Elsevier

NeuroImage

Volume 43, Issue 2, 1 November 2008, Pages 288-296
NeuroImage

Spinal cord functional MRI at 3 T: Gradient echo echo-planar imaging versus turbo spin echo

https://doi.org/10.1016/j.neuroimage.2008.07.024Get rights and content

Abstract

The purpose of this study was to evaluate and compare turbo spin echo (TSE) with gradient echo echo-planar imaging (GE-EPI) pulse sequences for functional magnetic resonance imaging (fMRI) of spinal cord activation at 3 T field strength. Healthy volunteers underwent TSE and GE-EPI spinal fMRI. The activation paradigm comprised the temporal alternation of finger motion and rest. Pulse sequences were optimized to obtain sufficient image quality and optimal sensitivity to small T2 or T2⁎ relaxation time changes. Spinal cord activation measured by the two pulse sequences was evaluated with respect to spatial distribution of activation, signal sensitivity, and reproducibility. For the GE-EPI sequence, fMRI activation was maximal in the spinal cord segments at the levels of the fifth cervical down to the first thoracic vertebra. For the TSE sequence, fMRI measurements showed no distinct location with maximal activation. Percentage signal change and number of activated voxels were approximately twice as high for GE-EPI compared to TSE fMRI. Reproducibility of the signal changes was much better for GE-EPI than for TSE imaging. To conclude, multi-subjects averaged GE-EPI is more location specific for blood-oxygen-level-dependent (BOLD) activation, more sensitive, and is suggested to be more reproducible than TSE fMRI.

Introduction

Extensive clinical and basic research in the field of functional magnetic resonance imaging (fMRI) of the human brain is performed in a large number of research institutes. Functional MRI of the brain is based on the blood oxygenation level dependent (BOLD) effect, measuring magnetic susceptibility changes of blood due to hemodynamically related deoxygenated hemoglobin concentration changes during neuronal activation. In the spinal cord, a similar BOLD effect is expected that should enable the measurement of spinal cord activation by fMRI. Besides being relevant to clinical neuroscience, in the future fMRI might be clinically relevant for assessing the viability of injured cord tissue in terms of vascular responses and potentially provides an alternative or adjunct to electrophysiological assessments.

Detection of the BOLD effect in the spinal cord poses some technical challenges that complicate functional imaging compared to similar experiments in the brain (Giove et al., 2004, Stroman, 2005). These complications include (i) cord motion related to cardiac cycle, pulsatile cerebrospinal fluid (CSF) flow, respiration, and swallowing, (ii) magnetic field inhomogeneities due to magnetic susceptibility differences between cord and surrounding tissues, and (iii) relative small cross-sectional and large longitudinal cord dimensions. The feasibility of spinal cord fMRI has now independently been demonstrated by a number of research groups and several possible solutions have been applied to overcome these issues. For instance, breath holding (Stroman et al., 1999), respiratory gating (Stroman and Ryner, 2001), and cardiac-gating (Backes et al., 2001) have been used to limit the cord motion. Flow-compensating gradients (Stroman, 2006, Stroman et al., 2001, Stroman et al., 2002a, Stroman et al., 1999, Stroman et al., 2002b, Yoshizawa et al., 1996) and presaturation slabs (Backes et al., 2001, Stroman et al., 2001) (Stroman, 2006, Stroman et al., 2002a, Stroman et al., 2002b) have been used to further reduce motion artifacts. Spatially selective saturation pulses were applied on each side of the cord to minimize the effect of inflow enhancement (Stroman et al., 1999). Volume shimming was used to increase field homogeneity (Maieron et al., 2007, Stracke et al., 2005, Stroman et al., 1999). Small voxels were used to avoid mixing of cord tissue and CSF signal in the same voxel, which inherently occurs due to the small axial cord dimensions.

Pulse sequences previously applied for spinal fMRI include T2-weighted turbo spin echo (TSE) (Stroman, 2006, Stroman et al., 2005, Stroman et al., 2001, Stroman et al., 2002a, Stroman et al., 2002b), T2⁎-weighted echo-planar imaging (EPI) (Backes et al., 2001, Govers et al., 2007, Madi et al., 2001, Maieron et al., 2007, Stracke et al., 2005), and T2⁎-weighted fast low angle shot (FLASH) (Stroman et al., 1999, Yoshizawa et al., 1996) methods. The BOLD signal consists of an extravascular and an intravascular component from both small and large vessels. GE-EPI (T2⁎-weighting) should provide a stronger signal response than spin echo imaging (T2-weighting) due to sensitivity to magnetic susceptibility changes associated with the BOLD effect. The ratio of signal changes in the brain of gradient echo versus spin echo imaging has been estimated to be almost 2:1 (Bandettini et al., 1994) at 1.5 T. However, spin echo imaging is expected to be more spatially accurate than gradient echo imaging as the signal becomes more weighted towards microvasculature, which is assumed to be closer to the location of neuronal activation, and less extravascular magnetic field changes associated with large (possibly distant) veins are measured with spin echo imaging (Bandettini et al., 1994, Duong et al., 2003, Poser and Norris, 2007). Furthermore, spin echo imaging provides better image quality as it is less sensitive to magnetic field inhomogeneities (e.g. through plane susceptibility gradients and in-plane distortions), which increase for EPI at higher field strengths (Poser and Norris, 2007).

As it is still unclear which technique is best for spinal cord fMRI, we compared TSE and GE-EPI and evaluated image quality and signal sensitivity, spatial distribution, and reproducibility of fMRI signal responses at a field strength of 3 T.

Section snippets

Subjects

Healthy volunteers underwent GE-EPI (n = 10) and TSE (n = 8) fMRI examinations of the spinal cord. Volunteers were seven men and three women with an age between 25 and 40 years. Three volunteers were studied twice (10 weeks interval) using both pulse sequences to assess the reproducibility of the techniques.

Exercise paradigm

Volunteers were instructed and trained to perform unilateral finger motion by flexing and stretching all five fingers at a frequency of approximately 1 Hz. The volunteers received vocal

Image quality

In Table 2 SNR and CNR values are listed for the optimized TSE and GE-EPI pulse sequences. Image quality of TSE was superior to GE-EPI in terms of both SNR and CNR. Fig. 4 provides examples of acquired GE and SE images. No obvious distortions or motion artifacts after motion correction were noticed. However, strong signal modulations and signal voids were noticed (Fig. 4, Fig. 6) for GE-EPI in the back of the spinal canal that accompany the inhomogeneities between bones and vertebral discs.

Activation

In

Discussion

In this study, signal changes in the cervical and upper thoracic spinal cord in response to a motor exercise task were demonstrated for both echo time optimized TSE and GE-EPI pulse sequences. Images of TSE and GE-EPI were evaluated and compared in terms of image quality and signal sensitivity, location specificity, and reproducibility of cord activation.

Conclusion

Spinal cord GE-EPI appeared to have better signal sensitivity, better location specificity, and was suggested to be more reproducible than TSE fMRI, despite the superior image quality of TSE images.

References (26)

  • BlandJ.M. et al.

    Statistical methods for assessing agreement between two methods of clinical measurement

    Lancet

    (1986)
  • BlandJ.M. et al.

    Measurement error

    Bmj

    (1996)
  • BrooksJ.C. et al.

    Physiological noise modelling for spinal functional magnetic resonance imaging studies

    Neuroimage

    (2008)
  • Cited by (0)

    View full text