High‐resolution functional MRI at 3 T: 3D/2D echo‐planar imaging with optimized physiological noise correction

Antoine Lutti,Nikolaus Weiskopf,Chloe Hutton,David L Thomas

doi:10.1002/mrm.24398

Abstract

High-resolution functional MRI (fMRI) offers unique possibilities for studying human functional neuroanatomy. Although high-resolution fMRI has proven its potential at 7 T, most fMRI studies are still performed at rather low spatial resolution at 3 T. We optimized and compared single-shot two-dimensional echo-planar imaging (EPI) and multishot three-dimensional EPI high-resolution fMRI protocols. We extended image-based physiological noise correction from two-dimensional EPI to multishot three-dimensional EPI. The functional sensitivity of both acquisition schemes was assessed in a visual fMRI experiment. The physiological noise correction increased the sensitivity significantly, can be easily applied, and requires simple recordings of pulse and respiration only. The combination of three-dimensional EPI with physiological noise correction provides exceptional sensitivity for 1.5 mm high-resolution fMRI at 3 T, increasing the temporal signal-to-noise ratio by more than 25% compared to two-dimensional EPI. Magn Reson Med, 2013. © 2012 Wiley Periodicals, Inc.

Highlights

High-resolution functional MRI has proven its importance to reveal the functional microstructure of the human brain in cortical and subcortical areas [1,2,3]
Maximal tSNR was achieved when the reference partition was set to the center partition, and this setting was used for all experimental results presented here
The largest increases in tSNR were obtained in cortical gray matter (GM), where the contribution of physiological noise to overall image variance is larger due to the larger image signal-to-noise ratio (SNR)

Summary

Introduction

High-resolution functional MRI (fMRI) has proven its importance to reveal the functional microstructure of the human brain in cortical and subcortical areas [1,2,3]. Ultrahigh fields (!7 T) facilitate high-resolution fMRI due to the increased sensitivity to the blood oxygen level-dependent (BOLD) effect and signal-to-noise ratio (SNR) [4,5,6]. Few comparisons of the methods optimal for high-resolution fMRI have yet been conducted at 3 T. The most complete approach based on an extensive model including regressors for cardiorespiratory and motion effects into the general linear model (GLM) used for statistical analyses has not been extended to 3D EPI yet. A recent study [8] applied only the RETROICOR subset of nuisance covariates to 3D data [13] but did not include regressors modeling motion [14] or respiratory volume effects [15]

Methods

Results

Conclusion