Epilepsy Seizure Prediction on EEG Using Common Spatial Pattern and Convolutional Neural Network.

Area of Science:

• Neurology
• Biomedical Engineering
• Machine Learning

Background:

• Epilepsy seizure prediction is crucial for patient intervention but challenging due to the similarity between pre-ictal and inter-ictal states.
• Limited research exists for seizure prediction compared to seizure detection.
• Distinguishing pre-ictal from inter-ictal states in EEG signals is a significant hurdle.

Purpose of the Study:

• To propose a novel solution for epilepsy seizure prediction.
• To address the challenge of distinguishing between pre-ictal and inter-ictal states.
• To improve the accuracy and efficiency of seizure prediction systems.

Main Methods:

• Generating artificial pre-ictal electroencephalogram (EEG) signals to balance data.
• Employing wavelet packet decomposition and common spatial pattern (CSP) for feature extraction.
• Utilizing a shallow convolutional neural network (CNN) for classification.

Main Results:

• Achieved a sensitivity of 92.2% and a false prediction rate of 0.12/h on the Boston Children's Hospital-MIT scalp EEG dataset.
• The proposed method demonstrated superior performance compared to existing state-of-the-art approaches.
• Feature extraction improved accuracy and reduced computational time.

Conclusions:

• The combined use of CSP and CNN offers a promising approach for accurate epilepsy seizure prediction.
• The developed method effectively distinguishes between pre-ictal and inter-ictal states.
• This technique has the potential to provide timely warnings for epilepsy patients, enabling proactive interventions.