Open AccessTemporal epilepsy seizures monitoring and prediction using cross‐correlation and chaos theory
Author(s) -
Haddad Tahar,
BenHamida Naim,
Talbi Larbi,
Lakhssassi Ahmed,
Aouini Sadok
Publication year - 2014
Publication title -
healthcare technology letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.45
H-Index - 19
ISSN - 2053-3713
DOI - 10.1049/htl.2013.0010
Subject(s) - correlation , epilepsy , correlation dimension , electroencephalography , hippocampal formation , pattern recognition (psychology) , entropy (arrow of time) , lyapunov exponent , spectral density , physics , statistics , mathematics , artificial intelligence , neuroscience , computer science , psychology , chaotic , fractal dimension , fractal , mathematical analysis , geometry , quantum mechanics
Temporal seizures due to hippocampal origins are very common among epileptic patients. Presented is a novel seizure prediction approach employing correlation and chaos theories. The early identification of seizure signature allows for various preventive measures to be undertaken. Electro‐encephalography signals are spectrally broken down into the following sub‐bands: delta; theta; alpha; beta; and gamma. The proposed approach consists of observing a high correlation level between any pair of electrodes for the lower frequencies and a decrease in the Lyapunov index (chaos or entropy) for the higher frequencies. Power spectral density and statistical analysis tools were used to determine threshold levels for the lower frequencies. After studying all five sub‐bands, the analysis has revealed that the seizure signature can be extracted from the delta band and the high frequencies. High frequencies are defined as both the gamma band and the ripples occurring within the 60–120 Hz sub‐band. To validate the proposed approach, six patients from both sexes and various age groups with temporal epilepsies originating from the hippocampal area were studied using the Freiburg database. An average seizure prediction of 30 min, an anticipation accuracy of 72%, and a false‐positive rate of 0% were accomplished throughout 200 h of recording time.