Computational analysis of epileptic focus localization

Authors:
Evrim Acar;Canan A. Bingöl;Haluk Bingöl;Bülent Yener
Affiliations:
Department of Computer Science, Rensselaer Polytechnic Institute, Troy, NY;Department of Neurology, Yeditepe University, Istanbul, Turkey;Department of Computer Engineering, Bogaziçi University, Istanbul, Turkey;Department of Computer Science, Rensselaer Polytechnic Institute, Troy, NY
Venue:
BioMed'06 Proceedings of the 24th IASTED international conference on Biomedical engineering
Year:
2006

Citing 3
Cited 1

Characterizing nonlinearity in invasive EEG recordings from temporal lobe epilepsy

Physica D
Kernel principal component analysis

Advances in kernel methods
Kernel Methods for Pattern Analysis

Kernel Methods for Pattern Analysis

PARAFAC algorithms for large-scale problems

Neurocomputing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Epilepsy surgergy outcome strongly depends on the localization of epileptic focus. The analysis of ictal EEG (scalp or intracranial) is a gold standard for definition of localization of epileptic focus. In order to automate visual analysis of large amounts of EEG data, we examine the correlations among electrodes captured by linear, nonlinear and multilinear data analysis techniques. We study the performance of these statistical tools to understand the complex structure of epilepsy seizure and localize seizure origin. Our analysis results on four patients with temporal lobe epilepsy reveal that multiway (Tucker3 [1]) and nonlinear multiway (Kernelized Tucker3) analysis techniques are capable of capturing epileptic focus precisely when validated with clinical findings whereas linear and nonlinear analysis techniques (SVD, Kernel PCA [2]) fail to localize seizure origin.