Automated Brain Volumetric Analysis in Children with Genetic Epilepsy and Negative Qualitative Magnetic Resonance Imaging Findings

Article information

Ann Child Neurol. 2026;34(3):208-214

Publication date (electronic) : 2026 July 1

doi : https://doi.org/10.26815/acn.2026.01466

Yun-Jung Lim, MD¹, Yun Jung Hur, MD, PhD^,2

¹Department of Radiology, Inje University Haeundae Paik Hospital, Inje University College of Medicine, Busan, Korea

²Department of Pediatrics, Inje University Haeundae Paik Hospital, Inje University College of Medicine, Busan, Korea

Corresponding author: Yun Jung Hur, MD, PhD Department of Pediatrics, Inje University Haeundae Paik Hospital, Inje University College of Medicine, 875 Haeun-daero, Haeundae-gu, Busan 48108, Korea Tel: +82-51-7972000 Fax: +82-51-7973194 E-mail: H00105@paik.ac.kr

Received 2026 March 6; Revised 2026 April 19; Accepted 2026 April 20.

Abstract

Introduction

Epilepsy is a common neurological disorder in childhood, and recent advances in genetic testing have identified genetic etiologies in a substantial proportion of pediatric cases [1-3]. Genetic epilepsies often present in early childhood and are frequently associated with adverse neurodevelopmental and cognitive outcomes, making early identification essential for appropriate management and prognostic counseling [1,3-5]. However, the clinical manifestations of genetic epilepsy are highly heterogeneous, and genotype–phenotype correlations remain inconsistent, limiting the ability to predict a genetic etiology based solely on clinical features [6]. Although next-generation sequencing techniques have improved diagnostic yield, genetic testing remains costly and is not routinely performed in all pediatric patients with epilepsy. Therefore, additional biomarkers are needed to identify patients at higher risk for genetic epilepsy and to guide targeted genetic testing [7].

Magnetic resonance imaging (MRI) plays an essential role in the evaluation of pediatric epilepsy. In certain genetic epilepsies, characteristic structural abnormalities—such as lissencephaly, polymicrogyria, or focal cortical dysplasia—can be identified, facilitating etiological diagnosis [8-10]. However, many children with genetic epilepsy do not exhibit visible structural abnormalities on conventional MRI and are classified as having genetic epilepsy with negative qualitative MRI findings [11,12]. In such cases, qualitative visual assessment alone may be insufficient to detect subtle brain abnormalities associated with genetic etiologies. Quantitative MRI techniques, including volumetric analysis and cortical morphometry, enable the detection of structural alterations that are not apparent on visual inspection [12,13]. Previous studies have reported reductions in brain volume and cortical thickness, as well as alterations in cortical surface area, in children with genetic epilepsy, suggesting that abnormal brain development is associated with underlying genetic mutations [14,15]. However, studies specifically examining volumetric alterations in children with genetic epilepsy and negative qualitative MRI findings remain limited [16].

Recent advances in deep learning-based automated brain segmentation have enabled rapid, reliable, and efficient volumetric analysis of brain MRI, providing quantitative measures of brain structure [17-20]. These techniques may facilitate the identification of imaging biomarkers for genetic epilepsy, even in patients without visible structural lesions on MRI. In this study, we hypothesized that children with genetic epilepsy and negative qualitative MRI findings would exhibit distinct volumetric brain alterations compared with healthy controls. To test this hypothesis, we performed automated brain segmentation using a deep learning-based volumetric analysis program and compared total and regional brain volumes between children with genetic epilepsy and negative MRI findings and age- and sex-matched healthy controls.

Materials and methods

This retrospective study was approved by the Institutional Review Board of Inje University Haeundae Paik Hospital (approval no. 2025-08-002), and all analyses were performed using deidentified data. Written informed consent by the patients or patient's guardians was waived due to a retrospective nature of our study.

