Precise localization of the epileptogenic zone is pivotal for planning minimally invasive surgeries in drug-resistant epilepsy. Here, we present a graph neural network (GNN) framework that integrates interictal intracranial EEG features, electrode topology, and MRI features to automate epilepsy surgery planning. We retrospectively evaluated the model using leave-one-patient-out cross-validation on a dataset of 80 drug-resistant epilepsy patients treated at St. Anne's University Hospital (Brno, Czech Republic), comprising 31 patients with good postsurgical outcomes (Engel I) and 49 with poor outcomes (Engel II-IV). The GNN predictions demonstrated a significantly better (P < 0.05, Mann-Whitney-U test) area under the precision-recall curve in patients with good outcomes (area under the precision-recall curve: 0.69) compared with those with poor outcomes (area under the precision-recall curve: 0.33), indicating that the model captures clinically relevant targets in successful cases. In patients with poor outcomes, the graph neural network proposed alternative intervention sites that diverged from the original clinical plans, highlighting its potential to identify alternative therapeutic targets. We show that topology-aware GNNs significantly outperformed (P < 0.05, Wilcoxon signed-rank test) traditional neural networks while using the same intracranial EEG features, emphasizing the importance of incorporating implantation topology into predictive models. These findings uncover the potential of GNNs to automatically suggest targets for epilepsy surgery, which can assist the clinical team during the planning process.

Brno Epilepsy Center Department of Neurology Member of ERN EpiCARE St Anne's University Hospital Faculty of Medicine Masaryk University Brno 625 00 Czech Republic

Department of Biomedical Engineering Pratt School of Engineering Duke University Durham NC 27705 USA

Department of Neurology Duke University Medical Center Durham NC 27705 USA

Department of Neurology Mayo Clinic Rochester MN 55905 USA

Department of Physiology and Biomedical Engineering Mayo Clinic Rochester MN 55905 USA

Institute of Scientific Instruments The Czech Academy of Sciences Brno 612 00 Czech Republic

International Clinical Research Center St Anne's University Hospital Brno 602 00 Czech Republic

Zobrazit více v PubMed

Frauscher  B. Localizing the epileptogenic zone. Curr Opin Neurol. 2020;33(2):198–206. PubMed

Worrell  GA, Gardner  AB, Stead  SM, et al.  High-frequency oscillations in human temporal lobe: Simultaneous microwire and clinical macroelectrode recordings. Brain J Neurol. 2008;131(Pt 4):928–937. PubMed PMC

Brázdil  M, Pail  M, Halámek  J, et al.  Very high-frequency oscillations: Novel biomarkers of the epileptogenic zone. Ann Neurol. 2017;82(2):299–310. PubMed

Janca  R, Jezdik  P, Cmejla  R, et al.  Detection of interictal epileptiform discharges using signal envelope distribution modelling: Application to epileptic and non-epileptic intracranial recordings. Brain Topogr. 2015;28(1):172–183. PubMed

Travnicek  V, Klimes  P, Cimbalnik  J, et al.  Relative entropy is an easy-to-use invasive electroencephalographic biomarker of the epileptogenic zone. Epilepsia. 2023;64(4):962–972. PubMed

Cimbalnik  J, Klimes  P, Sladky  V, et al.  Multi-feature localization of epileptic foci from interictal, intracranial EEG. Clin Neurophysiol. 2019;130(10):1945–1953. PubMed PMC

Chybowski  B, Klimes  P, Cimbalnik  J, et al.  Timing matters for accurate identification of the epileptogenic zone. Clin Neurophysiol. 2024;161:1–9. PubMed

Klimes  P, Nejedly  P, Hrtonova  V, et al.  Interictal stereo-electroencephalography features below 45 Hz are sufficient for correct localization of the epileptogenic zone and postsurgical outcome prediction. Epilepsia. 2024;65:2935–2945. PubMed

Gunnarsdottir  KM, Li  A, Smith  RJ, et al.  Source-sink connectivity: A novel interictal EEG marker for seizure localization. Brain. 2022;145(11):3901–3915. PubMed PMC

Bernabei  JM, Sinha  N, Arnold  TC, et al.  Normative intracranial EEG maps epileptogenic tissues in focal epilepsy. Brain. 2022;145(6):1949–1961. PubMed PMC

Conrad  EC, Revell  AY, Greenblatt  AS, et al.  Spike patterns surrounding sleep and seizures localize the seizure-onset zone in focal epilepsy. Epilepsia. 2023;64(3):754–768. PubMed PMC

Říha  P, Doležalová  I, Mareček  R, et al.  Multimodal combination of neuroimaging methods for localizing the epileptogenic zone in MR-negative epilepsy. Sci Rep. 2022;12(1):15158. PubMed PMC

Li  X, Yu  T, Ren  Z, et al.  Localization of the epileptogenic zone by multimodal neuroimaging and high-frequency oscillation. Front Hum Neurosci. 2021;15:677840. PubMed PMC

Zhou  J, Cui  G, Hu  S, et al.  Graph neural networks: A review of methods and applications. AI Open. 2020;1:57–81.

