epilepsy, the top five centers accounted for almost 70% of all studies, corresponding to 347 of 501 studies. A survey-based 10-year appraisal of clinical MEG-MSI practice between 2006 and 2016 further concluded that the number of MEG-MSI studies had reached a plateau and that MEG-MSI had not become part of the standard of care in the United States [13,14]. Contributing factors included limitations in capital investment, impractical software, and a shortage of trained practitioners. Furthermore, the clinical practice guidelines for MEG-MSI issued by the American Clinical Magnetoencephalography Society have had little impact on clinical practice. Greater education for practitioners and referrers remains needed, as well as clearer demonstration of the clinical value of MEG-MSI through high-quality examinations [14].
To our knowledge, South Korea had only one MEG center operating from the 2000s until the 2010s, and this center was eventually closed. During this period, MEG-MSI was considered a safe and useful presurgical evaluation modality in pediatric patients with lesion localization-related epilepsy [15], and MEG-MSI cortical mapping was particularly helpful when the epileptogenic zone was close to eloquent cortex [16]. The MEG center with the country’s only operating MEG system reopened in August 2023 at Severance Hospital in South Korea. Temporary operations began in March 2023 before the official opening. The operating cost of the MEG system was reduced through liquid helium recycling, although helium recycling was still required once per year. The reopening of the MEG center required financial support from both the South Korean government and Severance Hospital [17]. As described above for MEG centers in the United States [14], maintaining high-quality MEG-MSI examinations, improving capital investment, designing better software, and training practitioners are critical ongoing priorities.
In this study, we examined how incorporating MEG-MSI into routine presurgical evaluation influenced decision-making for epilepsy surgery at the Severance Children's Hospital pediatric epilepsy center during the first 18 months after the center opened.
Materials and Methods
1. Ethical statements
The requirement for informed consent was waived by the Institutional Review Board (IRB). This study was conducted in accordance with the ethical standards of the IRB of Severance Hospital (IRB number: 4-2024-1330) and the Declaration of Helsinki of 1964, as revised in 2000.
2. Patient selection
Patients who underwent presurgical evaluation for epilepsy surgery at our pediatric epilepsy center between March 2023 and September 2024 were discussed at multidisciplinary pediatric epilepsy surgery conferences involving pediatric neurologists, neurosurgeons, and/or neuroradiologists. At our center, patients are usually referred to an adult neurology clinic when they reach 18 years of age; however, many continue to attend our pediatric epilepsy center voluntarily. For all adult patients included in this study, the first visit to our center occurred before 18 years of age. No adult neurologist was involved or consulted at any stage of patient care, including for patients older than 18 years.
3. MEG-EEG acquisition
A whole-head 306-channel MEGIN system was used to acquire MEG recordings, with simultaneous recording from 21 EEG channels in a magnetically shielded room. Recordings were performed with patients in the supine position. EEG was recorded using a standard 10–20 electrode setup. The left mastoid was used as the reference, and the right mastoid electrode was used as the ground. Two diagonally positioned bipolar electrooculography electrodes were placed above and below each eye to monitor eye movements and blinking. A single bipolar electrocardiography channel was used to record cardiac activity. A total of 120 minutes of data were acquired using the following parameters: a 0.1-Hz high-pass filter, a 333.3-Hz low-pass filter, and a sampling rate of 1,000 Hz.
4. Analysis of spontaneous activity
The entire analysis was conducted using MEGIN data analysis software. Waveforms were visually inspected offline using variable settings, including a bandpass filter in the range of 1 to 70 Hz, with or without a notch filter and with or without a spatial filter, as appropriate. The automatic spike search function was not used to identify events that met predefined criteria based on the initial visual inspection.
EEG data were reviewed using established visual methods, and no source modeling was applied. Identified EEG spikes were used as landmarks to fit the corresponding MEG signals, regardless of morphology. Identified MEG spikes or sharp waves without corresponding EEG correlates were also fitted.
