Robust Epileptic Seizure Detection Based on Biomedical Signals Using an Advanced Multi-View Deep Feature Learning...

Area of Science:

• Biomedical Signal Processing (BSP)
• Machine Learning (ML)
• Neurology

Background:

• Epilepsy is a neurological disorder causing seizures due to abnormal neuronal activity.
• Electroencephalogram (EEG) signals are crucial for monitoring epilepsy and detecting epileptic seizures (ES).
• Accurate ES detection relies on identifying key EEG features, necessitating domain expertise.

Purpose of the Study:

• To present an Advanced Multi-View Deep Feature Learning (AMV-DFL) framework for enhanced EEG feature detection in epilepsy.
• To improve the accuracy and interpretability of epileptic seizure detection using machine learning.

Main Methods:

• Extracted traditional time and frequency domain features (TMV-F) using Fast Fourier Transform (FFT) and raw EEG signals.
• Utilized one-dimensional convolutional neural networks (1D CNN) for autonomous deep feature extraction (MV-DF).
• Employed a Multi-View Forest (MV-F) classifier and Tree-based SHAP explainable AI (T-XAI) for classification and interpretation.

Main Results:

• The AMV-DFL framework demonstrated superior performance compared to traditional multi-view features and single-view deep features.
• Achieved a 4% improvement over models using TMV-FL and SV-DF.
• Outperformed other state-of-the-art methods by an average of 3% in classification accuracy.

Conclusions:

• The AMV-DFL approach offers a robust and generalized classification for epileptic seizure detection.
• This framework aids clinicians in identifying critical EEG features for epilepsy diagnosis.
• Potential for discovering novel biomarkers and improving epilepsy management through enhanced diagnostic capabilities.