Epilepsy surgery is an important treatment option for patients with drug-resistant epilepsy (DRE). The primary goal of presurgical evaluation is to identify the epileptogenic zone (EZ). Subdural electrodes, including grids and strips, have long been used to delineate the EZ and map functional cortex [1]. However, the emergence of stereoelectroencephalography (SEEG) has shifted the paradigm in presurgical evaluation [2,3]. Over recent decades, advances in imaging, robotic guidance, and electrode technology have improved the precision and safety of SEEG electrode implantation [4,5]. Compared with scalp electroencephalography (EEG) or subdural EEG, SEEG offers several advantages: it enables three-dimensional sampling of the brain, including deep and bilateral regions, with minimal invasiveness [2,6]. Recent systematic reviews and meta-analyses have also shown that SEEG is associated with lower complication rates, particularly infection and hemorrhage, than subdural electrode implantation [7-9].

Several studies have evaluated the effectiveness of SEEG-guided epilepsy surgery, reporting that approximately 60% to 70% of patients achieve Engel class I outcomes [10-13]. As evidence has accumulated, SEEG has been adopted across epilepsy centers worldwide, and international data now support its utility in presurgical evaluation for both adults and children [7,8,14]. However, pediatric-specific outcome data from Korean epilepsy centers remain limited, and few studies have reported long-term surgical outcomes after SEEG-guided surgery in Korean pediatric cohorts. Additional data from Korean institutions are therefore needed to better define outcomes and optimize surgical strategies in children.

This study evaluated the surgical outcomes of SEEG-guided epilepsy surgery in pediatric patients with DRE. Specifically, we compared surgical outcomes achieved after SEEG-based evaluations with those achieved after conventional subdural grid-based approaches. We also sought to identify potential areas for improvement in SEEG-guided surgical approaches, with the goal of optimizing patient outcomes.

All pediatric patients with DRE who underwent SEEG-guided resective epilepsy surgery at Severance Children’s Hospital between 2015 and 2023 and had at least 1 year of postoperative follow-up were included. As a comparison group, all patients who underwent resective epilepsy surgery with subdural grid monitoring during the same period were included. Both SEEG and subdural EEG monitoring were performed for at least 3 days. Retrospective data were collected to analyze clinical characteristics and surgical outcomes.

Patients in both the SEEG and subdural EEG groups underwent standardized presurgical evaluation, including seizure history, scalp video-EEG, brain magnetic resonance imaging (MRI), 18F-fluorodeoxyglucose positron emission tomography/computed tomography, single-photon emission computed tomography, and neuropsychological testing. Intraoperative electrocorticography was performed in all cases.

In the SEEG group, electrodes were implanted with the Robotized Stereotactic Assistant (ROSA) system (Zimmer Biomet, Montpellier, France), guided by individualized hypotheses generated from noninvasive evaluations. Depth electrodes were stereotactically implanted along trajectories tailored to the suspected seizure onset zone and propagation pathways. Electrode placement was planned using co-registered high-resolution MRI and computed tomography. Continuous video-EEG monitoring was performed for 3–7 days to capture both interictal and ictal events before resection.

The following variables were collected and analyzed for each patient: demographic variables, epilepsy-related characteristics, surgical variables, histopathological diagnoses, and postoperative outcomes. Demographic variables included sex and age at surgery. Epilepsy-related characteristics included epilepsy duration, age at seizure onset, seizure frequency, seizure type, electroclinical syndrome, epilepsy laterality, localization of the EZ, generalized epileptiform discharges on scalp EEG, number of anti-seizure medications before surgery, and prior epilepsy surgery. Seizure types were classified according to the revised 2025 International League Against Epilepsy (ILAE) classification, and electroclinical syndromes were defined according to the current ILAE syndrome classification. Surgical variables included the type and duration of invasive monitoring, the number of implanted electrodes in the SEEG group, the surgical method used (such as resection or disconnection), the anatomical resection site, and the primary indication for SEEG selection, categorized as reoperation, deep-seated lesion, bilateral lesion, multilobar lesion, or MRI-negative epilepsy. Multilobar resection was defined as resection involving two or more lobes within the same hemisphere. Histopathological diagnoses were classified as gliosis, malformations of cortical development (MCD), tumors, tuber, and other diagnoses. MCD cases were further subclassified according to the ILAE 2022 classification of focal cortical dysplasia (FCD), and tumor cases were subclassified according to the World Health Organization 2021 classification of central nervous system tumors. Postoperative seizure outcomes were assessed using the Engel classification and the ILAE outcome scale at the last follow-up visit. A favorable outcome was defined as Engel class I with a minimum postoperative follow-up of 12 months. Follow-up duration was also recorded.

Data are presented as median (interquartile range) or number (percentage). The chi-square test (or Fisher exact test when appropriate), Cox regression, and the Mann–Whitney U test were used to compare the two groups. Statistical analyses were performed using IBM SPSS Statistics version 27.0 (IBM Corp., Armonk, NY, USA). Multivariate logistic regression with Firth bias reduction was used to identify variables independently associated with seizure outcome using R version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria). A two-sided P value less than 0.05 was considered to indicate statistical significance.

This retrospective study was approved by the Institutional Review Board (IRB No. 4-2026-0312) of Severance Hospital, Yonsei University Health System. The IRB waived the requirement for informed consent due to the retrospective design and use of anonymized clinical data.

