Spatiospectral Decomposition of Multi-subject EEG: Evaluating Blind Source Separation Algorithms on Real and Realistic Simulated Data

David A. Bridwell, Srinivas Rachakonda, Rogers F. Silva, Godfrey D. Pearlson, Vince Daniel Calhoun

Research output: Contribution to journalArticle

Abstract

Electroencephalographic (EEG) oscillations predominantly appear with periods between 1 s (1 Hz) and 20 ms (50 Hz), and are subdivided into distinct frequency bands which appear to correspond to distinct cognitive processes. A variety of blind source separation (BSS) approaches have been developed and implemented within the past few decades, providing an improved isolation of these distinct processes. Within the present study, we demonstrate the feasibility of multi-subject BSS for deriving distinct EEG spatiospectral maps. Multi-subject spatiospectral EEG decompositions were implemented using the EEGIFT toolbox (http://mialab.mrn.org/software/eegift/) with real and realistic simulated datasets (the simulation code is available at http://mialab.mrn.org/software/simeeg). Twelve different decomposition algorithms were evaluated. Within the simulated data, WASOBI and COMBI appeared to be the best performing algorithms, as they decomposed the four sources across a range of component numbers and noise levels. RADICAL ICA, ERBM, INFOMAX ICA, ICA EBM, FAST ICA, and JADE OPAC decomposed a subset of sources within a smaller range of component numbers and noise levels. INFOMAX ICA, FAST ICA, WASOBI, and COMBI generated the largest number of stable sources within the real dataset and provided partially distinct views of underlying spatiospectral maps. We recommend the multi-subject BSS approach and the selected algorithms for further studies examining distinct spatiospectral networks within healthy and clinical populations.

Original languageEnglish (US)
Pages (from-to)1-15
Number of pages15
JournalBrain Topography
DOIs
StateAccepted/In press - Feb 24 2016
Externally publishedYes

Fingerprint

Noise
Software
Population
Datasets

Keywords

  • Blind source separation
  • ICA
  • Multi-subject decomposition
  • Resting EEG
  • Simulated EEG
  • Wavelets

ASJC Scopus subject areas

  • Clinical Neurology
  • Anatomy
  • Neurology
  • Radiology Nuclear Medicine and imaging
  • Radiological and Ultrasound Technology

Cite this

Spatiospectral Decomposition of Multi-subject EEG : Evaluating Blind Source Separation Algorithms on Real and Realistic Simulated Data. / Bridwell, David A.; Rachakonda, Srinivas; Silva, Rogers F.; Pearlson, Godfrey D.; Calhoun, Vince Daniel.

In: Brain Topography, 24.02.2016, p. 1-15.

Research output: Contribution to journalArticle

@article{91898366232e44e5aa4e372c3fe45e89,
title = "Spatiospectral Decomposition of Multi-subject EEG: Evaluating Blind Source Separation Algorithms on Real and Realistic Simulated Data",
abstract = "Electroencephalographic (EEG) oscillations predominantly appear with periods between 1 s (1 Hz) and 20 ms (50 Hz), and are subdivided into distinct frequency bands which appear to correspond to distinct cognitive processes. A variety of blind source separation (BSS) approaches have been developed and implemented within the past few decades, providing an improved isolation of these distinct processes. Within the present study, we demonstrate the feasibility of multi-subject BSS for deriving distinct EEG spatiospectral maps. Multi-subject spatiospectral EEG decompositions were implemented using the EEGIFT toolbox (http://mialab.mrn.org/software/eegift/) with real and realistic simulated datasets (the simulation code is available at http://mialab.mrn.org/software/simeeg). Twelve different decomposition algorithms were evaluated. Within the simulated data, WASOBI and COMBI appeared to be the best performing algorithms, as they decomposed the four sources across a range of component numbers and noise levels. RADICAL ICA, ERBM, INFOMAX ICA, ICA EBM, FAST ICA, and JADE OPAC decomposed a subset of sources within a smaller range of component numbers and noise levels. INFOMAX ICA, FAST ICA, WASOBI, and COMBI generated the largest number of stable sources within the real dataset and provided partially distinct views of underlying spatiospectral maps. We recommend the multi-subject BSS approach and the selected algorithms for further studies examining distinct spatiospectral networks within healthy and clinical populations.",
keywords = "Blind source separation, ICA, Multi-subject decomposition, Resting EEG, Simulated EEG, Wavelets",
author = "Bridwell, {David A.} and Srinivas Rachakonda and Silva, {Rogers F.} and Pearlson, {Godfrey D.} and Calhoun, {Vince Daniel}",
year = "2016",
month = "2",
day = "24",
doi = "10.1007/s10548-016-0479-1",
language = "English (US)",
pages = "1--15",
journal = "Brain Topography",
issn = "0896-0267",
publisher = "Kluwer Academic/Human Sciences Press Inc.",

