Variational mode decomposition and binary grey wolf optimization-based automated epilepsy seizure classification...

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Machine Learning

Background:

• Epilepsy diagnosis relies heavily on electroencephalogram (EEG) signal analysis.
• Accurate and automated seizure classification is crucial for effective patient management.
• Existing methods face challenges in feature extraction and selection from complex EEG data.

Purpose of the Study:

• To propose a robust framework for automated seizure classification using EEG signals.
• To leverage Variational Mode Decomposition (VMD) for effective EEG signal decomposition.
• To employ Binary Grey Wolf Optimization (BGWO) for optimal feature selection and enhance classification accuracy.

Main Methods:

• EEG signals were decomposed into band-limited intrinsic mode functions (BL-IMFs) using VMD.
• A comprehensive set of features (frequency, time, information theory) were extracted from BL-IMFs.
• BGWO was utilized to select an optimal subset of these features.
• Six supervised machine learning algorithms, including Bayesian Regularized Shallow Neural Networks (BR-SNNs), were employed for classification.

Main Results:

• The proposed VMD-BGWO framework demonstrated high performance across two public EEG databases (CHB-MIT and Bonn University).
• Maximum classification accuracies of 99.53% and 99.64% were achieved for 1s and 2s epochs, respectively, on database 1 using BR-SNNs.
• Accuracies of 99.79% and 99.84% were obtained for 1s and 2s epochs, respectively, on database 2.

Conclusions:

• The VMD-BGWO framework offers a highly accurate and efficient approach for seizure classification from EEG data.
• The combination of VMD for signal decomposition and BGWO for feature selection significantly improves diagnostic performance.
• This framework holds promise for developing advanced, automated tools for epilepsy diagnosis and monitoring.