Elsevier

NeuroImage

Volume 80, 15 October 2013, Pages 234-245
NeuroImage

The Human Connectome Project and beyond: Initial applications of 300 mT/m gradients

https://doi.org/10.1016/j.neuroimage.2013.05.074Get rights and content

Highlights

  • Diffusion spectrum imaging to study traumatic coma recovery

  • In vivo human axon diameter measurements using 300 mT/m gradients

  • High-resolution (0.6 mm isotropic) diffusion imaging in whole, fixed human brain

Abstract

The engineering of a 3 T human MRI scanner equipped with 300 mT/m gradients – the strongest gradients ever built for an in vivo human MRI scanner – was a major component of the NIH Blueprint Human Connectome Project (HCP). This effort was motivated by the HCP's goal of mapping, as completely as possible, the macroscopic structural connections of the in vivo healthy, adult human brain using diffusion tractography. Yet, the 300 mT/m gradient system is well suited to many additional types of diffusion measurements. Here, we present three initial applications of the 300 mT/m gradients that fall outside the immediate scope of the HCP. These include: 1) diffusion tractography to study the anatomy of consciousness and the mechanisms of brain recovery following traumatic coma; 2) q-space measurements of axon diameter distributions in the in vivo human brain and 3) postmortem diffusion tractography as an adjunct to standard histopathological analysis. We show that the improved sensitivity and diffusion-resolution provided by the gradients are rapidly enabling human applications of techniques that were previously possible only for in vitro and animal models on small-bore scanners, thereby creating novel opportunities to map the microstructure of the human brain in health and disease.

Introduction

The NIH Blueprint for Neuroscience Research Human Connectome Project (HCP) is in the process of mapping the connections of the adult human brain as completely as possible using diffusion tractography, functional MRI (FMRI), magnetoencephalography (MEG), electroencephalography (EEG) and genetics. This comprehensive initiative involves two different consortia: the Washington University, the University of Minnesota and Oxford University consortium (WU–Minn HCP) and the Massachusetts General Hospital and the University of California, Los Angeles consortium (MGH–UCLA HCP). The “connectome” map will integrate complementary efforts from both HCP consortia and provide a new foundation for understanding neural network connectivity and structural neuroanatomy. In addition to its basic neuroscience merits, mapping of healthy adult human connectomes provides a basis for studying how connectomes change as humans learn, mature and age, and how a connectome becomes dysfunctional with injury or disease.

Beyond these high impact scientific contributions, the legacy of the HCP will also include its significant technological advancements. Prominent among these are MRI scanners equipped with significantly stronger magnetic gradients. The WU–Minn HCP involves a human 3 T MRI with 100 mT/m maximum gradients (Van Essen et al., 2012). The MGH–UCLA HCP involves the first ever human MRI scanner equipped with 300 mT/m gradients (7.5 × stronger than the clinical standard (40 mT/m)) and a 64-channel phased-array receiver coil (Keil et al., 2012). This MRI technology, which was engineered specifically for the HCP, has already been applied to several additional in vivo neuroimaging studies, which are impractical or underserved by existing technology. A detailed description of the novel hardware employed by the MGH-UCLA consortia and a characterization of the improvements in data quality provided by this hardware can found in Setsompop et al. (2013). Here we describe how the 300 mT/m gradients of the MGH–UCLA connectome scanner are creating new ways to probe the human brain and opening up new possibilities to serially investigate changes in brain tissue microstructure. This paper focuses on the first imaging studies conducted on the MGH–UCLA connectome scanner that fall outside the immediate scope of the HCP but are an impactful byproduct of its technological developments. Namely, we present initial results for: diffusion tractography of the pathways that mediate human consciousness and recovery after traumatic coma, in vivo mapping of axon diameter distributions and postmortem diffusion imaging.

