Elsevier

NeuroImage

Volume 44, Issue 3, 1 February 2009, Pages 827-838
NeuroImage

Sensitivity of voxel-based morphometry analysis to choice of imaging protocol at 3 T

https://doi.org/10.1016/j.neuroimage.2008.09.053Get rights and content

Abstract

The objective of this study was to determine which 3D T1-weighted acquisition protocol at 3 T is best suited to voxel-based morphometry (VBM), and to characterize the sensitivity of VBM to choice of acquisition. First, image quality of three commonly used protocols, FLASH, MP-RAGE and MDEFT, was evaluated in terms of SNR, CNR, image uniformity and point spread function. These image metrics were estimated from simulations, phantom imaging and human studies. We then performed a VBM study on nine subjects scanned twice using the three protocols to evaluate differences in grey matter (GM) density and scan–rescan variability between the protocols. These results reveal the relative bias and precision of the tissue classification obtained using the different protocols. MDEFT achieved the highest CNR between white and grey matter, and the lowest GM density variability of the three sequences. Each protocol is also characterized by a distinct regional bias in GM density due to the effect of transmission field inhomogeneity on image uniformity combined with spatially variant GM T1 values and the sequence's T1 contrast function. The required population sample size estimates to detect a difference in GM density in longitudinal VBM studies, i.e. based only on methodological variance, were lowest for MDEFT. Although MP-RAGE requires more subjects than FLASH, its higher cortical CNR improves the accuracy of the tissue classification results, particularly in the motor cortex. For cross-sectional VBM studies, the variance in morphology across the population is likely to be the primary source of variability in the power analysis.

Introduction

When investigating subtle structural brain differences between subject populations, or within the same population over time, using morphometric analyses, large populations are often required due to the biological variability in brain morphology between subjects. In such studies, automated quantitative analysis is essential as it allows for objective statistical analysis of the results. Using measures such as atrophy, cortical thickness and grey matter density, subtle morphometric group differences that are not directly discernible upon visual inspection can be detected.

There are three main branches of computational neuroanatomy that detect regional structural differences over the whole brain. The first two branches, deformation-based morphometry (Ashburner et al., 1998) and tensor-based morphometry (Chiang et al., 2007, Leow et al., 2006), are based on the analysis of the deformation field that non-linearly registers two data sets. Deformation-based morphometry examines the relative position of brain structures by analyzing displacement vectors. Tensor-based morphometry studies the local shape and volume of structures; spatial patterns of tissue expansion and contraction are determined from the analysis of the Jacobian determinants of the deformation field. Higher order moments of the determinant can be used to characterize strain or torque for example. Voxel-based morphometry (VBM) (Ashburner and Friston, 2000, Wright et al., 1995), on the other hand, detects local differences in tissue composition across subjects once gross anatomical differences have been accounted for by linear or non-linear registration to a model. Originally applied to study grey matter (GM) and white matter (WM) changes in schizophrenia patients (Wright et al., 1999, Wright et al., 1995), this brain mapping technique has provided insight into the structural changes in many neurologic and psychiatric disorders, including Alzheimer's disease (Baron et al., 2001, Whitwell et al., 2007), Parkinson's disease and dementia with Lewy bodies (Beyer et al., 2007, Burton et al., 2004), epilepsy (Bernasconi et al., 2004, Bonilha et al., 2007), autism (Abell et al., 1999), as well as normal brain development and aging (Sowell et al., 1999).

In VBM, images are pre-processed to reduce acquisition artifacts and improve sensitivity to biological differences in local tissue composition. The stability of the morphometric analysis process, over time and, if applicable, across sites, is crucial in such studies to enhance the power of statistical results and reduce the number of subjects required. The accuracy of the morphometric results is equally important, as there is questionable value in using a consistent measure of morphometric difference that is not related to true anatomical variation.

The accuracy and precision of VBM relies on the quality of the input images and the image processing algorithms used. An important feature of VBM analysis is that it is not biased to a particular region or structure of the brain. Likewise, the anatomical images should not be biased by regional variations in signal- and contrast-to-noise, or by imaging artifacts. These regional biases, whether caused by transmission inhomogeneity or natural biological variation in the MR properties of brain tissues, could lead to regional variations in the accuracy and precision of the VBM results.

