Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Optimizing EEG Graph Structure for Seizure Detection: An Information Bottleneck and Self-Supervised Learning Approach

Lincan Li, Rikuto Kotoge, Xihao Piao, Zheng Chen, Yushun Dong

Abstract

Seizure detection from EEG signals is highly challenging due to complex spatiotemporal dynamics and extreme inter-patient variability. To model them, recent methods construct dynamic graphs via statistical correlations, predefined similarity measures, or implicit learning, yet rarely account for EEG's noisy nature. Consequently, these graphs usually contain redundant or task-irrelevant connections, undermining model performance even with state-of-the-art architectures. In this paper, we present a new perspective for EEG seizure detection: jointly learning denoised dynamic graph structures and informative spatial-temporal representations guided by the Information Bottleneck (IB). Unlike prior approaches, our graph constructor explicitly accounts for the noisy characteristics of EEG data, producing compact and reliable connectivity patterns that better support downstream seizure detection. To further enhance representation learning, we employ a self-supervised Graph Masked AutoEncoder that reconstructs masked EEG signals based on dynamic graph context, promoting structure-aware and compact representations aligned with the IB principle. Bringing things together, we introduce Information Bottleneck-guided EEG SeizuRE DetectioN via SElf-Supervised Learning (IRENE), which explicitly learns dynamic graph structures and interpretable spatial-temporal EEG representations. IRENE addresses three core challenges: (i) Identifying the most informative nodes and edges; (ii) Explaining seizure propagation in the brain network; and (iii) Enhancing robustness against label scarcity and inter-patient variability. Extensive experiments on benchmark EEG datasets demonstrate that our method outperforms state-of-the-art baselines in seizure detection and provides clinically meaningful insights into seizure dynamics. The source code is available at https://github.com/LabRAI/IRENE.

Optimizing EEG Graph Structure for Seizure Detection: An Information Bottleneck and Self-Supervised Learning Approach

Abstract

Seizure detection from EEG signals is highly challenging due to complex spatiotemporal dynamics and extreme inter-patient variability. To model them, recent methods construct dynamic graphs via statistical correlations, predefined similarity measures, or implicit learning, yet rarely account for EEG's noisy nature. Consequently, these graphs usually contain redundant or task-irrelevant connections, undermining model performance even with state-of-the-art architectures. In this paper, we present a new perspective for EEG seizure detection: jointly learning denoised dynamic graph structures and informative spatial-temporal representations guided by the Information Bottleneck (IB). Unlike prior approaches, our graph constructor explicitly accounts for the noisy characteristics of EEG data, producing compact and reliable connectivity patterns that better support downstream seizure detection. To further enhance representation learning, we employ a self-supervised Graph Masked AutoEncoder that reconstructs masked EEG signals based on dynamic graph context, promoting structure-aware and compact representations aligned with the IB principle. Bringing things together, we introduce Information Bottleneck-guided EEG SeizuRE DetectioN via SElf-Supervised Learning (IRENE), which explicitly learns dynamic graph structures and interpretable spatial-temporal EEG representations. IRENE addresses three core challenges: (i) Identifying the most informative nodes and edges; (ii) Explaining seizure propagation in the brain network; and (iii) Enhancing robustness against label scarcity and inter-patient variability. Extensive experiments on benchmark EEG datasets demonstrate that our method outperforms state-of-the-art baselines in seizure detection and provides clinically meaningful insights into seizure dynamics. The source code is available at https://github.com/LabRAI/IRENE.

Paper Structure

This paper contains 14 sections, 7 equations, 4 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Notations
  3. Methodology
  4. Model Architecture
  5. Information Bottleneck-Guided Dynamic Graph Construction
  6. Structure-Aware Graph Transformer for EEG Representation Learning
  7. Model Training Strategy and Loss Function
  8. Experimental Evaluation
  9. Experimental Settings
  10. Performance Evaluation
  11. Effectiveness of Graph Structure
  12. Graph Structure Interpretability
  13. Ablation Study and Component Analysis
  14. Conclusion

Figures (4)

  • Figure 1: The architecture of the proposed IRENE framework.
  • Figure 2: (a) ROC curves for seizure detection on 12s EEG clips. (b) Confusion matrices for each baseline model on 12s clip seizure classification. IRENE shows improved per-class prediction accuracy, particularly for the most challenging CT class.
  • Figure 3: Performance of IRENE when using different graph construction methods. We illustrate how various graph structures, including distance-based graph (D-$\mathcal{G}$ + IRENE), cross-correlation (CC-$\mathcal{G}$ + IRENE), temporal similarity-based graph (TS-$\mathcal{G}$ + IRENE), semantic similarity graph (SS-$\mathcal{G}$), and our IB-based dynamic graph, impact the model's F1-Scores.
  • Figure 4: Visualization of learned dynamic EEG graphs across different brain states. Each subfigure in each row shows a learned dynamic graph structure generated by a specific method. We have the following observations for IRENE constructed graphs: In the normal state, the graphs exhibit sparse and weakly connected structures. As seizure onset approaches, the graphs display stronger and more centralized connections, closely resemble to the ground-truth structure patterns and connectivity dynamics.