1. Study design and population

From May 2000 to December 2025, 71 pediatric patients with genetically confirmed epilepsy who underwent brain MRI were initially identified. The exclusion criteria for the patient group were as follows: (1) abnormal signal intensity or abnormal enhancement on brain MRI suggestive of specific conditions, such as hypoxic–ischemic encephalopathy, demyelinating disease, brain tumor, or neurocutaneous syndrome; (2) cortical malformations, including lissencephaly, pachygyria, polymicrogyria, microcephaly with a simplified gyral pattern, or focal cortical dysplasia; (3) decreased cerebral or cerebellar volume on qualitative MRI assessment; (4) microcephaly or macrocephaly, defined as head circumference less than −2 standard deviations or greater than 2 standard deviations for age and sex; and (5) absence of three-dimensional T1-weighted MRI. Based on these criteria, 29 patients were excluded. The remaining 42 pediatric patients, aged 3 months to 17 years, with genetically confirmed epilepsy and normal brain MRI findings were enrolled. For each patient, two age- and sex-matched healthy controls were selected. The control group included 84 age- and sex-matched individuals diagnosed with primary headache or nonlesional trauma. None of the controls had a history of genetic diagnosis, congenital anomaly, developmental delay, family history of epilepsy, or genetic disease. In total, 42 patients with genetic epilepsy and normal brain MRI findings and 84 age- and sex-matched controls with normal brain MRI findings were finally included (Fig. 1).

Fig. 1.

Flowchart of the study population. MRI, magnetic resonance imaging.

Clinical data were obtained through retrospective review of the medical records. For the patient group, data on sex, age at seizure onset, age at MRI, number of antiseizure medications, seizure type, epilepsy syndrome, seizure-free status, and identified genetic variants were extracted. Seizure-free status was defined as the absence of seizures for at least 12 months. For the control group, data collection was limited to sex, age, and timing of MRI acquisition to ensure comparability.

Results

1. Clinical characteristics

The genetic epilepsy group was classified into five groups according to epilepsy syndrome and seizure type. The generalized epilepsy group (n=13) included patients with isodicentric 15 syndrome, Turner syndrome, and DDX3X/Xp22.33 duplication; pathogenic variants in CSNK2B, NHLRC1, GNAO1, MAT1A, RORB, CUX2, MECP2, and HECW2; and generalized epilepsy with febrile seizures plus (n=2) associated with SCN1A and SCN1B mutations. The focal epilepsy group (n=12) included patients with ATP6V1B2, PRRT2, KAT6B, KCNT1, SCN2A, CLCN4, and SCN8A variants (n=2), as well as 11q11.2 duplication, 16p11.2 deletion, and 8p11.23/8q22.3 duplication. The Lennox–Gastaut syndrome group (n=9) included patients with MECP2, HNRNPU, NEXMIF, MT-TL1, ALG13, and CACNA1A variants (n=2), as well as ring chromosome 20 and 15q11q13 microduplications. The developmental and epileptic encephalopathy group (n=5) included patients with WWOX, KCNQ2, HCN1, and DISP1 variants and 5q14.3 deletions. The severe myoclonic epilepsy of infancy group (n=3) included patients with SCN1A variants (n=2) and a PCDH19 variant.

The mean age was 7.17±5.42 years in the genetic epilepsy group and 7.15±5.27 years in the control group, with no significant between-group difference (P=0.987) (Table 1). The proportion of males was also similar between the genetic epilepsy and control groups (35.7% vs. 39.3%, respectively; P=0.697). In the genetic epilepsy group, 29 patients (69.0%) had developmental delays, primarily involving early childhood milestones. Intellectual disability was confirmed in 33 patients (78.6%) based on standardized cognitive assessments. Status epilepticus was observed in two patients (4.8%), and 22 patients (52.4%) achieved seizure-free status during follow-up (Table 1).

Table 1.

