Enhanced Seizure Prediction Through CNN-LSTM Based Network

In order to enhance the prediction of epileptic seizures and improve the quality of life for affected individuals, this paper employs a combined methodology. Initially, significant features of electroencephalography (EEG) signals are extracted in the time-frequency domain using Short-Time Fourier Transform (STFT). Subsequently, Convolutional and Long-Short Term Memory (CNN-LSTM) based neural network are utilized to classify signals into preictal and interictal categories, enabling highly accurate seizure prediction. Results obtained from the CHB-MIT dataset demonstrate an ۹۸.۵۷% specificity and a ۰.۲۳ false prediction rate (FPR) per hour. This research highlights that advanced signal processing techniques and deep learning networks can significantly contribute to better seizure prediction and management, potentially leading to substantial improvements in the quality of life for individuals affected by this condition

Keywords:

Electroencephalography , Convolutional Neural Network , Epileptic Seizure , Long Short Term Memory