Selected MEG-domain events were mapped using an SECD model. A single dipole was selected to represent each sharp wave or spike. The dipole selection criteria included the following eight elements: (1) a well-formed dipolar field configuration; (2) central location on the dipole contour map; (3) dipole moment from 100 to 400 nA·m; (4) goodness of fit >80%; (5) confidence volume <5,000 mm³; (6) reduced chi-square <2; (7) fitting probability >80%; and (8) physiological plausibility. In general, the SECD was selected at a time point between spike onset and maximum spike amplitude. SECD calculations were initially performed using 306 channels. Subsequently, regional selection of gradiometer channels was performed to best represent the contour plot of the magnetic field, as deemed appropriate. Accepted dipole locations representing sharp interictal waves or spikes were calculated using a spherical head model and projected onto each patient’s MRI data.
Results
1. Patient demographics
A total of 64 patients were included in this study, of whom 30 (47%) were male and 34 (53%) were female (Table 1). Patients’ ages at the time of MEG-MSI ranged from 0.39 to 34.83 years, with a median age of 13.67 years. Among patients who underwent invasive intracranial monitoring after MEG-MSI, the interval between MEG-MSI and the start of invasive intracranial monitoring ranged from 1 to 323 days, with a median interval of 91 days. Twenty-three patients (36%) had previously undergone one, two, three, or four epilepsy surgeries, including 15 patients with one prior surgery (23% of the total), four patients with two prior surgeries (6%), three patients with three prior surgeries (5%), and one patient with four prior surgeries (1.6%). Brain MRI showed 52 lesion positive cases (81%), including 47 cases (73%) with unilateral lesions and five cases (8%) with bilateral lesions, whereas 12 cases (19%) were lesion negative. On MEG-MSI, MEG dipoles were unilateral in 44 cases (69%), whereas 20 cases showed bilateral MEG dipoles (31%). MEG-MSI functional mapping was performed for sensory function in 58 cases (91%), motor function in 12 cases (19%), and visual function in three cases (5%), but not for auditory function. Detailed information on seizure onset, seizure type, semiology, video-EEG, MRI, PET, history of epilepsy surgery, and, if present, the type of neuromodulation received is shown for each patient in Supplementary Table 1.
Patient demographics
2. Flow of clinical decisions after presurgical evaluation including MEG-MSI
Based on a retrospective medical chart review, we reconstructed how presurgical evaluation, including MEG-MSI, informed subsequent clinical decisions in this patient cohort (Fig. 1). Of the 64 patients, 52 (81%) were initially considered candidates for resective surgery based on the results of other studies, including video-EEG, seizure semiology, PET-CT, SPECT, and/or brain MRI. MEG-MSI findings were concordant with those of other presurgical modalities in 37 of these 52 cases (71%). Among these 37 patients, 27 underwent resective surgery, four were scheduled for resective surgery, and five patients and/or their guardians decided not to proceed with resective surgery after it was recommended following presurgical evaluation, including MEG-MSI. In the remaining patient, SEEG was performed, and the epileptogenic zone was ultimately found to involve Broca’s area; therefore, resective surgery was canceled.
Flow of clinical decisions after presurgical evaluation including magnetoencephalography-magnetic source imaging (MEG-MSI). VNS, vagus nerve stimulation; OP, operation; SEEG, stereoelectroencephalography; ASM, antiseizure medication.
In 15 of the 52 cases, although the patients were initially considered suitable surgical candidates, the treatment plan changed after presurgical evaluation. In 11 of these 15 patients, the change in plan was attributable to MEG-MSI findings, whereas in the remaining four patients, factors other than MEG-MSI were responsible. MEG-MSI findings led to a decision not to proceed with epilepsy surgery in 11 of the 15 patients. Three of these 11 patients showed bilateral MEG dipoles on MEG-MSI and underwent corpus callosotomy in one case or vagus nerve stimulation (VNS) insertion in two cases instead. In two of the 11 patients, MEG dipoles were localized to the visual and motor cortices, and these patients subsequently underwent VNS insertion. Among the remaining six patients, bilateral, widespread MEG dipoles were recorded in three patients; surgery was no longer considered beneficial, and neither corpus callosotomy nor VNS insertion was performed. The other three patients had MEG-MSI findings that were discordant with those of other presurgical modalities. In four of the 15 patients, MEG-MSI did not contribute to the change in treatment plan. These patients were undergoing other treatments, including a new trial of antiseizure medication, a steroid trial, or a ketogenic diet, and showed improvement; therefore, surgery was not discussed further.
