Entropy-driven deep learning framework for epilepsy detection using electro encephalogram signals.

Area of Science:

• Neurology
• Signal Processing
• Machine Learning

Background:

• Epilepsy is a common neurological disorder requiring accurate and early detection.
• Electroencephalogram (EEG) signals are crucial for diagnosing epilepsy but are susceptible to noise.
• Existing detection methods often lack robustness and accuracy.

Purpose of the Study:

• To develop a novel, robust, and effective method for automatic epilepsy detection from EEG signals.
• To integrate advanced entropy measures with deep learning for enhanced feature extraction and classification.
• To improve the accuracy and reliability of epilepsy diagnosis using EEG data.

Main Methods:

• EEG data pre-processing using adaptive wavelet denoising.
• Extraction of multivariate entropy features: Multiple Variable Permutation Entropy (mvMPE) and Multiple Variable Multi-Scale Fuzzy Entropy (mvMFE).
• Non-linear dimensionality reduction using Uniform Manifold Approximation and Projection (UMAP).
• Classification using a Residual Convolutional Neural Network (ResNet) integrated with Bi-Directional Long Short-Term Memory (Bi-LSTM).

Main Results:

• The proposed model achieved a classification accuracy of 94%, F1-Score of 96%, recall of 93%, specificity of 87.70%, and precision of 82.21%.
• Demonstrated superior performance compared to traditional epilepsy detection approaches.
• Successfully captured temporal dynamics and spatial features from EEG signals.

Conclusions:

• The combination of advanced entropy measures and deep learning architectures provides a powerful approach for epilepsy detection.
• The developed method offers a robust and accurate solution for early identification of epilepsy from EEG signals.
• This study underscores the potential of integrating signal processing techniques with deep learning for neurological disorder diagnosis.