Random neural network based epileptic seizure episode detection exploiting electroencephalogram signals

Syed Yaseen Shah*, Hadi Larijani*, Ryan M. Gibson, Dimitrios Liarokapis

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

12 Citations (Scopus)
87 Downloads (Pure)

Abstract

Epileptic seizures are caused by abnormal electrical activity in the brain that manifests itself in a variety of ways, including confusion and loss of awareness. Correct identification of epileptic seizures is critical in the treatment and management of patients with epileptic disorders. One in four patients present resistance against seizures episodes and are in dire need of detecting these critical events through continuous treatment in order to manage the specific disease. Epileptic seizures can be identified by reliably and accurately monitoring the patients’ neuro and muscle activities, cardiac activity, and oxygen saturation level using state-of-the-art sensing techniques including electroencephalograms (EEGs), electromyography (EMG), electrocardiograms (ECGs), and motion or audio/video recording that focuses on the human head and body. EEG analysis provides a prominent solution to distinguish between the signals associated with epileptic episodes and normal signals; therefore, this work aims to leverage on the latest EEG dataset using cutting-edge deep learning algorithms such as random neural network (RNN), convolutional neural network (CNN), extremely random tree (ERT), and residual neural network (ResNet) to classify multiple variants of epileptic seizures from non-seizures. The results obtained highlighted that RNN outperformed all other algorithms used and provided an overall accuracy of 97%, which was slightly improved after cross validation.

Original languageEnglish
Article number2466
Number of pages17
JournalSensors
Volume22
Issue number7
Early online date23 Mar 2022
DOIs
Publication statusPublished - 1 Apr 2022

Keywords

  • machine learning
  • epilepsy
  • random neural network

ASJC Scopus subject areas

  • Analytical Chemistry
  • Information Systems
  • Atomic and Molecular Physics, and Optics
  • Biochemistry
  • Instrumentation
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Random neural network based epileptic seizure episode detection exploiting electroencephalogram signals'. Together they form a unique fingerprint.

Cite this