Intelligent Epileptic Seizure Detection and Classification Model Using Optimal Deep Canonical Sparse Autoencoder.

Area of Science:

• Neurology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Epileptic seizures are a chronic neurological disorder impacting brain function.
• Electroencephalogram (EEG) is a valuable, accessible tool for diagnosing brain disorders like epilepsy.
• Traditional epilepsy diagnosis is time-consuming; automated methods are needed for early detection.

Purpose of the Study:

• To develop an intelligent deep learning model for accurate epileptic seizure detection and classification using EEG signals.
• To enhance diagnostic efficiency and reduce the severity of epileptic seizures through early identification.
• To present a novel deep canonical sparse autoencoder-based epileptic seizure detection and classification (DCSAE-ESDC) model.

Main Methods:

• The DCSAE-ESDC model utilizes a deep canonical sparse autoencoder for seizure detection and classification.
• Optimal feature selection is performed using a novel coyote optimization algorithm (COA).
• Parameter tuning of the autoencoder model is achieved through the krill herd algorithm (KHA).

Main Results:

• The proposed DCSAE-ESDC technique demonstrated superior performance compared to existing methods.
• Achieved a maximum accuracy of 98.67% for binary classification of epileptic seizures.
• Achieved a maximum accuracy of 98.73% for multi-classification of epileptic seizures.

Conclusions:

• The DCSAE-ESDC model offers an effective and accurate automated solution for epileptic seizure detection and classification.
• The integration of COA for feature selection and KHA for parameter tuning enhances model performance.
• This deep learning approach holds significant potential for improving early diagnosis and management of epilepsy.