Deep Learning Approaches for Epileptic Seizure Prediction: A Review
- Resource Type
- Conference
- Authors
- Soliman, Seif; Fouad, Ahmed M.; Mourad, Emil; Hossam, Sara; Ehab, Mohamed; Selim, Sahar; Darweesh, M. Saeed
- Source
- 2022 4th Novel Intelligent and Leading Emerging Sciences Conference (NILES) Novel Intelligent and Leading Emerging Sciences Conference (NILES), 2022 4th. :01-06 Oct, 2022
- Subject
- Bioengineering
Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
Engineered Materials, Dielectrics and Plasmas
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Deep learning
Sensitivity
Frequency-domain analysis
Neural networks
Feature extraction
Prediction algorithms
Classification algorithms
Seizure prediction
Epilepsy
EEG
Preprocessing
Feature Extraction
Classification
Deep Learning (DL)
Convolutional Neural Network (CNN)
Long Short-Term Memory (LSTM)
- Language
Epilepsy is a chronic nervous disorder, which disturbs the normal daily routine of an epileptic patient due to sudden seizure onset that may cause loss of consciousness. Seizures are periods of aberrant brain activity patterns. Early prediction of an epileptic seizure is critical for those who suffer from it as it will give them time to prepare for an incoming seizure and alert anyone in their close circle of contacts to aid them. This has been an active field of study, powered by the decreasing cost of non-invasive electroencephalogram (EEG) collecting equipment and the rapid evolution of Deep Learning (DL) algorithms. This review paper offers the most recent evaluations of contemporary DL techniques for predicting epileptic seizures with a lot of emphasis on pre-processing, feature extraction and the classification techniques implemented many of which depend on Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) as well as the different datasets used. The study compares the claimed sensitivity and false alarm rate to conclude the described methodologies and their limitations.