Abstracts

SEQUENTIAL MULTICHANNEL GABOR ATOM DENSITY FOR SEIZURE LOCALIZATION

Abstract number : 1.126
Submission category :
Year : 2002
Submission ID : 1577
Source : www.aesnet.org
Presentation date : 12/7/2002 12:00:00 AM
Published date : Dec 1, 2002, 06:00 AM

Authors :
Piotr J. Franaszczuk, Christophe C. Jouny, Gregory K. Bergey. Department of Neurology, Epilepsy Research Lab, Johns Hopkins University School of Medicine, Baltimore, MD

RATIONALE: Seizure localization is important in the evaluation of patients undergoing intensive video EEG monitoring. Analysis of ictal EEG can determine the location of seizure onset; at times intracranial recordings are necessary when scalp recordings are indeterminate. We present application of the Gabor Atom Density (GAD) method, derived from the matching pursuit decomposition, as a means of seizure localization.
METHODS: The matching pursuit (MP) method developed by Mallat and Zhang (1993) allows for continuous time-frequency decomposition of rapidly changing signals without requiring linear or nonlinear assumptions. The GAD method (Jouny et al., 2001) is derived from the MP method and provides a value that represents the number of atoms necessary to represent the signal for each point in time. We applied GAD analyses sequentially to each channel of ictal activity recorded intracranially from 10 patients with mesial temporal onset complex partial seizures, patients with well-defined electrographic seizure onsets.
RESULTS: The GAD analyses of partial seizures from each of the 10 patients revealed that at the time of ictal EEG onset, the earliest increases of GAD were seen at the area of seizure onset, based on visual analysis. The sequential multichannel GAD display also revealed the patterns and duration of ictal evolution and subsequential seizure propagation.
CONCLUSIONS: These results demonstrate that the application of sequential multichannel GAD analysis reveals that the earliest elevations in GAD levels are seen in the channels near the seizure focus. These levels reflect changes in signal complexity that accompany the electrographic seizure onset. In addition, the GAD analyses reveal signal changes that reflect seizure propagation. Therefore application of GAD analyses can provide a measure for seizure localization and onset potentially more consistent than visual analysis.
[Supported by: Funded by NIH grant NS 33732]