There are two key advantages provided by an MRI scanner with 7.5 × stronger magnetic gradients, both of which have been instrumental in facilitating the neuroimaging applications discussed here. First, the stronger gradient amplitudes achieve a given diffusion-encoding gradient area in less time. This reduces the entire diffusion-encoding period and the echo time. As a result, the T2 decay is reduced and the signal-to-noise ratio (SNR) is increased. Second, the larger gradient amplitudes enable “stronger” diffusion encoding (larger diffusion-encoding gradient areas, larger q-values and b-values) to be achieved in practical times. The resulting enhancement in the “diffusion resolution” represents a smaller spatial scale at which differences in spin displacements can be resolved. Together, these advantages of stronger magnetic field gradients dramatically increase the sensitivity to tissues with short T2 and improve q-space diffusion imaging methods that demand large diffusion-gradient areas.

q-Space diffusion imaging (Callaghan, 1991, Cory and Garroway, 1990) samples multiple q-values, where a q-value is proportional to the diffusion-encoding gradient area. The Fourier Transform of the signal attenuation at multiple q-values describes the diffusion propagator, i.e. the probability density function (PDF) that a particle has been displaced a specific distance along a given orientation during a given diffusion time. The maximum q-value determines the diffusion resolution and hence the stronger connectome gradients probe sharper features of the diffusion PDF in less time and with less T2 loss, thereby pushing q-space imaging methods beyond prior capabilities. The stronger connectome gradients also allow a given q-value to be reached in a shorter diffusion time, probing the barriers experienced by the diffusing water molecules on a short time scale.

In this work we use a novel gradient system, 300 mT/m, slew-rate 200 T/m/s (Siemens AS302), integrated into a clinical 3 T MRI scanner (MAGNETOM Skyra CONNECTOM, Siemens Healthcare). We use these capabilities to explore a broader q and diffusion time (Δ) parameter space than was previously possible in humans in vivo. Here, we show that beyond mapping of the human connectome, there are a number of additional applications of the connectome's gradients that are rapidly advancing our ability to probe the structural basis of human brain function.

Traumatic coma affects more than 1 million people worldwide each year (Bruns and Hauser, 2003, Faul et al., 2010). In addition, hundreds of military personnel have experienced a severe traumatic brain injury (TBI) causing coma since the start of the wars in Afghanistan and Iraq (Bell et al., 2009, DuBose et al., 2011). For these civilians and military personnel with traumatic coma, recovery of consciousness requires the activation and integration of brain networks that support the two critical components of consciousness: arousal (wakefulness) and awareness (Kinney and Samuels, 1994). Advanced neuroimaging techniques have begun to identify the neural networks in the cerebral hemispheres that enable recovery of awareness (Fernandez-Espejo et al., 2012, Vanhaudenhuyse et al., 2010), but the pathways in the brainstem ascending reticular activating system (ARAS) that facilitate recovery of arousal after coma have yet to be elucidated. Since the discovery of the ARAS by Moruzzi and Magoun in 1949 (Moruzzi and Magoun, 1949), the vast majority of ARAS connectivity studies have been performed in animal models (Fuller et al., 2011, Gennarelli et al., 1982, Smith et al., 2000), with human studies mostly being limited to two-dimensional lesion mapping analyses (Parvizi and Damasio, 2003) and functional activation studies using positron emission tomography (Kinomura et al., 1996, Silva et al., 2010).

Diffusion tractography is a promising tool for elucidating the structural connectivity of ARAS pathways in the human brainstem. One of the principal challenges, however, is detecting the crossing fibers that are a prominent component of the ARAS network (Nauta and Kuypers, 1958). Diffusion Spectrum Imaging (DSI) (Callaghan et al., 1988, Wedeen et al., 2005) is one of several diffusion techniques that provides a means of detecting these crossing fibers. DSI requires the acquisition of multiple q-values and diffusion orientations arranged in a lattice in q-space (qx,qy,qz). A 3D Fourier transform of this q-space lattice yields the 3D spin displacement PDF. The orientation distribution function (ODF) along a given orientation is then the radial projection of the spin-displacement PDF. Maxima on the ODF are used to identify fiber orientations.