There is a growing interest in high field anatomical imaging due to increased sensitivity. The gain in baseline magnetization allows for more flexibility in the acquisition process. SNR can be traded off to vary contrast characteristics and/or increase spatial resolution, for instance. High field imaging also brings new technical challenges including an increase in transmission field inhomogeneity and modified relaxation properties. These challenges, in addition to non-standardized hardware and image acquisition protocols, originally limited the use of high field imaging in VBM studies. Now that these challenges have been largely addressed in commercial 3 Tesla (T) scanners, VBM (Chan et al., 2006, Chung et al., 2004, Meyer et al., 2007, Rocca et al., 2006) and other morphometric studies of data sets acquired at 3 T are slowing emerging, yet there have been very few studies on the impact of MRI acquisition protocols on subsequent morphometric measures at 3 T (Dickerson et al., 2008, Han et al., 2006). This question is important both for selecting acquisition protocols for new studies and assessing the extent to which existing studies can be compared to each other.

The objective of this paper is to evaluate three 3D T1-weighted acquisition protocols at 3 T from the perspective of VBM. We chose to evaluate the two most widespread acquisition sequences: FLASH (Frahm et al., 1986, Haase, 1990) and MP-RAGE (Bluml et al., 1996, Deichmann et al., 2000, Epstein et al., 1994, Mugler and Brookeman, 1990, Mugler and Brookeman, 1991, Runge et al., 1991). We also included our implementation of a third sequence that has been proposed for high field strengths: MDEFT (modified driven-equilibrium Fourier transform) (Deichmann et al., 2004, Lee et al., 1995, Thomas et al., 2005, Ugurbil et al., 1993). We first evaluated these sequences in terms of basic image quality metrics, including SNR, CNR, image uniformity and k-space weighting. The evaluation includes MR pulse sequence simulations, as well as phantom and human imaging studies. Next, we investigated the sensitivity of VBM results to these acquisition protocols to determine which sequence is best suited to this type of whole brain quantitative analysis technique. The data set acquired consists of nine healthy volunteers scanned twice using the three protocols. The data were analyzed using published image processing tools and a probabilistic neuro-anatomical atlas, developed at the Montreal Neurological Institute. Finally, to place these results into the context of a VBM study, we performed a power analysis for each acquisition protocol to determine the required population sample size to detect a predefined difference in GM density.

Section snippets

Theory: T1-weighted sequences

The FLASH pulse sequence consists of a gradient echo with a short repetition time (TR) and reduced flip angle, and both radio-frequency (RF) and gradient spoiling to remove the residual transverse magnetization before the next excitation. After several repetitions, the magnetization reaches a steady state. The sequence contrast is determined by the choice of TR and flip angle, and the echo time (TE) is usually kept to a minimum to maximize SNR and minimize T2⁎-weighting. Due to its simple

Phantom design

A uniform elliptical cylinder phantom with similar dimensions, electrical properties and relaxation times as the human brain was designed to study B1 transmission field and the resulting image non-uniformity (Sled and Pike, 1998). The phantom solution consists of distilled H20, 77 mM NaCl to adjust the conductivity to σ  0.6 S/m, and 103 μM MnCl2 to modify the T1 relaxation time to 879 ± 27 ms, estimated using DESPOT1 (Deoni et al., 2003) with B1 field correction (Sled and Pike, 1998). The

Image quality

Image uniformity was quantified by the standard deviation σ and signal range of the elliptical phantom images, which have been normalized by the mean signal intensity. The results in Table 1 show that FLASH signal intensity is the most insensitive to variations in the B1 field over the range expected in the human head. FLASH has the lowest standard deviation σ = 0.072, and the smallest range of signal intensity of 32%. The MDEFT implementation with an adiabatic inversion pulse showed the most

Image quality

FLASH has some important advantages over magnetization prepared sequences; first and foremost, its simple sequence design and availability on all commercial MRI systems. These features are particularly attractive in multi-center studies that use different MR imaging systems as the implementation is easiest to match and the contrast settings are straightforward and do not vary with number of partitions. The implementation of MP-RAGE may vary between manufacturers, and MDEFT is currently not

Conclusion

This study was carried out to investigate the sensitivity of VBM to acquisition protocols at 3 T. Our results are only strictly applicable to the FLASH, MP-RAGE and MDEFT protocols chosen in this study. However, since the implementation and acquisition parameters of these sequences are relatively stable across imaging studies, we believe that these results reflect the performance of the sequences in general.

