GLMdenoise: A fast, automated technique for denoising task-based fMRI data

Kendrick N. Kay, Ariel Rokem, Jonathan Winawer, Robert F. Dougherty, Brian A. Wandell

Research output: Contribution to journalArticle

Abstract

In task-based functional magnetic resonance imaging (fMRI), researchers seek to measure fMRI signals related to a given task or condition. In many circumstances, measuring this signal of interest is limited by noise. In this study, we present GLMdenoise, a technique that improves signal-to-noise ratio (SNR) by entering noise regressors into a general linear model (GLM) analysis of fMRI data. The noise regressors are derived by conducting an initial model fit to determine voxels unrelated to the experimental paradigm, performing principal components analysis (PCA) on the time-series of these voxels, and using cross-validation to select the optimal number of principal components to use as noise regressors. Due to the use of data resampling, GLMdenoise requires and is best suited for datasets involving multiple runs (where conditions repeat across runs). We show that GLMdenoise consistently improves cross-validation accuracy of GLM estimates on a variety of event-related experimental datasets and is accompanied by substantial gains in SNR. To promote practical application of methods, we provide MATLAB code implementing GLMdenoise. Furthermore, to help compare GLMdenoise to other denoising methods, we present the Denoise Benchmark (DNB), a public database and architecture for evaluating denoising methods. The DNB consists of the datasets described in this paper, a code framework that enables automatic evaluation of a denoising method, and implementations of several denoising methods, including GLMdenoise, the use of motion parameters as noise regressors, ICA-based denoising, and RETROICOR/RVHRCOR. Using the DNB, we find that GLMdenoise performs best out of all of the denoising methods we tested.

Original languageEnglish (US)
Article number247
JournalFrontiers in Neuroscience
Issue number7 DEC
DOIs
StatePublished - 2013

Fingerprint

Magnetic Resonance Imaging
Noise
Benchmarking
Signal-To-Noise Ratio
Linear Models
Principal Component Analysis
Research Personnel
Databases
Datasets

Keywords

  • BOLD fMRI
  • Correlated noise
  • Cross-validation
  • General linear model
  • ICA
  • Physiological noise
  • RETROICOR
  • Signal-to-noise ratio

ASJC Scopus subject areas

  • Neuroscience(all)

Cite this

Kay, K. N., Rokem, A., Winawer, J., Dougherty, R. F., & Wandell, B. A. (2013). GLMdenoise: A fast, automated technique for denoising task-based fMRI data. Frontiers in Neuroscience, (7 DEC), [247]. https://doi.org/10.3389/fnins.2013.00247

GLMdenoise : A fast, automated technique for denoising task-based fMRI data. / Kay, Kendrick N.; Rokem, Ariel; Winawer, Jonathan; Dougherty, Robert F.; Wandell, Brian A.

In: Frontiers in Neuroscience, No. 7 DEC, 247, 2013.

Research output: Contribution to journalArticle

Kay, Kendrick N. ; Rokem, Ariel ; Winawer, Jonathan ; Dougherty, Robert F. ; Wandell, Brian A. / GLMdenoise : A fast, automated technique for denoising task-based fMRI data. In: Frontiers in Neuroscience. 2013 ; No. 7 DEC.
@article{b3ccb56d1b8e45dcbf7ee384ecbd304d,
title = "GLMdenoise: A fast, automated technique for denoising task-based fMRI data",
abstract = "In task-based functional magnetic resonance imaging (fMRI), researchers seek to measure fMRI signals related to a given task or condition. In many circumstances, measuring this signal of interest is limited by noise. In this study, we present GLMdenoise, a technique that improves signal-to-noise ratio (SNR) by entering noise regressors into a general linear model (GLM) analysis of fMRI data. The noise regressors are derived by conducting an initial model fit to determine voxels unrelated to the experimental paradigm, performing principal components analysis (PCA) on the time-series of these voxels, and using cross-validation to select the optimal number of principal components to use as noise regressors. Due to the use of data resampling, GLMdenoise requires and is best suited for datasets involving multiple runs (where conditions repeat across runs). We show that GLMdenoise consistently improves cross-validation accuracy of GLM estimates on a variety of event-related experimental datasets and is accompanied by substantial gains in SNR. To promote practical application of methods, we provide MATLAB code implementing GLMdenoise. Furthermore, to help compare GLMdenoise to other denoising methods, we present the Denoise Benchmark (DNB), a public database and architecture for evaluating denoising methods. The DNB consists of the datasets described in this paper, a code framework that enables automatic evaluation of a denoising method, and implementations of several denoising methods, including GLMdenoise, the use of motion parameters as noise regressors, ICA-based denoising, and RETROICOR/RVHRCOR. Using the DNB, we find that GLMdenoise performs best out of all of the denoising methods we tested.",
keywords = "BOLD fMRI, Correlated noise, Cross-validation, General linear model, ICA, Physiological noise, RETROICOR, Signal-to-noise ratio",
author = "Kay, {Kendrick N.} and Ariel Rokem and Jonathan Winawer and Dougherty, {Robert F.} and Wandell, {Brian A.}",
year = "2013",
doi = "10.3389/fnins.2013.00247",
language = "English (US)",
journal = "Frontiers in Neuroscience",
issn = "1662-4548",
publisher = "Frontiers Research Foundation",
number = "7 DEC",

