Multi-Channel Vision Transformer for Epileptic Seizure Prediction

Area of Science:

• Neurology
• Artificial Intelligence
• Biomedical Signal Processing

Background:

• Epilepsy is a neurological disorder characterized by recurrent seizures, affecting approximately 30% of patients who remain refractory to anti-seizure medication.
• Predicting seizures is crucial for patient safety, enabling timely interventions and injury prevention.
• Current seizure prediction methods often struggle with effectively analyzing complex spatio-temporal-spectral features in electroencephalogram (EEG) data.

Purpose of the Study:

• To develop and evaluate a novel Transformer-based approach, the Multi-channel Vision Transformer (MViT), for automated seizure prediction using multi-channel EEG data.
• To simultaneously learn spatio-temporal and spectral features from EEG signals for enhanced seizure prediction accuracy.
• To demonstrate the superiority of the MViT algorithm compared to existing state-of-the-art seizure prediction techniques.

Main Methods:

• Utilized continuous wavelet transform to convert time-series EEG signals into time-frequency representations (Scalograms).
• Employed a Multi-channel Vision Transformer (MViT) model, processing Scalograms by splitting them into fixed-size patches for input.
• Conducted extensive experiments on three benchmark EEG datasets to validate the MViT algorithm's performance.

Main Results:

• The MViT algorithm achieved an average prediction sensitivity of 99.80% for surface EEG data.
• For invasive EEG data, the MViT model demonstrated high prediction sensitivity, ranging from 90.28% to 91.15%.
• The proposed MViT approach significantly outperformed current state-of-the-art seizure prediction methods across the evaluated datasets.

Conclusions:

• The Multi-channel Vision Transformer (MViT) is a highly effective deep learning model for automated seizure prediction from multi-channel EEG.
• The MViT's ability to learn spatio-temporal-spectral features offers a significant advancement in epilepsy management.
• This approach holds promise for improving the clinical management of epilepsy by enabling accurate and timely seizure prediction.