Use of complex networks in the detection of epileptic seizures and in the location of the epileptogenic zone in patients with epilepsy

Author(s)

Gustavo Henrique Tomanik

• Document type: Master's Dissertation
• Press: Botucatu. 2020-04-22.
• Institution: Universidade Estadual Paulista (Unesp). Instituto de Biociências. Botucatu
• Defense date: 2020-02-27
• Advisor: Andriana Susana Lopes de Oliveira Campanharo

Abstract

Epilepsy is a cerebral pathology characterized by the presence of epileptic seizures that affects around 1% of the entire world population. Epileptic seizures can cause loss of consciousness generating great impact on the lives of people who suffer from the disease. Currently, the ElectroEncefaloGraphy(EEG) technique is one of the most important diagnostic tests for the identification of epileptic seizures and events between seizures, called Intertictal Epileptiform Discharges (DEIs). The epileptogenic zone is the neuronal region responsible for generating seizures in patients with focal epilepsy, and in some cases, treatment consists of surgery to remove this region. Studies that relate time series to the theory of complex networks have gained great prominence and importance, showing that it is possible to map a time series in a complex network without great loss of information. In this sense, the objective of this work is to use the mapping proposed by Campanharo et al., using the concept of bins, in a novel application: in the automatic detection of epileptic seizures, of DEIs through the analysis of EEG signals of patients with epilepsy and the location of the epileptogenic zone using a non-invasive technique. The proposed methodology presented satisfactory performance in the identification of epileptic seizures and EIDs, characterized by a high variation in amplitude and frequency. Average sensitivity values 83,3% and 93,2% were reached in the detection of epileptic seizures and IEDs, respectively. In the identification of the epileptogenic zone, the methodology was not successful in identifying the DEIs, which were not characterized by the high variation in amplitude, and consequently, its location from the proposed methodology remained inconclusive. (AU)

• FAPESP's process: 18/02014-2 - The use of complex networks in the seizures prediction and in the Epileptogenic Zone location of epileptic patients
• Grantee: Gustavo Henrique Tomanik
• Support Opportunities: Scholarships in Brazil - Master