Of the 64 cases, 12 (19%) were considered non-candidates for resective surgery before MEG-MSI, but MEG-MSI was performed as a confirmatory tool to establish concordance with other presurgical modalities. In one of these 12 patients, the epileptogenic lesion was previously known to be located in the motor cortex, which was confirmed by MEG-MSI. The remaining 11 patients were already known to have bilateral lesions, and MEG-MSI confirmed the bilaterality of the seizure foci. Five of these 11 patients underwent either corpus callosotomy in two cases or VNS insertion in three cases.
3. Illustrative cases
Here, we present two illustrative cases in which MEG-MSI was helpful in localizing presumed seizure foci.
Case 1 was an 18-year-old female patient whose seizures began at 9 years of age. The frequency of focal impaired awareness seizures increased to 10–30 episodes per day. Brain MRI findings were normal, but EEG showed slow activity in both temporal regions (Fig. 2B). However, MEG-MSI revealed a clear right temporal focus, with MEG dipoles clustered in the right temporal region (Fig. 2). This case demonstrates the lateralizing value of MEG-MSI: MEG dipoles were localized to the right temporal area, whereas EEG did not show right temporal lateralization.
(A) Axial brain magnetic resonance images with magnetoencephalography (MEG) dipoles distributed in the right temporal area. (B) MEG and electroencephalography (EEG) recordings from the same patient showing clear right temporal interictal epileptiform discharges on MEG but bilateral temporal slowing on EEG. Lt, left; Rt, right.
Case 2 was a 23-year-old female patient whose seizures began at 5 months of age. She first visited our hospital at 6 years of age. At our hospital, she underwent left parieto-occipital lobectomy, left parieto-occipital mesial lobectomy, and left parietal corticotomy; the resected tissue confirmed focal cortical dysplasia type IIb. After left parietal corticotomy at 13 years of age, a scalp wound defect persisted, and she developed a wound infection caused by Pseudomonas aeruginosa. At 15 years of age, she underwent cranioplasty with titanium plate insertion (Fig. 3A and B). However, she continued to experience right tonic and focal aware seizures multiple times per day. EEG revealed only muscle artifacts. In contrast, MEG-MSI showed MEG dipoles near the previous resection sites, including the left temporo-occipital area, left insular area, and left supplementary motor area (Fig. 3C).
(A) Anteroposterior skull radiograph of the patient. (B) Lateral skull radiograph of the patient. Note the titanium plate. (C) Axial brain magnetic resonance images with magnetoencephalography dipoles distributed near the previous resection sites.
Discussion
Herein, we retrospectively reviewed the records of 64 patients at our hospital who underwent MEG-MSI as part of presurgical evaluation. Our results are comparable to those of a previous report showing that MEG-MSI provides nonredundant information in 33% of patients [6]. In our retrospective analysis, 11 of the 52 patients initially considered suitable surgical candidates (21%) did not undergo resective surgery because of their MEG-MSI findings. In these patients, MEG-MSI provided information that was discordant with that from other presurgical modalities, demonstrating the importance of incorporating MEG-MSI into routine presurgical evaluation [18].
In 12 patients, representing 19% of the total cohort, resective surgery was not considered beneficial even before MEG-MSI, and MEG-MSI provided additional evidence that resective surgery should not be performed. In this subgroup, MEG-MSI was used as a confirmatory tool. This represented a special circumstance because these patients underwent MEG-MSI free of charge during the temporary opening period. However, given the cost and effectiveness considerations associated with MEG-MSI, its use solely as a confirmatory tool should be avoided.
In 37 of the 52 patients (71%) in whom MEG-MSI findings were concordant with those of other presurgical modalities, the MEG-MSI findings were considered valuable by pediatric neurologists. Because MEG dipoles are registered onto each patient’s brain MRI, they provide clearer anatomical guidance for targeting SEEG electrodes and mapping the epileptogenic focus as accurately as possible [7,8]. Thus, concordant findings across presurgical modalities, including MEG-MSI, may provide reliable guidance for SEEG electrode placement during invasive intracranial monitoring. However, future studies are required to assess the contribution of MEG-MSI findings to seizure outcomes.
