EEG-Based Seizure Prediction Using Hybrid DenseNet-ViT Network with Attention Fusion.

Area of Science:

• Neurology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Epilepsy seizure prediction is crucial for improving patient quality of life.
• Current prediction methods require enhancement for greater precision.

Purpose of the Study:

• To introduce a novel hybrid deep learning architecture for epilepsy seizure prediction.
• To leverage the strengths of DenseNet and Vision Transformer (ViT) with an attention fusion layer.

Main Methods:

• EEG signals were preprocessed using short-time Fourier transform (STFT) for time-frequency analysis.
• A hybrid DenseNet-ViT model with an attention fusion layer was developed for end-to-end seizure prediction.
• The CHB-MIT dataset and leave-one-out cross-validation were used for evaluation.

Main Results:

• The proposed hybrid model demonstrated superior performance in seizure prediction.
• High accuracy and low redundancy were achieved, indicating effective feature amalgamation.
• The model successfully converted EEG signals into time-frequency matrices for analysis.

Conclusions:

• Combining DenseNet, ViT, and attention mechanisms significantly enhances epilepsy seizure prediction capabilities.
• The developed model facilitates more precise therapeutic interventions for individuals with epilepsy.
• This hybrid deep learning approach offers a promising direction for advanced seizure forecasting.