Modeling low-frequency fluctuation and hemodynamic response timecourse in event-related fMRI

Kendrick N. Kay, Stephen David, Ryan J. Prenger, Kathleen A. Hansen, Jack L. Gallant

Research output: Contribution to journalArticle

36 Citations (Scopus)

Abstract

Functional magnetic resonance imaging (fMRI) suffers from many problems that make signal estimation difficult. These include variation in the hemodynamic response across voxels and low signal-to-noise ratio (SNR). We evaluate several analysis techniques that address these problems for event-related fMRI. (1) Many fMRI analyses assume a canonical hemodynamic response function, but this assumption may lead to inaccurate data models. By adopting the finite impulse response model, we show that voxel-specific hemodynamic response functions can be estimated directly from the data. (2) There is a large amount of low-frequency noise fluctuation (LFF) in blood oxygenation level dependent (BOLD) time-series data. To compensate for this problem, we use polynomials as regressors for LFF. We show that this technique substantially improves SNR and is more accurate than high-pass filtering of the data. (3) Model overfitting is a problem for the finite impulse response model because of the low SNR of the BOLD response. To reduce overfitting, we estimate a hemodynamic response timecourse for each voxel and incorporate the constraint of time-event separability, the constraint that hemodynamic responses across event types are identical up to a scale factor. We show that this technique substantially improves the accuracy of hemodynamic response estimates and can be computed efficiently. For the analysis techniques we present, we evaluate improvement in modeling accuracy via 10-fold cross-validation.

Original languageEnglish (US)
Pages (from-to)142-156
Number of pages15
JournalHuman Brain Mapping
Volume29
Issue number2
DOIs
StatePublished - Feb 2008
Externally publishedYes

Fingerprint

Hemodynamics
Magnetic Resonance Imaging
Signal-To-Noise Ratio
Noise

Keywords

  • Cross-validation
  • Hemodynamic response function
  • Low-frequency noise
  • Model evaluation
  • Reverse correlation

ASJC Scopus subject areas

  • Clinical Neurology
  • Neuroscience(all)
  • Radiological and Ultrasound Technology

Cite this

Modeling low-frequency fluctuation and hemodynamic response timecourse in event-related fMRI. / Kay, Kendrick N.; David, Stephen; Prenger, Ryan J.; Hansen, Kathleen A.; Gallant, Jack L.

In: Human Brain Mapping, Vol. 29, No. 2, 02.2008, p. 142-156.

Research output: Contribution to journalArticle

Kay, Kendrick N. ; David, Stephen ; Prenger, Ryan J. ; Hansen, Kathleen A. ; Gallant, Jack L. / Modeling low-frequency fluctuation and hemodynamic response timecourse in event-related fMRI. In: Human Brain Mapping. 2008 ; Vol. 29, No. 2. pp. 142-156.
@article{a288ddce36924afba4ebe7cd47d26cbc,
title = "Modeling low-frequency fluctuation and hemodynamic response timecourse in event-related fMRI",
abstract = "Functional magnetic resonance imaging (fMRI) suffers from many problems that make signal estimation difficult. These include variation in the hemodynamic response across voxels and low signal-to-noise ratio (SNR). We evaluate several analysis techniques that address these problems for event-related fMRI. (1) Many fMRI analyses assume a canonical hemodynamic response function, but this assumption may lead to inaccurate data models. By adopting the finite impulse response model, we show that voxel-specific hemodynamic response functions can be estimated directly from the data. (2) There is a large amount of low-frequency noise fluctuation (LFF) in blood oxygenation level dependent (BOLD) time-series data. To compensate for this problem, we use polynomials as regressors for LFF. We show that this technique substantially improves SNR and is more accurate than high-pass filtering of the data. (3) Model overfitting is a problem for the finite impulse response model because of the low SNR of the BOLD response. To reduce overfitting, we estimate a hemodynamic response timecourse for each voxel and incorporate the constraint of time-event separability, the constraint that hemodynamic responses across event types are identical up to a scale factor. We show that this technique substantially improves the accuracy of hemodynamic response estimates and can be computed efficiently. For the analysis techniques we present, we evaluate improvement in modeling accuracy via 10-fold cross-validation.",
keywords = "Cross-validation, Hemodynamic response function, Low-frequency noise, Model evaluation, Reverse correlation",
author = "Kay, {Kendrick N.} and Stephen David and Prenger, {Ryan J.} and Hansen, {Kathleen A.} and Gallant, {Jack L.}",
year = "2008",
month = "2",
doi = "10.1002/hbm.20379",
language = "English (US)",
volume = "29",
pages = "142--156",
journal = "Human Brain Mapping",
issn = "1065-9471",
publisher = "Wiley-Liss Inc.",
number = "2",

