Epileptic Seizure Detection Using Geometric Features Extracted from SODP Shape of EEG Signals and AsyLnCPSO-GA

Area of Science:

• Neurology
• Biomedical Engineering
• Signal Processing

Background:

• Epilepsy is a neurological disorder characterized by abnormal brain electrical activity.
• Electroencephalogram (EEG) analysis is crucial for diagnosing epilepsy and detecting seizures.
• Accurate and efficient seizure detection methods are needed for clinical evaluation.

Purpose of the Study:

• To develop a hybrid optimization model for selecting optimal features from EEG signals for epilepsy detection.
• To improve the accuracy and efficiency of epileptic seizure detection using a combination of advanced algorithms.
• To establish an effective tool for clinicians to aid in the expeditious diagnosis of epilepsy.

Main Methods:

• Application of the second-order difference plot (SODP) method to derive ten geometric features from EEG signals across frequency bands (δ, θ, α, β).
• Development of a hybrid optimization algorithm, AsyLnCPSO-GA, integrating a modified PSO with asynchronous learning and a genetic algorithm for feature selection.
• Utilizing a naïve Bayesian classifier to identify epileptic seizures based on the selected optimal feature combinations.

Main Results:

• The proposed hybrid model achieved a classification accuracy of 95.35% using a tenfold cross-validation strategy.
• Optimal feature combinations derived from inter-frequency band analysis significantly improved seizure detection performance.
• The method demonstrated effectiveness in distinguishing between seizure and seizure-free EEG signals.

Conclusions:

• The developed AsyLnCPSO-GA hybrid model offers a highly accurate and effective approach for epileptic seizure detection.
• SODP analysis combined with advanced feature selection provides a robust method for EEG-based epilepsy diagnosis.
• This technique can assist clinicians in making faster and more reliable epilepsy diagnoses.