From the image quality analysis and the VBM results presented above, we can first

Acknowledgment

This research was funded by the Fonds de Recherche sur la Nature et les Technologies (C.L. Tardif) and the Canadian Institutes of Health Research (G.B. Pike).

References (57)

  • DickersonB.C. et al.

    Detection of cortical thickness correlates of cognitive performance: reliability across MRI scan sessions, scanners, and field strengths

    NeuroImage

    (2008)
  • ErnstT. et al.

    Absolute quantitation of water and metabolites in the human brain. I. Compartments and water

    J. Magn. Reson. Ser. B

    (1993)
  • HanX. et al.

    Reliability of MRI-derived measurements of human cerebral cortical thickness: the effects of field strength, scanner upgrade and manufacturer

    NeuroImage

    (2006)
  • LeowA.D. et al.

    Longitudinal stability of MRI for mapping brain change using tensor-based morphometry

    NeuroImage

    (2006)
  • MuellerS.G. et al.

    The Alzheimer's disease neuroimaging initiative

    Neuroimaging Clin. N. Am.

    (2005)
  • NorrisD.G. et al.

    MDEFT imaging of the human brain at 8 T

    MAGMA

    (1999)
  • SowellE.R. et al.

    Localizing age-related changes in brain structure between childhood and adolescence using statistical parametric mapping

    NeuroImage

    (1999)
  • WrightI.C. et al.

    A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia

    NeuroImage

    (1995)
  • WrightI.C. et al.

    Mapping of grey matter changes in schizophrenia

    Schizophr. Res.

    (1999)
  • AbellF. et al.

    The neuroanatomy of autism: a voxel-based whole brain analysis of structural scans

    NeuroReport

    (1999)
  • AshburnerJ. et al.

    Identifying global anatomical differences: deformation-based morphometry

    Hum. Brain Mapp.

    (1998)
  • BeyerM.K. et al.

    Gray matter atrophy in Parkinson disease with dementia and dementia with Lewy bodies

    Neurology

    (2007)
  • BonilhaL. et al.

    Extrahippocampal gray matter atrophy and memory impairment in patients with medial temporal lobe epilepsy

    Hum. Brain Mapp.

    (2007)
  • BurtonE.J. et al.

    Cerebral atrophy in Parkinson's disease with and without dementia: a comparison with Alzheimer's disease, dementia with Lewy bodies and controls

    Brain

    (2004)
  • ChanC.H. et al.

    Thalamic atrophy in childhood absence epilepsy

    Epilepsia

    (2006)
  • CohenJ.A.

    Statistical Power Analysis for the Behavioral Sciences

    (1988)
  • CollinsD.L. et al.

    Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space

    J. Comput. Assist. Tomogr.

    (1994)
  • CollinsD.L. et al.

    Automatic 3D model-based neuroanatomical segmentation

    Hum. Brain Mapp.

    (1995)
  • Cited by (60)

    • Improving structural brain images acquired with the 3D FLASH sequence

      2017, Magnetic Resonance Imaging
      Citation Excerpt :

      Recently, computer simulation has been used to optimize imaging parameters [12,13]. However, there is no general agreement on the optimal 3D FLASH parameters for brain image acquisition [12–14] because different objective functions (such as maximal SNR or CNR) have been used to determine the optimal imaging parameters. In this paper, we use the gray matter (GM) and white matter (WM) CNR efficiency as an objective function to optimize imaging parameters and minimize contrast inhomogeneity.

    View all citing articles on Scopus
    View full text