Automatic detection of ictal activity in EEG using synchronization and chaos-based attributes

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Existing automatic seizure onset detectors (SODs) are often complex and difficult for real-time implementation due to large, redundant feature sets.
• There is a need for simpler, more efficient SODs to improve the quality of life for epileptic patients by providing timely seizure alerts.

Purpose of the Study:

• To develop a simple and lightweight SOD that effectively detects seizures using minimal, relevant EEG features.
• To investigate the utility of EEG channel synchronization and signal chaoticity as indicators of seizure onset.

Main Methods:

• Utilized condition number to measure synchronization between EEG channels.
• Employed recurrence period density entropy to quantify the chaoticity of EEG signals.
• Trained and tested a support vector machine (SVM) on scalp EEG data from 10 patients.

Main Results:

• The proposed SOD achieved 100% sensitivity in detecting seizures.
• The system demonstrated a low false positive rate of 0.5 per hour.
• Synchronization and chaos features effectively reflect seizure manifestations in EEG data.

Conclusions:

• Simple, relevant features like synchronization and chaos can be sufficient for developing high-performing SODs.
• The developed lightweight SOD shows comparable performance to more complex systems, highlighting the potential for efficient real-time seizure detection.