Multi-echo fMRI: A review of applications in fMRI denoising and analysis of BOLD signals

Prantik Kundu*, Valerie Voon, Priti Balchandani, Michael V Lombardo, Benedikt A. Poser, Peter A Bandettini

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

574 Downloads (Pure)

Abstract

In recent years the field of fMRI research has enjoyed expanded technical abilities related to resolution, as well as use across many fields of brain research. At the same time, the field has also dealt with uncertainty related to many known and unknown effects of artifact in fMRI data. In this review we discuss an emerging fMRI technology, called multi-echo (ME)-fMRI, which focuses on improving the fidelity and interpretability of fMRI. Where the essential problem of standard single-echo fMRI is the indeterminacy of sources of signals, whether BOLD or artifact, this is not the case for ME-fMRI. By acquiring multiple echo images per slice, the ME approach allows T2* decay to be modeled at every voxel at every time point. Since BOLD signals arise by changes in T2* over time, an fMRI experiment sampling the T2* signal decay can be analyzed to distinguish BOLD from artifact signal constituents. While the ME approach has a long history of use in theoretical and validation studies, modern MRI systems enable whole-brain multi-echo fMRI at high resolution. This review covers recent multi-echo fMRI acquisition methods, and the analysis steps for this data to make fMRI at once more principled, straightforward, and powerful. After a brief overview of history and theory, T2* modeling and applications will be discussed. These applications include T2* mapping and combining echoes from ME data to increase BOLD contrast and mitigate dropout artifacts. Next, the modeling of fMRI signal changes to detect signal origins in BOLD-related T2* versus artifact-related S0 changes will be reviewed. A focus is on the use of ME-fMRI data to extract and classify components from spatial ICA, called multi-echo ICA (ME-ICA). After describing how ME-fMRI and ME-ICA lead to a general model for analysis of fMRI signals, applications in animal and human imaging will be discussed. Applications include removing motion artifacts in resting state data at subject and group level. New imaging methods such as multi-band multi-echo fMRI and imaging at 7T are demonstrated throughout the review, and a practical analysis pipeline is described. The review culminates with evidence from recent studies of major boosts in statistical power from using multi-echo fMRI for detecting activation and connectivity in healthy individuals and patients with neuropsychiatric disease. In conclusion, the review shows evidence that the multi-echo approach expands the range of experiments that is practicable using fMRI. These findings suggest a compelling future role of the multi-echo approach in subject-level and clinical fMRI.

Original languageEnglish
Pages (from-to)59-80
Number of pages22
JournalNeuroimage
Volume154
Early online date29 Mar 2017
DOIs
Publication statusPublished - 1 Jul 2017

Keywords

  • INDEPENDENT COMPONENT ANALYSIS
  • RESTING STATE FMRI
  • FUNCTIONAL CONNECTIVITY
  • TIME-SERIES
  • SPIN-ECHO
  • DIFFERENTIATING BOLD
  • CONTRAST SENSITIVITY
  • BRAIN ACTIVATION
  • REMOVING MOTION
  • GLOBAL SIGNAL

Cite this