Epileptic Seizure Detection Using Deep CNNs with Spectral and Complexity Features from EEG Signals

Journal Publication · 2025

We present a clinically oriented deep learning approach for automated detection of epileptic seizures from scalp electroencephalography (EEG) recordings. The proposed method integrates domain-specific features, including spectral band-power and signal complexity measures such as Shannon entropy and Hjorth parameters, into a compact one-dimensional convolutional neural network (CNN). We evaluate the model on the Children's Hospital Boston and Massachusetts Institute of Technology (CHB-MIT) pediatric EEG dataset. While the model achieves high discrimination, with an area under the ROC curve of about 0.98 and accuracy of about 99% on held-out balanced data, the default decision threshold of 0.5 yields poor seizure recall. By lowering the threshold to 0.10 based on validation analysis, seizure sensitivity improves to about 91%, with a manageable false-alarm rate of about two to three per hour. Beyond accuracy, we report cross-validation with confidence intervals, event-based detection metrics, and ablation studies comparing spectral, complexity, and raw EEG inputs.

📊 Key Metrics

0.98

AUC-ROC

99%

Accuracy

91%

Sensitivity

2-3/hour

False Alarm Rate

🎯 Problem Statement

Epileptic seizure detection in clinical settings requires continuous monitoring of EEG signals, which is labor-intensive and prone to human error. Automated systems are needed to provide real-time, accurate seizure detection with minimal false alarms to support clinical decision-making.

🔬 Methodology

We developed a deep learning approach that combines domain-specific feature engineering with a compact 1D CNN. The model integrates spectral band-power features and signal complexity measures (Shannon entropy, Hjorth parameters) extracted from EEG signals. We evaluated on the CHB-MIT pediatric EEG dataset, using cross-validation and careful threshold tuning to optimize clinical utility.

💡 Key Contributions

→Clinically-oriented approach with optimized decision threshold (0.10) for high sensitivity
→Integration of spectral and complexity features for robust seizure detection
→Comprehensive evaluation with cross-validation, confidence intervals, and event-based metrics
→Achieved 91% sensitivity with manageable false-alarm rate (2-3/hour)

🧪 Experiments & Evaluation

Trained on CHB-MIT pediatric EEG dataset with multiple patients
Conducted ablation studies comparing spectral, complexity, and raw EEG inputs
Performed cross-validation with confidence intervals
Evaluated event-based detection metrics for clinical relevance
Optimized decision threshold through validation analysis

📈 Results

The model achieved an AUC-ROC of 0.98 and accuracy of 99% on balanced test data. By optimizing the decision threshold to 0.10 (instead of default 0.5), we improved seizure sensitivity to 91% while maintaining a manageable false-alarm rate of 2-3 per hour, making it suitable for clinical deployment.

🚀 Future Work

Real-time deployment in clinical monitoring systems
Extension to other neurological conditions
Integration with wearable EEG devices
Multi-modal fusion with other physiological signals

📚 Citation (BibTeX)

@article{ejigu2025epileptic,
  title={Epileptic Seizure Detection Using Deep CNNs with Spectral and Complexity Features from EEG Signals},
  author={Ejigu, Yohannes Ayana and Tadesse, Surafel Amsalu and Asfaw, Tesfa Tegegne and Tegegne, Anteneh Yehalem and Chekol, Adane Kassie},
  journal={Journal Publication},
  year={2025},
  note={AUC: 0.98, Accuracy: 99%, Sensitivity: 91%}
}