Effective Early Detection of Epileptic Seizures through EEG Signals Using Classification Algorithms Based on...

Area of Science:

• Neurology
• Biomedical Engineering
• Machine Learning

Background:

• Epilepsy is a neurological disorder characterized by recurrent seizures due to abnormal brain activity.
• Visual inspection of electroencephalogram (EEG) data for epilepsy diagnosis is time-consuming and subjective.
• Automated computer-aided diagnosis systems are crucial for efficient and reliable EEG analysis.

Purpose of the Study:

• To develop an effective automated approach for the early detection of epilepsy from EEG signals.
• To enhance the accuracy and efficiency of epilepsy diagnosis through advanced feature extraction and classification techniques.

Main Methods:

• Feature extraction using Discrete Wavelet Transform (DWT).
• Dimensionality reduction via Principal Component Analysis (PCA) and t-distributed Stochastic Neighbor Embedding (t-SNE).
• Classification using K-means clustering combined with PCA/t-SNE, followed by Extreme Gradient Boosting, K-NN, Decision Tree, Random Forest, and MLP classifiers.

Main Results:

• The Random Forest (RF) classifier with DWT and PCA achieved 97.96% accuracy.
• The RF classifier with DWT and t-SNE reached 98.09% accuracy.
• The Multilayer Perceptron (MLP) classifier with PCA + K-means demonstrated superior performance with 98.98% accuracy, 99.16% precision, 95.69% recall, and 97.4% F1 score.

Conclusions:

• The proposed automated approach significantly improves early epilepsy detection accuracy compared to existing methods.
• The combination of DWT, PCA/t-SNE, and advanced classifiers like MLP offers a robust solution for EEG-based epilepsy diagnosis.
• This automated system has the potential to aid clinicians in faster and more accurate epilepsy diagnosis.