The physiological plausibility of time-varying Granger-causal modeling: Normalization and weighting by spectral power

Gijs Plomp, Charles Quairiaux, Christoph M. Michel, Laura Astolfi

Research output: Contribution to journalArticle

40 Citations (Scopus)

Abstract

Time-varying connectivity methods are increasingly used to study directed interactions between brain regions from electrophysiological signals. These methods often show good results in simulated data but it is unclear to what extent connectivity results obtained from real data are physiologically plausible. Here we introduce a benchmark approach using multichannel somatosensory evoked potentials (SEPs) measured across rat cortex, where the structural and functional connectivity is relatively simple and well-understood. Rat SEPs to whisker stimulation are exclusively initiated by contralateral primary sensory cortex (S1), at known latencies, and with activity spread from S1 to specific cortical regions. This allows for a comparison of time-varying connectivity measures according to fixed criteria. We thus evaluated the performance of time-varying Partial Directed Coherence (PDC) and the Directed Transfer Function (DTF), comparing row- and column-wise normalization and the effect of weighting by the power spectral density (PSD). The benchmark approach revealed clear differences between methods in terms of physiological plausibility, effect size and temporal resolution. The results provide a validation of time-varying directed connectivity methods in an animal model and suggest a driving role for ipsilateral S1 in the later part of the SEP. The benchmark SEP dataset is made freely available.

Original languageEnglish
Pages (from-to)206-216
Number of pages11
JournalNeuroImage
Volume97
DOIs
Publication statusPublished - Aug 15 2014

Fingerprint

Somatosensory Evoked Potentials
Benchmarking
Vibrissae
Animal Models
Brain

Keywords

  • Barrel cortex
  • Dynamics
  • EEG
  • Effective connectivity
  • Functional networks
  • Granger causality

ASJC Scopus subject areas

  • Cognitive Neuroscience
  • Neurology
  • Medicine(all)

Cite this

The physiological plausibility of time-varying Granger-causal modeling : Normalization and weighting by spectral power. / Plomp, Gijs; Quairiaux, Charles; Michel, Christoph M.; Astolfi, Laura.

In: NeuroImage, Vol. 97, 15.08.2014, p. 206-216.

Research output: Contribution to journalArticle

@article{a47b1f7d71544d349a071a3435d243ae,
title = "The physiological plausibility of time-varying Granger-causal modeling: Normalization and weighting by spectral power",
abstract = "Time-varying connectivity methods are increasingly used to study directed interactions between brain regions from electrophysiological signals. These methods often show good results in simulated data but it is unclear to what extent connectivity results obtained from real data are physiologically plausible. Here we introduce a benchmark approach using multichannel somatosensory evoked potentials (SEPs) measured across rat cortex, where the structural and functional connectivity is relatively simple and well-understood. Rat SEPs to whisker stimulation are exclusively initiated by contralateral primary sensory cortex (S1), at known latencies, and with activity spread from S1 to specific cortical regions. This allows for a comparison of time-varying connectivity measures according to fixed criteria. We thus evaluated the performance of time-varying Partial Directed Coherence (PDC) and the Directed Transfer Function (DTF), comparing row- and column-wise normalization and the effect of weighting by the power spectral density (PSD). The benchmark approach revealed clear differences between methods in terms of physiological plausibility, effect size and temporal resolution. The results provide a validation of time-varying directed connectivity methods in an animal model and suggest a driving role for ipsilateral S1 in the later part of the SEP. The benchmark SEP dataset is made freely available.",
keywords = "Barrel cortex, Dynamics, EEG, Effective connectivity, Functional networks, Granger causality",
author = "Gijs Plomp and Charles Quairiaux and Michel, {Christoph M.} and Laura Astolfi",
year = "2014",
month = "8",
day = "15",
doi = "10.1016/j.neuroimage.2014.04.016",
language = "English",
volume = "97",
pages = "206--216",
journal = "NeuroImage",
issn = "1053-8119",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - The physiological plausibility of time-varying Granger-causal modeling

T2 - Normalization and weighting by spectral power

AU - Plomp, Gijs

AU - Quairiaux, Charles

AU - Michel, Christoph M.

AU - Astolfi, Laura

PY - 2014/8/15

Y1 - 2014/8/15

N2 - Time-varying connectivity methods are increasingly used to study directed interactions between brain regions from electrophysiological signals. These methods often show good results in simulated data but it is unclear to what extent connectivity results obtained from real data are physiologically plausible. Here we introduce a benchmark approach using multichannel somatosensory evoked potentials (SEPs) measured across rat cortex, where the structural and functional connectivity is relatively simple and well-understood. Rat SEPs to whisker stimulation are exclusively initiated by contralateral primary sensory cortex (S1), at known latencies, and with activity spread from S1 to specific cortical regions. This allows for a comparison of time-varying connectivity measures according to fixed criteria. We thus evaluated the performance of time-varying Partial Directed Coherence (PDC) and the Directed Transfer Function (DTF), comparing row- and column-wise normalization and the effect of weighting by the power spectral density (PSD). The benchmark approach revealed clear differences between methods in terms of physiological plausibility, effect size and temporal resolution. The results provide a validation of time-varying directed connectivity methods in an animal model and suggest a driving role for ipsilateral S1 in the later part of the SEP. The benchmark SEP dataset is made freely available.

AB - Time-varying connectivity methods are increasingly used to study directed interactions between brain regions from electrophysiological signals. These methods often show good results in simulated data but it is unclear to what extent connectivity results obtained from real data are physiologically plausible. Here we introduce a benchmark approach using multichannel somatosensory evoked potentials (SEPs) measured across rat cortex, where the structural and functional connectivity is relatively simple and well-understood. Rat SEPs to whisker stimulation are exclusively initiated by contralateral primary sensory cortex (S1), at known latencies, and with activity spread from S1 to specific cortical regions. This allows for a comparison of time-varying connectivity measures according to fixed criteria. We thus evaluated the performance of time-varying Partial Directed Coherence (PDC) and the Directed Transfer Function (DTF), comparing row- and column-wise normalization and the effect of weighting by the power spectral density (PSD). The benchmark approach revealed clear differences between methods in terms of physiological plausibility, effect size and temporal resolution. The results provide a validation of time-varying directed connectivity methods in an animal model and suggest a driving role for ipsilateral S1 in the later part of the SEP. The benchmark SEP dataset is made freely available.

KW - Barrel cortex

KW - Dynamics

KW - EEG

KW - Effective connectivity

KW - Functional networks

KW - Granger causality

UR - http://www.scopus.com/inward/record.url?scp=84901346892&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84901346892&partnerID=8YFLogxK

U2 - 10.1016/j.neuroimage.2014.04.016

DO - 10.1016/j.neuroimage.2014.04.016

M3 - Article

C2 - 24736179

AN - SCOPUS:84901346892

VL - 97

SP - 206

EP - 216

JO - NeuroImage

JF - NeuroImage

SN - 1053-8119

ER -