Seizure prediction using EEG spatiotemporal correlation structure

James R. Williamson, Daniel Bliss, David W. Browne, Jaishree T. Narayanan

Research output: Contribution to journalArticle

83 Citations (Scopus)

Abstract

A seizure prediction algorithm is proposed that combines novel multivariate EEG features with patient-specific machine learning. The algorithm computes the eigenspectra of space-delay correlation and covariance matrices from 15?s blocks of EEG data at multiple delay scales. The principal components of these features are used to classify the patient's preictal or interictal state. This is done using a support vector machine (SVM), whose outputs are averaged using a running 15-minute window to obtain a final prediction score. The algorithm was tested on 19 of 21 patients in the Freiburg EEG data set who had three or more seizures, predicting 71 of 83 seizures, with 15 false predictions and 13.8. h in seizure warning during 448.3. h of interictal data. The proposed algorithm scales with the number of available EEG signals by discovering the variations in correlation structure among any given set of signals that correlate with seizure risk.

Original languageEnglish (US)
Pages (from-to)230-238
Number of pages9
JournalEpilepsy and Behavior
Volume25
Issue number2
DOIs
StatePublished - Oct 2012
Externally publishedYes

Fingerprint

Electroencephalography
Seizures

Keywords

  • Correlation structure
  • Eigenvalues
  • Electroencephalogram
  • Epilepsy
  • Machine learning
  • Multivariate features
  • Principal components
  • Seizure prediction
  • Support vector machines

ASJC Scopus subject areas

  • Clinical Neurology
  • Behavioral Neuroscience
  • Neurology

Cite this

Seizure prediction using EEG spatiotemporal correlation structure. / Williamson, James R.; Bliss, Daniel; Browne, David W.; Narayanan, Jaishree T.

In: Epilepsy and Behavior, Vol. 25, No. 2, 10.2012, p. 230-238.

Research output: Contribution to journalArticle

Williamson, James R. ; Bliss, Daniel ; Browne, David W. ; Narayanan, Jaishree T. / Seizure prediction using EEG spatiotemporal correlation structure. In: Epilepsy and Behavior. 2012 ; Vol. 25, No. 2. pp. 230-238.
@article{5cc5aedbbf2c49588c0127326e34e83c,
title = "Seizure prediction using EEG spatiotemporal correlation structure",
abstract = "A seizure prediction algorithm is proposed that combines novel multivariate EEG features with patient-specific machine learning. The algorithm computes the eigenspectra of space-delay correlation and covariance matrices from 15?s blocks of EEG data at multiple delay scales. The principal components of these features are used to classify the patient's preictal or interictal state. This is done using a support vector machine (SVM), whose outputs are averaged using a running 15-minute window to obtain a final prediction score. The algorithm was tested on 19 of 21 patients in the Freiburg EEG data set who had three or more seizures, predicting 71 of 83 seizures, with 15 false predictions and 13.8. h in seizure warning during 448.3. h of interictal data. The proposed algorithm scales with the number of available EEG signals by discovering the variations in correlation structure among any given set of signals that correlate with seizure risk.",
keywords = "Correlation structure, Eigenvalues, Electroencephalogram, Epilepsy, Machine learning, Multivariate features, Principal components, Seizure prediction, Support vector machines",
author = "Williamson, {James R.} and Daniel Bliss and Browne, {David W.} and Narayanan, {Jaishree T.}",
year = "2012",
month = "10",
doi = "10.1016/j.yebeh.2012.07.007",
language = "English (US)",
volume = "25",
pages = "230--238",
journal = "Epilepsy and Behavior",
issn = "1525-5050",
publisher = "Academic Press Inc.",
number = "2",

}

TY - JOUR

T1 - Seizure prediction using EEG spatiotemporal correlation structure

AU - Williamson, James R.

AU - Bliss, Daniel

AU - Browne, David W.

AU - Narayanan, Jaishree T.

PY - 2012/10

Y1 - 2012/10

N2 - A seizure prediction algorithm is proposed that combines novel multivariate EEG features with patient-specific machine learning. The algorithm computes the eigenspectra of space-delay correlation and covariance matrices from 15?s blocks of EEG data at multiple delay scales. The principal components of these features are used to classify the patient's preictal or interictal state. This is done using a support vector machine (SVM), whose outputs are averaged using a running 15-minute window to obtain a final prediction score. The algorithm was tested on 19 of 21 patients in the Freiburg EEG data set who had three or more seizures, predicting 71 of 83 seizures, with 15 false predictions and 13.8. h in seizure warning during 448.3. h of interictal data. The proposed algorithm scales with the number of available EEG signals by discovering the variations in correlation structure among any given set of signals that correlate with seizure risk.

AB - A seizure prediction algorithm is proposed that combines novel multivariate EEG features with patient-specific machine learning. The algorithm computes the eigenspectra of space-delay correlation and covariance matrices from 15?s blocks of EEG data at multiple delay scales. The principal components of these features are used to classify the patient's preictal or interictal state. This is done using a support vector machine (SVM), whose outputs are averaged using a running 15-minute window to obtain a final prediction score. The algorithm was tested on 19 of 21 patients in the Freiburg EEG data set who had three or more seizures, predicting 71 of 83 seizures, with 15 false predictions and 13.8. h in seizure warning during 448.3. h of interictal data. The proposed algorithm scales with the number of available EEG signals by discovering the variations in correlation structure among any given set of signals that correlate with seizure risk.

KW - Correlation structure

KW - Eigenvalues

KW - Electroencephalogram

KW - Epilepsy

KW - Machine learning

KW - Multivariate features

KW - Principal components

KW - Seizure prediction

KW - Support vector machines

UR - http://www.scopus.com/inward/record.url?scp=84867919804&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84867919804&partnerID=8YFLogxK

U2 - 10.1016/j.yebeh.2012.07.007

DO - 10.1016/j.yebeh.2012.07.007

M3 - Article

C2 - 23041171

AN - SCOPUS:84867919804

VL - 25

SP - 230

EP - 238

JO - Epilepsy and Behavior

JF - Epilepsy and Behavior

SN - 1525-5050

IS - 2

ER -