}

TY - JOUR

T1 - GLMdenoise

T2 - A fast, automated technique for denoising task-based fMRI data

AU - Kay, Kendrick N.

AU - Rokem, Ariel

AU - Winawer, Jonathan

AU - Dougherty, Robert F.

AU - Wandell, Brian A.

PY - 2013

Y1 - 2013

N2 - In task-based functional magnetic resonance imaging (fMRI), researchers seek to measure fMRI signals related to a given task or condition. In many circumstances, measuring this signal of interest is limited by noise. In this study, we present GLMdenoise, a technique that improves signal-to-noise ratio (SNR) by entering noise regressors into a general linear model (GLM) analysis of fMRI data. The noise regressors are derived by conducting an initial model fit to determine voxels unrelated to the experimental paradigm, performing principal components analysis (PCA) on the time-series of these voxels, and using cross-validation to select the optimal number of principal components to use as noise regressors. Due to the use of data resampling, GLMdenoise requires and is best suited for datasets involving multiple runs (where conditions repeat across runs). We show that GLMdenoise consistently improves cross-validation accuracy of GLM estimates on a variety of event-related experimental datasets and is accompanied by substantial gains in SNR. To promote practical application of methods, we provide MATLAB code implementing GLMdenoise. Furthermore, to help compare GLMdenoise to other denoising methods, we present the Denoise Benchmark (DNB), a public database and architecture for evaluating denoising methods. The DNB consists of the datasets described in this paper, a code framework that enables automatic evaluation of a denoising method, and implementations of several denoising methods, including GLMdenoise, the use of motion parameters as noise regressors, ICA-based denoising, and RETROICOR/RVHRCOR. Using the DNB, we find that GLMdenoise performs best out of all of the denoising methods we tested.

AB - In task-based functional magnetic resonance imaging (fMRI), researchers seek to measure fMRI signals related to a given task or condition. In many circumstances, measuring this signal of interest is limited by noise. In this study, we present GLMdenoise, a technique that improves signal-to-noise ratio (SNR) by entering noise regressors into a general linear model (GLM) analysis of fMRI data. The noise regressors are derived by conducting an initial model fit to determine voxels unrelated to the experimental paradigm, performing principal components analysis (PCA) on the time-series of these voxels, and using cross-validation to select the optimal number of principal components to use as noise regressors. Due to the use of data resampling, GLMdenoise requires and is best suited for datasets involving multiple runs (where conditions repeat across runs). We show that GLMdenoise consistently improves cross-validation accuracy of GLM estimates on a variety of event-related experimental datasets and is accompanied by substantial gains in SNR. To promote practical application of methods, we provide MATLAB code implementing GLMdenoise. Furthermore, to help compare GLMdenoise to other denoising methods, we present the Denoise Benchmark (DNB), a public database and architecture for evaluating denoising methods. The DNB consists of the datasets described in this paper, a code framework that enables automatic evaluation of a denoising method, and implementations of several denoising methods, including GLMdenoise, the use of motion parameters as noise regressors, ICA-based denoising, and RETROICOR/RVHRCOR. Using the DNB, we find that GLMdenoise performs best out of all of the denoising methods we tested.

KW - BOLD fMRI

KW - Correlated noise

KW - Cross-validation

KW - General linear model

KW - ICA

KW - Physiological noise

KW - RETROICOR

KW - Signal-to-noise ratio

UR - http://www.scopus.com/inward/record.url?scp=84898026889&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84898026889&partnerID=8YFLogxK

U2 - 10.3389/fnins.2013.00247

DO - 10.3389/fnins.2013.00247

M3 - Article

AN - SCOPUS:84898026889

JO - Frontiers in Neuroscience

JF - Frontiers in Neuroscience

SN - 1662-4548

IS - 7 DEC

M1 - 247

ER -