Interictal Epileptiform Discharge Detection Using Multi-Head Deep Convolutional Neural Network.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Artificial Intelligence

Background:

• Interictal epileptiform discharges (IEDs) are crucial biomarkers in epilepsy, indicating cortical irritation between seizures.
• Manual interpretation of electroencephalogram (EEG) for IED detection is time-consuming and prone to inter-observer variability.
• Automated detection systems are needed to assist clinicians in identifying IEDs and predicting seizure recurrence.

Purpose of the Study:

• To develop and evaluate a novel deep learning approach for accurate and efficient automated detection of IEDs.
• To leverage unique morphological patterns from EEG sub-bands for improved IED identification.
• To reduce the reliance on manual EEG interpretation by physicians.

Main Methods:

• A novel deep learning model combining 1D local binary pattern symbolization with a regularized multi-head 1D convolutional neural network was developed.
• The model was trained to learn morphological patterns from different EEG sub-bands.
• The approach was validated using the Temple University Events corpus scalp EEG data.

Main Results:

• The proposed deep learning method achieved a promising F1-score of 87.18% for IED detection.
• The system demonstrated effective learning of unique morphological patterns from EEG sub-bands.
• The results indicate a significant advancement in automated IED detection capabilities.

Conclusions:

• The novel deep learning approach offers a promising solution for automated IED detection in epilepsy.
• This method has the potential to enhance clinical decision-making and reduce the burden of manual EEG analysis.
• Further validation on diverse EEG datasets is warranted to confirm generalizability.