Clinical characteristics of the genetic epilepsy and control groups

2. Total brain volumetric differences

After adjustment for age and sex using ANCOVA, the genetic epilepsy group had significantly smaller total brain volumes than the control group (Table 2). Specifically, total brain volume was significantly lower in the genetic epilepsy group (adjusted mean difference, −88.451; P<0.001). Total cerebrum volume (adjusted mean difference, −80.226; P<0.001), total subcortical GM volume (adjusted mean difference, −3.372; P<0.001), and total cerebral WM volume (adjusted mean difference, −23.864; P<0.001) were also significantly lower in the genetic epilepsy group.

Table 2.

Total brain volumetric differences between children with genetic epilepsy and negative qualitative MRI findings and healthy controls

Total cerebral GM volume (adjusted mean difference, −55.581; P<0.001), total cortical GM volume (adjusted mean difference, −52.129; P<0.001), and total cerebellar volume (adjusted mean difference, −8.311; P=0.001) were also significantly smaller in the genetic epilepsy group. In contrast, total lateral ventricle volume was significantly larger in the genetic epilepsy group than in controls (adjusted mean difference, 2.390; P=0.018).

3. Subcortical volumetric differences

Significant reductions in subcortical structure volumes were observed in the genetic epilepsy group compared with controls (Table 3). Total lentiform nucleus volume was significantly smaller in the genetic epilepsy group (adjusted mean difference, −0.954; P=0.001), with significant bilateral reductions (left, P=0.001; right, P=0.001). Similarly, total caudate nucleus volume was significantly reduced (adjusted mean difference, −0.619; P<0.001), with reductions in both hemispheres (left, P<0.001; right, P=0.001). Total thalamic volume was also significantly lower in the genetic epilepsy group (adjusted mean difference, −1.202; P<0.001), with significant reductions on both the left and right sides (both P<0.001). In addition, significant bilateral reductions were observed in the hippocampus (adjusted mean difference, −0.422; P<0.001) and amygdala (adjusted mean difference, −0.198; P=0.004).

Table 3.

Subcortical volumetric differences between children with genetic epilepsy and negative qualitative magnetic resonance imaging findings and healthy controls

4. Lobar volumetric differences

Regional analysis revealed widespread reductions in lobar volumes in the genetic epilepsy group (Table 4). Total frontal lobe volume was significantly smaller in the genetic epilepsy group (adjusted mean difference, −19.299; P<0.001), with significant bilateral reductions. Similarly, total parietal lobe volume (adjusted mean difference, −13.008; P=0.001), total temporal lobe volume (adjusted mean difference, −10.514; P<0.001), and total occipital lobe volume (adjusted mean difference, −4.870; P<0.001) were significantly reduced. The genetic epilepsy group also showed significantly lower cingulate cortex volume (adjusted mean difference, −1.807; P<0.001) and insular volume (adjusted mean difference, −1.001; P=0.002), with consistent bilateral reductions.

Table 4.

Lobar volumetric differences between children with genetic epilepsy and negative qualitative magnetic resonance imaging findings and healthy controls

Discussion

In this study, children with genetic epilepsy and negative qualitative MRI findings had significantly smaller total and regional brain volumes than age- and sex-matched controls, despite the absence of visible structural abnormalities on conventional MRI. Specifically, total brain volume, total cerebral volume, cortical GM volume, subcortical GM volume, WM volume, and cerebellar volume were significantly reduced, whereas lateral ventricle volume was increased. Significant volumetric reductions were also observed in key subcortical structures, including the thalamus, caudate nucleus, lentiform nucleus, hippocampus, and amygdala, as well as in widespread cortical regions, including the frontal, parietal, temporal, and occipital lobes. These findings suggest that genetic epilepsy, even in the absence of visible MRI abnormalities, is associated with widespread structural brain alterations that may reflect abnormal neurodevelopment rather than focal structural lesions.

