A Multivariate Approach for Patient-Specific EEG Seizure Detection Using Empirical Wavelet Transform

Area of Science:

• Signal Processing
• Biomedical Engineering
• Neurology

Background:

• Epileptic seizure detection from electroencephalogram (EEG) signals is crucial for patient care.
• Analyzing the multivariate oscillatory nature of EEG signals presents a complex challenge.

Purpose of the Study:

• To investigate the multivariate oscillatory nature of EEG signals using adaptive frequency scales for improved epileptic seizure detection.
• To develop and validate a novel method for enhanced EEG seizure detection.

Main Methods:

• A multivariate extension of the empirical wavelet transform (EWT) was applied to analyze joint instantaneous amplitudes and frequencies in adaptive scales.
• Features were extracted from multivariate EEG signal epochs and processed for seizure and seizure-free discrimination.
• The method was evaluated on the CHB-MIT scalp EEG database using a moving-window analysis.

Main Results:

• The proposed method achieved high performance metrics: 97.91% average sensitivity, 99.57% specificity, and 99.41% accuracy.
• These results surpass those of previously compared state-of-the-art methods on the same database.
• The method demonstrated efficient detection of long-duration epileptic seizure events in EEG recordings.

Conclusions:

• The developed method effectively utilizes the time-frequency plane for multivariate signals.
• Patient-specific models for EEG seizure detection were successfully built, enhancing diagnostic capabilities.