Automatic epileptic seizure detection based on EEG using a moth-flame optimization of one-dimensional convolutional...

Area of Science:

• Biomedical Engineering
• Computational Neuroscience

Background:

• Epileptic seizures can cause irreversible brain damage.
• Early detection and intervention are crucial for managing epilepsy.
• Current EEG analysis methods for seizure detection are often time-consuming and require manual hyperparameter tuning.

Purpose of the Study:

• To develop an automatic electroencephalogram (EEG) based epileptic seizure detection model.
• To optimize the detection model using moth-flame optimization (MFO).
• To address challenges of small-sample and single-channel EEG data.

Main Methods:

• Proposed a data augmentation technique for EEG signals.
• Utilized moth-flame optimization (MFO) to tune hyperparameters of a one-dimensional convolutional neural network (1D-CNN).
• Employed a softmax classifier for EEG classification.

Main Results:

• The model demonstrated feasibility in classifying five-class EEG data from the Bonn dataset, including healthy, preictal, and ictal states.
• The MFO-optimized 1D-CNN model outperformed other advanced optimization algorithms like PSO, GA, and GWO.
• Achieved accurate classification of EEG signals for epileptic seizure detection.

Conclusions:

• The proposed MFO-optimized 1D-CNN model offers an efficient and automatic approach for epileptic seizure detection.
• This model can be integrated into clinical applications for real-time seizure detection systems.
• The study highlights the potential of MFO in enhancing deep learning models for biomedical signal processing.