}

TY - JOUR

T1 - Modeling low-frequency fluctuation and hemodynamic response timecourse in event-related fMRI

AU - Kay, Kendrick N.

AU - David, Stephen

AU - Prenger, Ryan J.

AU - Hansen, Kathleen A.

AU - Gallant, Jack L.

PY - 2008/2

Y1 - 2008/2

N2 - Functional magnetic resonance imaging (fMRI) suffers from many problems that make signal estimation difficult. These include variation in the hemodynamic response across voxels and low signal-to-noise ratio (SNR). We evaluate several analysis techniques that address these problems for event-related fMRI. (1) Many fMRI analyses assume a canonical hemodynamic response function, but this assumption may lead to inaccurate data models. By adopting the finite impulse response model, we show that voxel-specific hemodynamic response functions can be estimated directly from the data. (2) There is a large amount of low-frequency noise fluctuation (LFF) in blood oxygenation level dependent (BOLD) time-series data. To compensate for this problem, we use polynomials as regressors for LFF. We show that this technique substantially improves SNR and is more accurate than high-pass filtering of the data. (3) Model overfitting is a problem for the finite impulse response model because of the low SNR of the BOLD response. To reduce overfitting, we estimate a hemodynamic response timecourse for each voxel and incorporate the constraint of time-event separability, the constraint that hemodynamic responses across event types are identical up to a scale factor. We show that this technique substantially improves the accuracy of hemodynamic response estimates and can be computed efficiently. For the analysis techniques we present, we evaluate improvement in modeling accuracy via 10-fold cross-validation.

AB - Functional magnetic resonance imaging (fMRI) suffers from many problems that make signal estimation difficult. These include variation in the hemodynamic response across voxels and low signal-to-noise ratio (SNR). We evaluate several analysis techniques that address these problems for event-related fMRI. (1) Many fMRI analyses assume a canonical hemodynamic response function, but this assumption may lead to inaccurate data models. By adopting the finite impulse response model, we show that voxel-specific hemodynamic response functions can be estimated directly from the data. (2) There is a large amount of low-frequency noise fluctuation (LFF) in blood oxygenation level dependent (BOLD) time-series data. To compensate for this problem, we use polynomials as regressors for LFF. We show that this technique substantially improves SNR and is more accurate than high-pass filtering of the data. (3) Model overfitting is a problem for the finite impulse response model because of the low SNR of the BOLD response. To reduce overfitting, we estimate a hemodynamic response timecourse for each voxel and incorporate the constraint of time-event separability, the constraint that hemodynamic responses across event types are identical up to a scale factor. We show that this technique substantially improves the accuracy of hemodynamic response estimates and can be computed efficiently. For the analysis techniques we present, we evaluate improvement in modeling accuracy via 10-fold cross-validation.

KW - Cross-validation

KW - Hemodynamic response function

KW - Low-frequency noise

KW - Model evaluation

KW - Reverse correlation

UR - http://www.scopus.com/inward/record.url?scp=39749174312&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=39749174312&partnerID=8YFLogxK

U2 - 10.1002/hbm.20379

DO - 10.1002/hbm.20379

M3 - Article

C2 - 17394212

AN - SCOPUS:39749174312

VL - 29

SP - 142

EP - 156

JO - Human Brain Mapping

JF - Human Brain Mapping

SN - 1065-9471

IS - 2

ER -