Improved decoding of neural activity from fMRI signals using non-separable spatiotemporal deconvolutions

Felix Bießmann, Yusuke Murayama, Nikos K. Logothetis, Klaus Muller, Frank C. Meinecke

Research output: Contribution to journalArticle

11 Citations (Scopus)

Abstract

The goal of most functional Magnetic Resonance Imaging (fMRI) analyses is to investigate neural activity. Many fMRI analysis methods assume that the temporal dynamics of the hemodynamic response function (HRF) to neural activation is separable from its spatial dynamics. Although there is empirical evidence that the HRF is more complex than suggested by space-time separable canonical HRF models, it is difficult to assess how much information about neural activity is lost when assuming space-time separability. In this study we directly test whether spatiotemporal variability in the HRF that is not captured by separable models contains information about neural signals. We predict intracranially measured neural activity from simultaneously recorded fMRI data using separable and non-separable spatiotemporal deconvolutions of voxel time series around the recording electrode. Our results show that abandoning the spatiotemporal separability assumption consistently improves the decoding accuracy of neural signals from fMRI data. We compare our findings with results from optical imaging and fMRI studies and discuss potential implications for classical fMRI analyses without invasive electrophysiological recordings.

Original languageEnglish
Pages (from-to)1031-1042
Number of pages12
JournalNeuroImage
Volume61
Issue number4
DOIs
Publication statusPublished - 2012 Jul 16

Fingerprint

Magnetic Resonance Imaging
Hemodynamics
Optical Imaging
Electrodes

Keywords

  • EEG-fMRI
  • Multivoxel pattern analysis
  • Neurovascular coupling
  • Spatiotemporal hemodynamic response function

ASJC Scopus subject areas

  • Cognitive Neuroscience
  • Neurology

Cite this

Bießmann, F., Murayama, Y., Logothetis, N. K., Muller, K., & Meinecke, F. C. (2012). Improved decoding of neural activity from fMRI signals using non-separable spatiotemporal deconvolutions. NeuroImage, 61(4), 1031-1042. https://doi.org/10.1016/j.neuroimage.2012.04.015

Improved decoding of neural activity from fMRI signals using non-separable spatiotemporal deconvolutions. / Bießmann, Felix; Murayama, Yusuke; Logothetis, Nikos K.; Muller, Klaus; Meinecke, Frank C.

In: NeuroImage, Vol. 61, No. 4, 16.07.2012, p. 1031-1042.

Research output: Contribution to journalArticle

Bießmann, F, Murayama, Y, Logothetis, NK, Muller, K & Meinecke, FC 2012, 'Improved decoding of neural activity from fMRI signals using non-separable spatiotemporal deconvolutions', NeuroImage, vol. 61, no. 4, pp. 1031-1042. https://doi.org/10.1016/j.neuroimage.2012.04.015
Bießmann, Felix ; Murayama, Yusuke ; Logothetis, Nikos K. ; Muller, Klaus ; Meinecke, Frank C. / Improved decoding of neural activity from fMRI signals using non-separable spatiotemporal deconvolutions. In: NeuroImage. 2012 ; Vol. 61, No. 4. pp. 1031-1042.
@article{ed13439052d644e79ab47f2f9accc078,
title = "Improved decoding of neural activity from fMRI signals using non-separable spatiotemporal deconvolutions",
abstract = "The goal of most functional Magnetic Resonance Imaging (fMRI) analyses is to investigate neural activity. Many fMRI analysis methods assume that the temporal dynamics of the hemodynamic response function (HRF) to neural activation is separable from its spatial dynamics. Although there is empirical evidence that the HRF is more complex than suggested by space-time separable canonical HRF models, it is difficult to assess how much information about neural activity is lost when assuming space-time separability. In this study we directly test whether spatiotemporal variability in the HRF that is not captured by separable models contains information about neural signals. We predict intracranially measured neural activity from simultaneously recorded fMRI data using separable and non-separable spatiotemporal deconvolutions of voxel time series around the recording electrode. Our results show that abandoning the spatiotemporal separability assumption consistently improves the decoding accuracy of neural signals from fMRI data. We compare our findings with results from optical imaging and fMRI studies and discuss potential implications for classical fMRI analyses without invasive electrophysiological recordings.",
keywords = "EEG-fMRI, Multivoxel pattern analysis, Neurovascular coupling, Spatiotemporal hemodynamic response function",
author = "Felix Bie{\ss}mann and Yusuke Murayama and Logothetis, {Nikos K.} and Klaus Muller and Meinecke, {Frank C.}",
year = "2012",
month = "7",
day = "16",
doi = "10.1016/j.neuroimage.2012.04.015",
language = "English",
volume = "61",
pages = "1031--1042",
journal = "NeuroImage",
issn = "1053-8119",
publisher = "Academic Press Inc.",
number = "4",

}

TY - JOUR

T1 - Improved decoding of neural activity from fMRI signals using non-separable spatiotemporal deconvolutions

AU - Bießmann, Felix

AU - Murayama, Yusuke

AU - Logothetis, Nikos K.

AU - Muller, Klaus

AU - Meinecke, Frank C.

PY - 2012/7/16

Y1 - 2012/7/16

N2 - The goal of most functional Magnetic Resonance Imaging (fMRI) analyses is to investigate neural activity. Many fMRI analysis methods assume that the temporal dynamics of the hemodynamic response function (HRF) to neural activation is separable from its spatial dynamics. Although there is empirical evidence that the HRF is more complex than suggested by space-time separable canonical HRF models, it is difficult to assess how much information about neural activity is lost when assuming space-time separability. In this study we directly test whether spatiotemporal variability in the HRF that is not captured by separable models contains information about neural signals. We predict intracranially measured neural activity from simultaneously recorded fMRI data using separable and non-separable spatiotemporal deconvolutions of voxel time series around the recording electrode. Our results show that abandoning the spatiotemporal separability assumption consistently improves the decoding accuracy of neural signals from fMRI data. We compare our findings with results from optical imaging and fMRI studies and discuss potential implications for classical fMRI analyses without invasive electrophysiological recordings.

AB - The goal of most functional Magnetic Resonance Imaging (fMRI) analyses is to investigate neural activity. Many fMRI analysis methods assume that the temporal dynamics of the hemodynamic response function (HRF) to neural activation is separable from its spatial dynamics. Although there is empirical evidence that the HRF is more complex than suggested by space-time separable canonical HRF models, it is difficult to assess how much information about neural activity is lost when assuming space-time separability. In this study we directly test whether spatiotemporal variability in the HRF that is not captured by separable models contains information about neural signals. We predict intracranially measured neural activity from simultaneously recorded fMRI data using separable and non-separable spatiotemporal deconvolutions of voxel time series around the recording electrode. Our results show that abandoning the spatiotemporal separability assumption consistently improves the decoding accuracy of neural signals from fMRI data. We compare our findings with results from optical imaging and fMRI studies and discuss potential implications for classical fMRI analyses without invasive electrophysiological recordings.

KW - EEG-fMRI

KW - Multivoxel pattern analysis

KW - Neurovascular coupling

KW - Spatiotemporal hemodynamic response function

UR - http://www.scopus.com/inward/record.url?scp=84861337769&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84861337769&partnerID=8YFLogxK

U2 - 10.1016/j.neuroimage.2012.04.015

DO - 10.1016/j.neuroimage.2012.04.015

M3 - Article

C2 - 22537598

AN - SCOPUS:84861337769

VL - 61

SP - 1031

EP - 1042

JO - NeuroImage

JF - NeuroImage

SN - 1053-8119

IS - 4

ER -