Classifying event-related desynchronization in EEG, ECoG and MEG signals

N. Jeremy Hill, Thomas Navin Lal, Michael Schröder, Thilo Hinterberger, Guido Widman, Christian E. Elger, Bernhard Schölkopf, Niels Birbaumer

Research output: Chapter in Book/Report/Conference proceedingConference contribution

27 Citations (Scopus)

Abstract

We employed three different brain signal recording methods to perform Brain-Computer Interface studies on untrained subjects. In all cases, we aim to develop a system that could be used for fast, reliable preliminary screening in clinical BCI application, and we are interested in knowing how long screening sessions need to be. Good performance could be achieved, on average, after the first 200 trials in EEG, 75-100 trials in MEG, or 25-50 trials in ECoG. We compare the performance of Independent Component Analysis and the Common Spatial Pattern algorithm in each of the three sensor types, finding that spatial filtering does not help in MEG, helps a little in ECoG, and improves performance a great deal in EEG. In all cases the unsupervised ICA algorithm performed at least as well as the supervised CSP algorithm, which can suffer from poor generalization performance due to overfitting, particularly in ECoG and MEG.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages404-413
Number of pages10
Volume4174 LNCS
ISBN (Print)3540444122, 9783540444121
Publication statusPublished - 2006
Event28th Symposium of the German Association for Pattern Recognition, DAGM 2006 - Berlin, Germany
Duration: Sep 12 2006Sep 14 2006

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume4174 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other28th Symposium of the German Association for Pattern Recognition, DAGM 2006
CountryGermany
CityBerlin
Period9/12/069/14/06

Fingerprint

Desynchronization
Electroencephalography
Independent component analysis
Screening
Brain computer interface
Spatial Filtering
Overfitting
Brain
Spatial Pattern
Independent Component Analysis
Sensors
Sensor
Electroencephalogram

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Hill, N. J., Lal, T. N., Schröder, M., Hinterberger, T., Widman, G., Elger, C. E., ... Birbaumer, N. (2006). Classifying event-related desynchronization in EEG, ECoG and MEG signals. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 4174 LNCS, pp. 404-413). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4174 LNCS). Springer Verlag.

Classifying event-related desynchronization in EEG, ECoG and MEG signals. / Hill, N. Jeremy; Lal, Thomas Navin; Schröder, Michael; Hinterberger, Thilo; Widman, Guido; Elger, Christian E.; Schölkopf, Bernhard; Birbaumer, Niels.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4174 LNCS Springer Verlag, 2006. p. 404-413 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4174 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Hill, NJ, Lal, TN, Schröder, M, Hinterberger, T, Widman, G, Elger, CE, Schölkopf, B & Birbaumer, N 2006, Classifying event-related desynchronization in EEG, ECoG and MEG signals. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 4174 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 4174 LNCS, Springer Verlag, pp. 404-413, 28th Symposium of the German Association for Pattern Recognition, DAGM 2006, Berlin, Germany, 9/12/06.
Hill NJ, Lal TN, Schröder M, Hinterberger T, Widman G, Elger CE et al. Classifying event-related desynchronization in EEG, ECoG and MEG signals. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4174 LNCS. Springer Verlag. 2006. p. 404-413. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
Hill, N. Jeremy ; Lal, Thomas Navin ; Schröder, Michael ; Hinterberger, Thilo ; Widman, Guido ; Elger, Christian E. ; Schölkopf, Bernhard ; Birbaumer, Niels. / Classifying event-related desynchronization in EEG, ECoG and MEG signals. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4174 LNCS Springer Verlag, 2006. pp. 404-413 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{c75fea81820a4b2dbeceeece32772065,
title = "Classifying event-related desynchronization in EEG, ECoG and MEG signals",
abstract = "We employed three different brain signal recording methods to perform Brain-Computer Interface studies on untrained subjects. In all cases, we aim to develop a system that could be used for fast, reliable preliminary screening in clinical BCI application, and we are interested in knowing how long screening sessions need to be. Good performance could be achieved, on average, after the first 200 trials in EEG, 75-100 trials in MEG, or 25-50 trials in ECoG. We compare the performance of Independent Component Analysis and the Common Spatial Pattern algorithm in each of the three sensor types, finding that spatial filtering does not help in MEG, helps a little in ECoG, and improves performance a great deal in EEG. In all cases the unsupervised ICA algorithm performed at least as well as the supervised CSP algorithm, which can suffer from poor generalization performance due to overfitting, particularly in ECoG and MEG.",
author = "Hill, {N. Jeremy} and Lal, {Thomas Navin} and Michael Schr{\"o}der and Thilo Hinterberger and Guido Widman and Elger, {Christian E.} and Bernhard Sch{\"o}lkopf and Niels Birbaumer",
year = "2006",
language = "English",
isbn = "3540444122",
volume = "4174 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "404--413",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Classifying event-related desynchronization in EEG, ECoG and MEG signals

AU - Hill, N. Jeremy

AU - Lal, Thomas Navin

AU - Schröder, Michael

AU - Hinterberger, Thilo

AU - Widman, Guido

AU - Elger, Christian E.

AU - Schölkopf, Bernhard

AU - Birbaumer, Niels

PY - 2006

Y1 - 2006

N2 - We employed three different brain signal recording methods to perform Brain-Computer Interface studies on untrained subjects. In all cases, we aim to develop a system that could be used for fast, reliable preliminary screening in clinical BCI application, and we are interested in knowing how long screening sessions need to be. Good performance could be achieved, on average, after the first 200 trials in EEG, 75-100 trials in MEG, or 25-50 trials in ECoG. We compare the performance of Independent Component Analysis and the Common Spatial Pattern algorithm in each of the three sensor types, finding that spatial filtering does not help in MEG, helps a little in ECoG, and improves performance a great deal in EEG. In all cases the unsupervised ICA algorithm performed at least as well as the supervised CSP algorithm, which can suffer from poor generalization performance due to overfitting, particularly in ECoG and MEG.

AB - We employed three different brain signal recording methods to perform Brain-Computer Interface studies on untrained subjects. In all cases, we aim to develop a system that could be used for fast, reliable preliminary screening in clinical BCI application, and we are interested in knowing how long screening sessions need to be. Good performance could be achieved, on average, after the first 200 trials in EEG, 75-100 trials in MEG, or 25-50 trials in ECoG. We compare the performance of Independent Component Analysis and the Common Spatial Pattern algorithm in each of the three sensor types, finding that spatial filtering does not help in MEG, helps a little in ECoG, and improves performance a great deal in EEG. In all cases the unsupervised ICA algorithm performed at least as well as the supervised CSP algorithm, which can suffer from poor generalization performance due to overfitting, particularly in ECoG and MEG.

UR - http://www.scopus.com/inward/record.url?scp=33750225611&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33750225611&partnerID=8YFLogxK

M3 - Conference contribution

SN - 3540444122

SN - 9783540444121

VL - 4174 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 404

EP - 413

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

PB - Springer Verlag

ER -