These findings are consistent with previous neuroimaging studies showing that epilepsy is associated with widespread structural abnormalities beyond the epileptogenic focus. The Enhancing Neuro Imaging Genetics through Meta-Analysis (ENIGMA) Consortium-Epilepsy study, which included more than 2,000 patients with epilepsy, reported significant volume reductions in subcortical structures, particularly the thalamus and hippocampus, supporting the concept of epilepsy as a network disorder involving distributed brain regions [14]. Similarly, structural brain alterations have been reported in pediatric patients with genetic epilepsy, including those with SCN1A mutations, with widespread cortical and subcortical volume reductions associated with impaired neurodevelopment [15]. The present results extend these findings by demonstrating similar volumetric alterations in children with genetically confirmed epilepsy who had no visible lesions on conventional MRI. However, the genetic epilepsy group in this study comprised a heterogeneous set of genes and syndromes, each with distinct pathophysiological mechanisms and developmental trajectories. Accordingly, gene-specific effects may vary, and these findings should be interpreted as reflecting shared neurodevelopmental vulnerability rather than gene-specific morphological characteristics.

The observed thalamic volume reduction may be clinically relevant because the thalamus plays a key role in regulating cortical excitability and synchronizing neuronal activity across thalamocortical networks involved in seizure generation and propagation [14]. Structural abnormalities in the hippocampus and amygdala, essential components of the limbic system, may contribute to seizure susceptibility and neurocognitive dysfunction. Genetic mutations associated with epilepsy can disrupt neuronal migration, synaptic function, and cortical network development, resulting in widespread structural alterations [3,5,7].

In addition to subcortical abnormalities, significant cortical volume reductions were observed across multiple lobes, including the frontal, temporal, parietal, and occipital lobes, as well as the cingulate and insular cortices. These findings support previous reports of cortical volume and thickness reductions in children with genetic epilepsy and developmental and epileptic encephalopathies [5,15]. These structural abnormalities likely reflect atypical brain development caused by genetic mutations affecting neurogenesis, neuronal differentiation, and synaptic connectivity [3,6]. This interpretation is consistent with the clinical characteristics of the cohort, in which most patients had developmental delay and intellectual disability, both of which are well-recognized features of genetic epilepsy [1,5].

Importantly, these volumetric alterations were detected using automated deep learning-based brain segmentation, highlighting the utility of quantitative MRI analysis for identifying subtle structural abnormalities that are not visible on conventional MRI. Quantitative morphometric analysis and automated volumetric techniques can detect structural abnormalities in patients with nonlesional epilepsy and may improve diagnostic accuracy [12,16]. Recent advances in deep learning-based brain segmentation have enabled reliable and efficient volumetric analysis, facilitating the identification of imaging biomarkers for neurological disorders, including epilepsy [17-20]. These findings support the potential role of automated volumetric analysis as an imaging biomarker for identifying children with genetic epilepsy and guiding genetic testing.

This study has several important clinical implications. First, quantitative volumetric analysis may serve as an objective imaging biomarker of genetic epilepsy, particularly in patients with normal qualitative MRI findings. Given that genetic testing remains costly and is not always readily available, imaging biomarkers may help identify patients at higher risk for genetic etiologies and guide genetic testing [3,7]. Second, the results suggest that genetic epilepsy may involve volumetric brain alterations even when qualitative MRI findings are normal and that these alterations are associated with widespread neurodevelopmental abnormalities, supporting the concept of epilepsy as a network disorder rather than a purely focal disease [14]. Third, early identification of structural brain abnormalities may provide valuable information for neurodevelopmental prognosis and clinical management.

This study had several limitations. First, it was a retrospective, single-center study with a relatively small sample size, which may limit the generalizability of the findings. Second, the genetic epilepsy group included patients with heterogeneous genetic mutations and epilepsy syndromes that may have different effects on brain development. Third, MRI data were acquired using different scanner field strengths, which may have introduced variability, although previous studies have demonstrated the reliability of automated volumetric analyses across different MRI platforms [19,20]. Future longitudinal studies with larger cohorts are needed to investigate the relationships among genetic mutations, structural brain alterations, and clinical outcomes.

In conclusion, children with genetic epilepsy showed significant total and regional brain volume reductions despite normal qualitative MRI findings. These findings support the potential use of automated volumetric MRI analysis as an imaging biomarker and provide supportive imaging evidence of neurodevelopmental abnormalities in genetic epilepsy.

Article information Continued