Classification of Epileptic Seizure From EEG Signal Based on Hilbert Vibration Decomposition and Deep Learning

Area of Science:

• * Neuroscience
• * Biomedical Engineering
• * Signal Processing

Background:

• * Epilepsy is a neurological disorder characterized by recurrent seizures.
• * Accurate seizure classification from electroencephalogram (EEG) signals is crucial for diagnosis and treatment.
• * Traditional methods often face challenges in effectively analyzing complex EEG data.

Purpose of the Study:

• * To develop and evaluate a novel method for classifying epileptic seizures using EEG signals.
• * To leverage the power of convolutional neural networks (CNNs) for image-based seizure detection.
• * To combine Hilbert vibration decomposition (HVD) with CNNs for enhanced classification performance.

Main Methods:

• * Electroencephalogram (EEG) signals were decomposed into mono-components using Hilbert vibration decomposition (HVD), focusing on delta, theta, alpha, and beta rhythms.
• * Two-dimensional (2D) images were generated from the 1D decomposed mono-components via continuous wavelet transform (CWT).
• * A CNN architecture was employed to process these 2D images for feature extraction and seizure classification, utilizing a 5-fold cross-validation technique on the Bonn University EEG dataset.

Main Results:

• * The proposed method achieved high classification accuracy, reaching an average of 98.6%.
• * Excellent performance metrics were observed, including an average sensitivity of 97.2% and specificity of 100%.
• * The CNN model demonstrated robust and generalized performance in classifying epileptic seizures.

Conclusions:

• * The integration of HVD and CNN provides an efficient and effective approach for epileptic seizure classification from EEG signals.
• * The data-driven CNN model excels at processing the 2D image representations derived from EEG mono-components.
• * This technique offers a promising advancement in automated seizure detection and analysis.