Epilepsy surgery is an established treatment for DRE, with long-term seizure freedom rates ranging from 27% to 84% depending on resection site and pathology [11,12]. The Early Randomized Surgical Epilepsy Trial (ERSET) further demonstrated the superiority of early surgical intervention over continued medical therapy in mesial temporal lobe epilepsy [15]. Within this framework, adoption of SEEG has refined the concept of the EZ from a localized cortical focus to a dynamic network of rapidly synchronizing discharges [16-18], a perspective that underpins its contemporary role in presurgical evaluation.

Compared with subdural grid monitoring, SEEG offers three key advantages: three-dimensional sampling of cortical and subcortical structures, which is particularly useful in deep-seated, bilateral, or non-lesional cases [19,20]; improved localization precision that may spare eloquent cortex [19]; and minimally invasive implantation through burr holes rather than craniotomy, with correspondingly lower rates of infection, hemorrhage, and cerebrospinal fluid leakage [21,22]. Recent multicenter studies and meta-analyses have confirmed that SEEG provides comparable or superior localization accuracy with substantially lower morbidity, particularly in pediatric populations [19,20,22].

In the present retrospective study, we evaluated surgical outcomes in pediatric patients with epilepsy who underwent invasive monitoring with SEEG, with patients who underwent subdural grid monitoring serving as a comparison group. SEEG-guided surgery was associated with favorable seizure outcomes, with 75% of patients achieving Engel class I and 69% achieving ILAE class 1. These results are consistent with recent pediatric SEEG studies and meta-analyses, which have reported comparable efficacy and safety profiles in children across age groups [22,23]. Taussig et al. [23] reported that SEEG was followed by resective surgery in 78% of pediatric patients, with favorable seizure freedom rates and outcomes in younger children comparable to those in older children. Recent systematic reviews and meta-analyses have likewise reported pooled Engel class I rates of approximately 55% to 65% across pediatric SEEG cohorts [20,22,23]. The somewhat higher seizure freedom rate in our SEEG group may reflect the relatively high proportion of patients with lesional pathology, including MCD and tumors, as well as the shorter follow-up period.

A notable feature of our cohort was that the SEEG group included a higher proportion of patients with prior epilepsy surgery and complex multilobar or non-lesional epilepsy. Despite these clinical challenges, the SEEG group had outcomes comparable to or better than those of the grid group. These findings suggest that SEEG could be particularly useful in complex cases in which subdural grid placement may not provide sufficient spatial resolution or access to deep structures [19,20]. The three-dimensional coverage provided by SEEG may improve identification of the epileptogenic network and facilitate accurate surgical planning [16,17]. Similar findings have been reported in previous pediatric cohorts, in which SEEG-guided reoperations achieved favorable seizure outcomes in approximately 50% to 60% of carefully selected patients [24], reinforcing the utility of SEEG in this challenging subgroup.

Histopathological analysis identified significant differences between the two groups. Patients who underwent SEEG-guided epilepsy surgery had a higher proportion of tumors than those in the grid group (19.4% vs. 1.5%), whereas gliosis accounted for a larger proportion of diagnoses in the grid group than in the SEEG group (56.9% vs. 36.1%). These differences may partly reflect selection bias, because SEEG is more often used for cases with deep-seated lesions, such as mesial temporal tumors, or for multiple potential epileptogenic foci without clear boundaries on MRI [19,20]. Nevertheless, the capacity of SEEG to localize EZs in non-lesional or diffuse pathologies highlights its versatility.

The lower morbidity associated with SEEG compared with subdural grid implantation has been well documented [19,25-27]. In our study, no severe complications were reported after SEEG implantation; two of 36 patients had mild transient hemiparesis, and one developed hemianopsia after occipital lobectomy. In contrast, the subdural grid group had one case of epidural hematoma, two cases of transient hemiparesis, one case of visual field defect, and one case of aphasia. In addition, many grid-monitored patients developed fever after electrode implantation and required postoperative intensive care unit (ICU) monitoring. Subdural grid placement often requires larger craniotomies, which are associated with higher risks of infection and hemorrhage [26,27]. This safety advantage supports the use of SEEG in appropriate clinical scenarios.

The duration of invasive monitoring was significantly shorter in the SEEG group than in the grid group (median, 5 days vs. 7 days; P<0.001). The longer monitoring duration in the grid group may reflect several factors: (1) the larger craniotomy and associated postoperative recovery period, which often required ICU monitoring before reliable seizure recording could be obtained; (2) the need for prolonged extraoperative cortical mapping when grids overlaid eloquent cortex; and (3) the higher likelihood of capturing habitual seizures requiring extensive surface coverage [26,27]. In contrast, the targeted three-dimensional sampling and minimally invasive implantation used in SEEG generally enabled earlier seizure capture and faster decision-making regarding resection planning.

This study has several limitations. First, the follow-up period for the SEEG group was relatively short. Because late seizure recurrence is a recognized phenomenon, extended follow-up is essential to validate the long-term durability of the favorable outcomes observed in the SEEG group [11]. Second, the modest sample size reduced statistical power and limited generalizability. Third, the retrospective design introduced potential selection bias, as the choice between SEEG and subdural grid monitoring was based on clinical judgment rather than randomization.

Despite these limitations, our findings support the clinical utility of SEEG as a minimally invasive, highly informative approach for defining the epileptogenic network in difficult-to-localize cases. Given its favorable diagnostic accuracy and safety profile, SEEG may be considered an increasingly valuable invasive monitoring modality in pediatric populations, in whom conventional methods may pose greater surgical risks [19,20,23].

In conclusion, SEEG may serve as a preferred invasive evaluation tool for pediatric patients with DRE, particularly those with deep, multilobar, or non-lesional EZs. Future prospective multicenter studies using standardized protocols are warranted to further define the role of SEEG and optimize surgical outcomes across diverse patient populations.