Determinant of Covariance Matrix Model Coupled with AdaBoost Classification Algorithm for EEG Seizure Detection

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Machine Learning for Healthcare

Background:

• Epilepsy diagnosis from electroencephalogram (EEG) is labor-intensive and time-consuming.
• Accurate and efficient EEG feature extraction is vital for clinical diagnosis.

Purpose of the Study:

• To develop an automated system for reducing EEG dimensionality and classifying seizure activity.
• To improve the accuracy and efficiency of epileptic seizure detection.

Main Methods:

• A determinant of covariance matrix (Cov-Det) model was used for EEG dimensionality reduction.
• Statistical features were extracted and ranked using Kolmogorov-Smirnov (KST) and Mann-Whitney U (MWUT) tests.
• An AdaBoost Back-Propagation neural network (AB_BP_NN) was employed for classifying EEG signals into seizure and non-seizure segments.

Main Results:

• The proposed AB_BP_NN achieved high accuracy (100% on Bern-Barcelona, 98.86% on Bonn dataset).
• The method demonstrated insignificant false positive rates and robustness in classification.
• The system offers a simpler design compared to classical machine learning techniques.

Conclusions:

• The developed Cov-Det and AB_BP_NN model provides an effective and efficient approach for automated epileptic seizure detection.
• This technique represents a significant improvement over existing state-of-the-art methods for EEG analysis.
• The findings suggest potential for widespread clinical application in epilepsy diagnosis.