Elsevier

NeuroImage

Volume 124, Part A, 1 January 2016, Pages 157-167
NeuroImage

Comparison of accelerated T1-weighted whole-brain structural-imaging protocols

https://doi.org/10.1016/j.neuroimage.2015.08.026Get rights and content

Highlights

  • Qualitative and quantitative comparison of 4-fold accelerated T1w protocols.

  • 4-fold accelerated T1w protocols have measurable impact on CNR and SNR.

  • CAIPIRINHA scans perceived to be on average of equal value for reading as ADNI-2.

  • 4-fold accelerated T1w protocols bias the segmentation results compared to ADNI-2.

Abstract

Imaging in neuroscience, clinical research and pharmaceutical trials often employs the 3D magnetisation-prepared rapid gradient-echo (MPRAGE) sequence to obtain structural T1-weighted images with high spatial resolution of the human brain. Typical research and clinical routine MPRAGE protocols with ~ 1 mm isotropic resolution require data acquisition time in the range of 5–10 min and often use only moderate two-fold acceleration factor for parallel imaging.

Recent advances in MRI hardware and acquisition methodology promise improved leverage of the MR signal and more benign artefact properties in particular when employing increased acceleration factors in clinical routine and research. In this study, we examined four variants of a four-fold-accelerated MPRAGE protocol (2D-GRAPPA, CAIPIRINHA, CAIPIRINHA elliptical, and segmented MPRAGE) and compared clinical readings, basic image quality metrics (SNR, CNR), and automated brain tissue segmentation for morphological assessments of brain structures. The results were benchmarked against a widely-used two-fold-accelerated 3T ADNI MPRAGE protocol that served as reference in this study.

22 healthy subjects (age = 20–44 yrs.) were imaged with all MPRAGE variants in a single session. An experienced reader rated all images of clinically useful image quality. CAIPIRINHA MPRAGE scans were perceived on average to be of identical value for reading as the reference ADNI-2 protocol. SNR and CNR measurements exhibited the theoretically expected performance at the four-fold acceleration.

The results of this study demonstrate that the four-fold accelerated protocols introduce systematic biases in the segmentation results of some brain structures compared to the reference ADNI-2 protocol. Furthermore, results suggest that the increased noise levels in the accelerated protocols play an important role in introducing these biases, at least under the present study conditions.

Introduction

The 3D magnetisation-prepared rapid acquisition gradient-echo (MPRAGE) pulse sequence (Mugler and Brookeman, 1990) with Alzheimer's Disease Neuroimaging Initiative protocol (ADNI-2) (Jack et al., 2010) parameters is a well-established standard for multi-site and longitudinal MRI studies that involve T1-weighted imaging of the human brain. Both in routine clinical and research settings, reduced MRI scan times are desirable for increased patient throughput, improved patient comfort, and better management of patient motion. The 3T imaging protocol of the ADNI employs conventional two-fold parallel imaging (Griswold et al., 2002) to reduce the acquisition time of a whole-brain MPRAGE scan from 9 min in the original ADNI-1 protocol (Jack et al., 2008) to 5 min in ADNI-2. Recently, several strategies have been proposed to further reduce the acquisition time in 3D structural-brain-imaging protocols beyond conventional parallel imaging, e.g. 2D-GRAPPA (Blaimer et al., 2006), CAIPIRINHA (Breuer et al., 2006, Brenner et al., 2014), and segmented MPRAGE (Falkovskiy et al., 2013).

Because of the excellent tissue contrast, the 3D T1-weighted MPRAGE images are appreciated for radiological reading. Moreover, quantitative assessment of brain tissues and the volume of individual brain structures has become an important tool in more research-oriented applications of the MPRAGE (Miller et al., 2009, Camicioli et al., 2009, Jack, 2011, Mills and Tamnes, 2014, Schmitter et al., 2015). For instance, studies aiming at quantification of disease progression, drug efficiency or normal aging apply serial imaging to assess structural changes over time. Since the expected effect sizes in normal aging and disease (e.g. increased atrophy rates) are often subtle over time, it is vital to understand the reliability and reproducibility of the imaging-based quantitative measurements and to understand any inconsistencies that may appear when changing the pulse sequence between longitudinal repeat scans.

