Epilepsy diagnosis based on generalized feed forward neural network

Area of Science:

• Neurology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Epilepsy is a common neurological disorder affecting 0.5-2% of the population, characterized by recurrent seizures.
• Diagnosis of epilepsy can be challenging, necessitating advanced tools for accurate identification.
• Electroencephalogram (EEG) signals are crucial for understanding brain activity in epilepsy.

Purpose of the Study:

• To develop a simple, reliable, and economical clinical decision support system (DSS) for epilepsy diagnosis.
• To utilize a Neural Network (NN) approach for automated epilepsy detection.
• To evaluate the performance of the proposed DSS using real-world EEG data.

Main Methods:

• A generalized feed forward neural network (GFFNN) was designed as the core of the DSS.
• Eleven statistical parameters and 64 Fast Fourier Transform (FFT) features were extracted from EEG signals.
• The system was trained and validated using EEG data from the University of Bonn.

Main Results:

• The GFFNN achieved a 100% classification rate on the training dataset.
• The system demonstrated an 86.67% classification rate on the cross-validation dataset.
• The proposed DSS shows significant potential for accurate epilepsy diagnosis.

Conclusions:

• The developed Neural Network-based DSS is a promising tool for epilepsy diagnosis.
• The GFFNN approach offers a reliable and economical method for analyzing EEG signals.
• Further validation and implementation of this DSS could aid clinicians in epilepsy management.