Epileptic Disorder Detection of Seizures Using EEG Signals.

Area of Science:

• Neurology
• Biomedical Engineering
• Data Science

Background:

• Epilepsy is a neurological disorder diagnosed using electroencephalography (EEG).
• Existing EEG datasets are often imbalanced, with fewer epileptic than non-epileptic signals.
• This imbalance poses challenges for developing accurate diagnostic AI models.

Purpose of the Study:

• To develop a novel framework for integrating local EEG data from King Abdulaziz University hospital into the CHB-MIT dataset.
• To address data imbalance issues in epilepsy EEG datasets.
• To enhance the performance of deep learning models for seizure detection.

Main Methods:

• A compatibility framework was developed for EEG data integration.
• Key functions include dominant channel selection and a novel algorithm for reading XLtek EEG data.
• The integrated dataset was evaluated using 1D-CNN, Bi-LSTM, and attention-based deep learning models.

Main Results:

• The integrated dataset, using selective channels, achieved high performance metrics.
• Accuracy reached 96.87%, precision 96.98%, and sensitivity 96.85%.
• The proposed method outperformed existing systems utilizing more EEG channels.

Conclusions:

• The developed framework effectively integrates local EEG data, mitigating dataset imbalance.
• The enhanced dataset significantly improves the accuracy of deep learning models for epilepsy diagnosis.
• This approach offers a promising strategy for developing more robust AI tools in clinical neurology.