SYNCHRONIZATION AND INTERDEPENDENCE IN INTRACRANIAL EEG RECORDINGS FROM EPILEPSY PATIENTS: A COMPARATIVE STUDY
Abstract number :
2.167
Submission category :
Year :
2003
Submission ID :
602
Source :
www.aesnet.org
Presentation date :
12/6/2003 12:00:00 AM
Published date :
Dec 1, 2003, 06:00 AM
Authors :
Hannes Osterhage, Florian Mormann, Thomas Kreuz, Christian E. Elger, Klaus Lehnertz Department of Epileptology, University of Bonn, Bonn, Germany; Helmholtz-Institute of Radiation and Nuclear Physics, University of Bonn, Bonn, Germany; John-von-Neumann In
During the last years a number of bivariate nonlinear EEG analysis techniques have been developed in the field of nonlinear dynamics allowing an improved characterization of synchronization and interdependence phenomena. Although previous studies have demonstrated the usefulness of the respective techniques to characterize the spatio-temporal dynamics of the epileptogenic process a direct comparison of the performance is still missing.
Intracranial multi-channel EEG recordings lasting up to days from 5 patients suffering from unilateral mesial temporal lobe epilepsy were analyzed. Selective amygdalohippocampectomy led to freedom of seizures in all patients. As characterizing measures we used the mutual false nearest neighbor parameter, the mutual predictability, the nonlinear interdependences as well as the cross-correlation sum. In order to check the performance of linear techniques nonlinear measures were accompanied by the cross-correlation function. Using a moving-window approach we investigated the performance of these measures with respect to (a) an interictal focus localization and (b) the detection of pre-seizure states.
None of the bivariate measures allowed to sufficiently localize the epileptic focus from interictal EEG recordings only. In contrast, apart from the cross-correlation sum all other nonlinear measures as well as the linear cross-correlation function allowed to detect a pre-seizure state, however, with a varying degree of success. In some patients the asymmetry of some of the nonlinear measures provided additional information as to the spatio-temporal interactions between the epileptic focus and other areas of the brain.
Apart from the cross-correlation sum all other nonlinear measures exhibited a similar performance with respect to the problems investigated here. In some cases the linear cross-correlation function yielded information comparable to that obtained from the nonlinear measures. Although a general assessment of the performance of the investigated methods can actually not be given they can nevertheless be regarded valuable for an improvement of the presurgical evaluation and for the detection of pre-seizure states.
[Supported by: Deutsche Forschungsgemeinschaft]