}

TY - JOUR

T1 - Spatiospectral Decomposition of Multi-subject EEG

T2 - Evaluating Blind Source Separation Algorithms on Real and Realistic Simulated Data

AU - Bridwell, David A.

AU - Rachakonda, Srinivas

AU - Silva, Rogers F.

AU - Pearlson, Godfrey D.

AU - Calhoun, Vince Daniel

PY - 2016/2/24

Y1 - 2016/2/24

N2 - Electroencephalographic (EEG) oscillations predominantly appear with periods between 1 s (1 Hz) and 20 ms (50 Hz), and are subdivided into distinct frequency bands which appear to correspond to distinct cognitive processes. A variety of blind source separation (BSS) approaches have been developed and implemented within the past few decades, providing an improved isolation of these distinct processes. Within the present study, we demonstrate the feasibility of multi-subject BSS for deriving distinct EEG spatiospectral maps. Multi-subject spatiospectral EEG decompositions were implemented using the EEGIFT toolbox (http://mialab.mrn.org/software/eegift/) with real and realistic simulated datasets (the simulation code is available at http://mialab.mrn.org/software/simeeg). Twelve different decomposition algorithms were evaluated. Within the simulated data, WASOBI and COMBI appeared to be the best performing algorithms, as they decomposed the four sources across a range of component numbers and noise levels. RADICAL ICA, ERBM, INFOMAX ICA, ICA EBM, FAST ICA, and JADE OPAC decomposed a subset of sources within a smaller range of component numbers and noise levels. INFOMAX ICA, FAST ICA, WASOBI, and COMBI generated the largest number of stable sources within the real dataset and provided partially distinct views of underlying spatiospectral maps. We recommend the multi-subject BSS approach and the selected algorithms for further studies examining distinct spatiospectral networks within healthy and clinical populations.

AB - Electroencephalographic (EEG) oscillations predominantly appear with periods between 1 s (1 Hz) and 20 ms (50 Hz), and are subdivided into distinct frequency bands which appear to correspond to distinct cognitive processes. A variety of blind source separation (BSS) approaches have been developed and implemented within the past few decades, providing an improved isolation of these distinct processes. Within the present study, we demonstrate the feasibility of multi-subject BSS for deriving distinct EEG spatiospectral maps. Multi-subject spatiospectral EEG decompositions were implemented using the EEGIFT toolbox (http://mialab.mrn.org/software/eegift/) with real and realistic simulated datasets (the simulation code is available at http://mialab.mrn.org/software/simeeg). Twelve different decomposition algorithms were evaluated. Within the simulated data, WASOBI and COMBI appeared to be the best performing algorithms, as they decomposed the four sources across a range of component numbers and noise levels. RADICAL ICA, ERBM, INFOMAX ICA, ICA EBM, FAST ICA, and JADE OPAC decomposed a subset of sources within a smaller range of component numbers and noise levels. INFOMAX ICA, FAST ICA, WASOBI, and COMBI generated the largest number of stable sources within the real dataset and provided partially distinct views of underlying spatiospectral maps. We recommend the multi-subject BSS approach and the selected algorithms for further studies examining distinct spatiospectral networks within healthy and clinical populations.

KW - Blind source separation

KW - ICA

KW - Multi-subject decomposition

KW - Resting EEG

KW - Simulated EEG

KW - Wavelets

UR - http://www.scopus.com/inward/record.url?scp=84959115818&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84959115818&partnerID=8YFLogxK

U2 - 10.1007/s10548-016-0479-1

DO - 10.1007/s10548-016-0479-1

M3 - Article

C2 - 26909688

AN - SCOPUS:84959115818

SP - 1

EP - 15

JO - Brain Topography

JF - Brain Topography

SN - 0896-0267

ER -