Automatic Detection of Ictal Activity in EEG Using Synchronization and Chaos-Based Attributes

Epileptic patients are at risk of experiencing unexpected and random seizures and this lowers their quality of life (QoL) considerably. Automatic seizure onset detectors (SODs) have been proposed to alert the patients before a seizure and in turn improve their QoL. SODs can aid doctors in the diagnosis of epilepsy and detection of seizures which is usually tiring and time consuming; the process requires examining the brain’s electrical activity recorded in an Electroencephalogram (EEG) record. However, the SODs developed in literature are computationally extensive and may not be suitable for clinical adoption. The aim of this thesis is to produce a simple and lightweight SOD that uses characteristics that reflect the neuronal behavior during a seizure. It has been shown that during a seizure, the neurons fire in a synchronized manner and the electrical activity is less chaotic. Hence, two features, the synchronization between EEG channels and the chaoticity of the EEG, have been used to build a patient-specific SOD. Synchronization was measured by the condition number while the recurrence period density entropy estimated the chaoticity of an EEG signal. The extracted features were fed to a support vector machine (SVM) and a neural network classifier for seizure detection. The detectors were validated using a scalp EEG dataset and the SVM-based SOD was able to detect the considered seizures with a sensitivity of 100% and a false positives rate of 0.5 per hour. The results indicate that synchronization and chaos attributes can effectively be used for seizure detection. This work emphasizes the importance of investigating characteristics that are reflective of seizures before building an SOD.