Mohammed  AH, Cabrerizo  M, Pinzon  A, Yaylali  I, Jayakar  P, Adjouadi  M. Graph neural networks in EEG spike detection. Artif Intell Med. 2023;145:102663. PubMed

Lian  Q, Qi  Y, Pan  G, Wang  Y. Learning graph in graph convolutional neural networks for robust seizure prediction. J Neural Eng. 2020;17(3):035004. PubMed

Li  Z, Hwang  K, Li  K, Wu  J, Ji  T. Graph-generative neural network for EEG-based epileptic seizure detection via discovery of dynamic brain functional connectivity. Sci Rep. 2022;12(1):18998. PubMed PMC

Zhao  Y, Zhang  G, Dong  C, Yuan  Q, Xu  F, Zheng  Y. Graph attention network with focal loss for seizure detection on electroencephalography signals. Int J Neural Syst. 2021;31(7):2150027. PubMed

Díaz-Montiel  AA, Lankarany  M. Graph Representations of iEEG Data for Seizure Detection with Graph Neural Networks. bioRxiv. [Preprint] doi:10.1101/2023.06.02.543277 DOI

Jia  Z, Lin  Y, Wang  J, et al.  Graphsleepnet: Adaptive spatial-temporal graph convolutional networks for sleep stage classification. In: IJCAI'20: Twenty-Ninth International Joint Conference on Artificial Intelligence Yokohama Yokohama, Japan, January 7-15, 2021. International Joint Conferences on Artificial Intelligence Organization. 2020:1324-1330.

Jehi  L. Machine learning for precision epilepsy surgery. Epilepsy Curr. 2023;23(2):78–83. PubMed PMC

Engel  J. Surgical treatment of the epilepsies. 2nd edn. Raven Press; 1993.

Penny WD, Friston KJ, Ashburner JT, Kiebel SJ, Nichols TE, eds . Statistical Parametric Mapping: The Analysis of Functional Brain Images. 2006. Accessed 15 August 2024. https://shop.elsevier.com/books/statistical-parametric-mapping-the-analysis-of-functional-brain-images/penny/978-0-12-372560-8

Kleinberg  JM. Authoritative sources in a hyperlinked environment. Journal of the ACM (JACM). 1999;46:604–632.

Roehri  N, Pizzo  F, Lagarde  S, et al.  High-frequency oscillations are not better biomarkers of epileptogenic tissues than spikes. Ann Neurol. 2018;83(1):84–97. PubMed

von Ellenrieder  N, Andrade-Valença  LP, Dubeau  F, Gotman  J. Automatic detection of fast oscillations (40–200 Hz) in scalp EEG recordings. Clin Neurophysiol. 2012;123(4):670–680. PubMed PMC

von Ellenrieder  N, Frauscher  B, Dubeau  F, Gotman  J. Interaction with slow waves during sleep improves discrimination of physiologic and pathologic high-frequency oscillations (80–500 Hz). Epilepsia. 2016;57(6):869–878. PubMed

von Ellenrieder  N, Khoo  HM, Dubeau  F, Gotman  J. What do intracerebral electrodes measure?  Clin Neurophysiol. 2021;132(5):1105–1115. PubMed

Veličković  P, Cucurull  G, Casanova  A, Romero  A, Liò  P, Bengio  Y. Graph Attention Networks. arXiv.Org [Preprint]. 2017. http://arxiv.org/abs/1710.10903

Hrtonova  V, Nejedly  P, Travnicek  V, et al.  Metrics for evaluation of automatic epileptogenic zone localization in intracranial electrophysiology. Clin Neurophysiol. 2025;169:33–46. PubMed

Hess  M, Kromney  J. Robust confidence intervals for effect sizes: A comparative study of Cohen’s d and cliff’s Delta under non-normality and heterogeneous variances. In: Paper Presented at the Annual Meeting of the American Educational Research Association; 2004.

Bernabei  JM, Li  A, Revell  AY, et al.  Quantitative approaches to guide epilepsy surgery from intracranial EEG. Brain. 2023;146(6):2248–2258. PubMed PMC

Rusch  TK, Bronstein  MM, Mishra  S. A Survey on Oversmoothing in Graph Neural Networks. arXiv [Preprint]. doi:10.48550/arXiv.2303.10993 DOI

Kipf  TN, Welling  M. Semi-Supervised Classification with Graph Convolutional Networks. arXiv [Preprint]. doi:10.48550/arXiv.1609.02907 DOI

Guo  MH, Cai  JX, Liu  ZN, Mu  TJ, Martin  RR, Hu  SM. PCT: Point cloud transformer. Comput Vis Media. 2021;7(2):187–199.

Klimes  P, Cimbalnik  J, Brazdil  M, et al.  NREM sleep is the state of vigilance that best identifies the epileptogenic zone in the interictal electroencephalogram. Epilepsia. 2019;60(12):2404–2415. PubMed