The reproducibility of the reference protocol (5-minute ADNI-2) has been studied extensively (Jovicich et al., 2006, Jovicich et al., 2009, Jovicich et al., 2013, Wonderlick et al., 2009, Krueger et al., 2012, Kruggel et al., 2010, Morey et al., 2010, Leung et al., 2014, Ching et al., 2015, Reuter et al., 2012). However, to our knowledge, the influence of further accelerations obtained by applying 2D-GRAPPA, CAIPIRINHA or segmented MPRAGE protocols on volumetric brain measurements and clinical readings has not been reported so far.

In this work, we aim to assess the reliability of the data obtained on the same platform across these accelerated protocols both qualitatively and quantitatively. The qualitative analysis is carried out by an experienced observer. The quantitative analysis is performed through assessing the reproducibility of volume measurements with automated brain segmentation algorithm (Schmitter et al., 2015, Roche et al., 2011) contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR) measurements (Robson et al., 2008).

Section snippets

Theory

Conventional GRAPPA- (Griswold et al., 2002) or SENSE-type accelerations (Pruessmann et al., 1999) are often used in clinical settings. Both methods obtain acceleration by undersampling of the k-space data in one phase-encoding direction (1D). In most cases, only moderate accelerations are employed due to increased aliasing and noise amplification (g-factor penalty) with higher acceleration factors. In order to overcome those limitations partly, and assuming that there is sufficient SNR, it is

MR acquisition

All experiments were performed on a standard clinical 3T MRI (MAGNETOM Skyra, Siemens Healthcare, Erlangen, Germany) equipped with a 32-channel head coil array.

The measurement protocol consisted of five 3D MPRAGE volume acquisitions with protocol parameters similar to the ADNI-2 MPRAGE protocol settings (TR/TI = 2300/900 ms, α = 9 deg., BW = 240 Hz/pixel, readout in superior-inferior direction) (Jack et al., 2010) but with 1 mm isotropic resolution at a FoV of 256 × 240 × 176 mm3: (a) T1w ADNI-2 protocol with

Results

Each row in Fig. 2 demonstrates images from the same subject (three shown in total) obtained within one session using the reference ADNI-2 protocol and the 3-minute accelerated protocol variants. The windowing was set to the same level for all shown images. In the brain stem region, there is a visible increase of noise in accelerated protocols when compared to the reference ADNI-2 protocol.

Discussion

In longitudinal studies that use morphometric assessments of brain tissues and structures, the choice of the imaging protocol can potentially influence qualitative readings and degrade reproducibility of serial automated brain segmentations. The main objective of this study was to investigate the impact that accelerated protocols (2D-GRAPPA, CAIPIRINHA, segmented MPRAGE, and CAIPIRINHA with elliptical scanning) have on the observer's qualitative readings and on an automated brain segmentation

Conclusion

In summary, there are three main practical results of this study. First, accelerations of 3D structural brain scans beyond the routinely used acceleration factor of two have a measureable impact on some image analysis metrics (CNR, SNR, and noise). However, obtained images provide at least very similar information for qualitative readings in this 3T setting. Second, our analysis suggests that using or combining data from different variants of MPRAGE protocols should be done with caution. This

References (39)

  • K.L. Mills et al.

    Methods and considerations for longitudinal structural brain imaging analysis across development

    Dev. Cogn. Neurosci.

    (2014)
  • M. Reuter et al.

    Within-subject template estimation for unbiased longitudinal image analysis

    NeuroImage

    (2012)
  • A. Roche et al.

    On the convergence of EM-like algorithms for image segmentation using Markov random fields

    Med. Image Anal.

    (2011)
  • D. Schmitter et al.

    An evaluation of volume-based morphometry for prediction of mild cognitive impairment and Alzheimer's disease

    NeuroImage. Clin.

    (2015)
  • A.J.W. van der Kouwe et al.

    Brain morphometry with multiecho MPRAGE

    NeuroImage

    (2008)
  • J.S. Wonderlick et al.

    Reliability of MRI-derived cortical and subcortical morphometric measures: effects of pulse sequence, voxel geometry, and parallel imaging

    NeuroImage

    (2009)
  • M.A. Bernstein et al.

    Effect of windowing and zero-filled reconstruction of MRI data on spatial resolution and acquisition strategy

    J. Magn. Reson. Imaging

    (2001)
  • M. Blaimer et al.

    2D-GRAPPA-operator for faster 3D parallel MRI

    Magn. Reson. Med.

    (2006)
  • D. Brenner et al.

    Two-dimensional accelerated MP-RAGE imaging with flexible linear reordering

    MAGMA

    (2014)
  • Cited by (15)

    View all citing articles on Scopus
    View full text