Classification of Resampled Pediatric Epilepsy EEG Data Using Artificial Neural Networks with Discrete Fourier Transforms

Abstract

Epilepsy is a neurological disorder commonly observed in children. Currently, electroencephalography (EEG) is widely used as the most important diagnostic method for epilepsy in medical practice. The diagnosis of epilepsy in pediatric patients is challenging due to their high level of activity and incomplete brain development. In this study, data sampled at 256 Hz were obtained from patients between the ages of 7–12, collected by Boston Children’s Hospital. First, the image intervals that contain seizure waves were identified in the datasets, and the discrete-time Fourier transform (DFT) was applied. The amplitude-frequency features of the frequency spectrum in seizure and nonseizure states were obtained, and patients were classified for seizure detection using a multilayer perceptron (MLP) based on an artificial neural network (ANN) architecture. In the next step, the EEG signals were resampled at low frequencies, and the same analyses were repeated to minimise the disadvantages of limiting factors such as storage space and processing power, resulting in reduced storage space usage and more efficient performance.

Keywords

• electroencephalography
• epileptic seizure
• discrete transforms
• machine learning