Patient specific seizure prediction system using Hilbert spectrum and Bayesian networks classifiers.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Epilepsy affects millions globally, necessitating improved seizure prediction methods.
• Intracranial electroencephalogram (EEG) signals offer rich data for seizure forecasting.
• Current prediction systems often lack real-time viability and high accuracy.

Purpose of the Study:

• To develop an automated system for epileptic seizure prediction.
• To utilize Hilbert-Huang Transform (HHT) and Bayesian classifiers for enhanced prediction accuracy.
• To evaluate the system's performance on a real-world intracranial EEG database.

Main Methods:

• Signal decomposition using Hilbert-Huang Transform (HHT) to extract intrinsic mode functions (IMFs).
• Feature extraction from IMFs and correlation-based feature selection.
• Binary classification of preictal and interictal states using Bayesian networks.
• System training and testing on the Freiburg EEG database.

Main Results:

• Achieved 96.55% sensitivity with 0.21 false alarms per hour.
• Demonstrated 13.896% average proportion of time spent in warning.
• Reported an average detection latency of 33.21 minutes.
• Average testing time per EEG segment was 4.1451 seconds.

Conclusions:

• HHT-based features are effective for patient-specific epileptic seizure prediction from intracranial EEG.
• The developed system shows high sensitivity and a low false positive rate.
• The system's real-time processing capability makes it suitable for seizure control applications.