The 10 notable indications for MEG-MSI are as follows: (1) absent or imprecise hypotheses regarding seizure onset; (2) nonlesional MRI with suspected mesial temporal onset; (3) multiple brain lesions on MRI; (4) a large brain lesion on MRI; (5) diagnostic or therapeutic reoperations; (6) ambiguous EEG findings suggestive of a bilateral or generalized seizure pattern; (7) suspected intrasylvian seizure onset; (8) suspected interhemispheric onset; (9) suspected insular seizure onset; and (10) negative EEG findings without spikes [18]. At our center, the proportion of patients who had already undergone epilepsy surgery before MEG-MSI was relatively high, with 23 patients representing 36% of the total cohort; this was one of the 10 indications for MEG-MSI [18]. MEG-MSI is recommended for postoperative patients in whom reoperation is being considered [1], because both operative and traumatic skull defects distort electrical fields and render EEG unreliable. By contrast, MEG-MSI is not affected by these distortions [18]. Preoperative MEG-MSI in epilepsy surgery reoperations can support surgical decision-making [19]. MEG-MSI was highly valuable for localizing potentially epileptogenic lesions during presurgical evaluation for epilepsy surgery reoperation, as illustrated in case 2 and Fig. 3. However, future studies of seizure outcomes after reoperation are warranted.
This study has several limitations. First, it was not a prospective study but rather a retrospective analysis of a nonblinded observational case series. This was a preliminary study, and long-term seizure outcomes after epilepsy surgery remain unknown; these outcomes will need to be addressed in future studies. Although we attempted to evaluate the impact of MEG-MSI, it was difficult to isolate the influence of MEG-MSI from that of other presurgical investigations, including PET-CT, brain MRI, SPECT, and/or video-EEG, as is inherent in most studies of presurgical evaluation for epilepsy surgery. The number of patients with prior epilepsy surgery was high; therefore, our cohort may not represent a typical population of patients undergoing epilepsy surgery. MEG-MSI was analyzed using the SECD method in this study; however, other, more advanced source localization methods are available, including dipole scanning, current distribution analysis, beamforming, maximum entropy on the mean, and connectivity analysis [20]. These algorithms will be implemented in the future.
We hope that this study will contribute to future long-term prospective investigations of the influence and usefulness of MEG-MSI in improving seizure outcomes in patients undergoing epilepsy surgery.
In conclusion, among the 64 patients who underwent presurgical evaluation for epilepsy surgery with MEG-MSI at our pediatric epilepsy center, 12 (19%) were considered non-candidates for resective surgery before MEG-MSI, and MEG-MSI subsequently confirmed that these patients were not candidates for resective surgery, as expected. In addition, 52 patients (81%) were considered candidates for resective surgery before MEG-MSI, and MEG-MSI findings were concordant with those of other presurgical modalities in 37 patients (71%), supporting clinical decisions to proceed with resective surgery. In these cases, MEG-MSI findings provided additional information for accurately targeting SEEG electrodes during invasive intracranial monitoring.
Since MEG-MSI was incorporated into routine presurgical evaluation for epilepsy surgery, clinical decision-making in at least 21% of initially eligible surgical candidates, corresponding to 11 patients, was strongly influenced by MEG-MSI findings. MEG-MSI findings, together with findings from other modalities, have become an indispensable component of presurgical evaluation at our pediatric epilepsy center.
However, MEG-MSI infrastructure remains limited in South Korea, and reopening the MEG center was a national priority that required financial support from the government. We hope that the results of this study will underscore the importance of continued support for the MEG center in South Korea.
Supplementary material
Supplementary materials related to this article can be found online at https://doi.org/10.26815/acn.2026.01501
Individual patient characteristics
Notes
Conflicts of interest
Ara Ko and Se Hee Kim are managing editors, Hoon-Chul Kang is an associate editor, Joon Soo Lee is an editorial board member, and Heung-Dong Kim is an emeritus editor of the journal, but they were not involved in the peer reviewer selection, evaluation, or decision process of this article. No other potential conflicts of interest relevant to this article were reported.
Author contribution
Conceptualization: SJL, AK, SHK, HCK, JSL, and HDK. Data curation: SJL and AK. Formal analysis: SJL. Methodology: SJL and SHK. Project administration: SJL, AK, SHK, HCK, JSL, and HDK. Visualization: SJL. Writing-original draft: SJL. Writing-review & editing: SJL and AK.
Acknowledgments
The authors would like to acknowledge the members of our MEG center at Severance.