Here, we analyze ARAS connectivity with deterministic tractography on DSI ODFs and probabilistic tractography on the ball and two stick model using the connectome scanner's 300 mT/m gradients in a 27-year-old man who recovered from traumatic coma. Six years and seven months prior to being imaged on the MGH-UCLA connectome scanner, the patient had experienced a severe TBI caused by a car that hit him while he was riding his bike. He was in a coma for 14 days, and a clinical MRI scan performed on day 7 post-injury revealed multiple microhemorrhages in the cerebral hemispheres, splenium of the corpus callosum, and brainstem (bilaterally), consistent with severe grade 3 diffuse axonal injury (Adams et al., 1989). After emergence from coma, the patient underwent one and a half months of inpatient rehabilitation, followed by one year of outpatient rehabilitation. By the time of his MGH–UCLA connectome scan, the patient's Glasgow Outcome Scale—Extended score was 7 (out of a maximal score of 8), indicating “good recovery,” and his Disability Rating Scale score was 0, indicating that he was completely independent with activities of daily living and had returned to a normal work environment — one of few patients with bilateral traumatic axonal injury lesions in the brainstem who has been reported to experience this level of recovery (Skandsen et al., 2011).

A different genre of q-space imaging focuses on sampling temporal q-space rather than spatial q-space (i.e. acquiring multiple q-values at multiple diffusion times). If sufficiently high q-values and a sufficiently wide range of diffusion times are acquired, it is possible to extract information about the size of restricted compartments, such as axons, based on theoretical models of the diffusion diffraction patterns that they will generate.

The diameter of an axon is proportional to the speed at which action potentials are conducted along its length (Hoffmeister et al., 1991, Hursh, 1939). Therefore, any change in the distribution of axon diameters in a white matter tract can putatively impact the operation of neural networks (Ringo et al., 1994). Axon diameter has also been linked to several neurological disorders. For example, large diameter axons are thought to be selectively damaged in amyotrophic lateral sclerosis patients (Cluskey and Ramsden, 2001, Heads et al., 1991). Smaller diameter axons are thought to be maldeveloped in autistic children (Piven et al., 1997). Smaller diameter axons are also thought to be the most vulnerable to neurodegeneration associated with aging (Marner et al., 2003).

The majority of axon diameter mapping using MRI has been performed in vitro (Assaf et al., 2008, Ong and Wehrli, 2010, Stanisz et al., 1997) and in animal models (Barazany et al., 2009) due to the demands for strong encoding gradients which have, thus far, only been available on small bore MRI scanners. Despite this, at least one group has now demonstrated estimation of an axon diameter index, a′, in the in vivo human brain using a standard clinical gradient strength of 40–60 mT/m (Alexander et al., 2010, Zhang et al., 2011). The ActiveAx (a′ contrast) and the AxCaliber technique developed by Assaf et al. (2008) are very similar. The output for ActiveAx provides a single statistic of the axon diameter distribution (the weighted mean) whereas AxCaliber estimates the mean and variance of the gamma distribution estimated by AxCaliber. ActiveAx is also orientationally invariant whereas the canonical version of AxCaliber is not. Therefore, ActiveAx shows significant promise for rapid clinical translation of axon diameter estimates. The sensitivity of both methods is limited by the SNR and range of achievable q-values and diffusion times, and these factors are directly linked to the maximum available gradient strength. AxCaliber has also recently been demonstrated in vivo at 40 mT/m maximum gradient strengths (Horowitz et al., 2012). Although the data presented here do not represent the first attempt at axon diameter mapping in the in vivo human brain, they are the first AxCaliber in vivo human data acquired with a similar gradient strength (Gmax = 300 mT/m) to the animal studies used to validate the methodology (Barazany et al., 2009).

While AxCaliber has previously been applied in the in vivo human brain using 40 mT/m gradients (Horowitz et al., 2012), here we present the first ever in vivo human brain AxCaliber data using 300 mT/m gradients. We focus on the corpus callosum because of the availability of gold standard electron microscopy in this region and because of the critical role that inter-hemispheric trans-callosal connections play in many cognitive functions.

Diffusion imaging of whole, fixed human brains has gained traction in recent years as a method of obtaining a type of gold-standard, high-spatial resolution, high-SNR data without the motion artifacts or distortions that constrain in vivo acquisitions. Postmortem diffusion images also provide a useful link between in vivo imaging and histology. Postmortem scans allow us to see what additional microstructural information we might be able to visualize if/when we are able to improve the sensitivity and image fidelity of in vivo acquisitions. For example, diffusion anisotropy in the cerebral cortex was first identified in postmortem diffusion imaging (Leuze et al., 2012, McNab et al., 2009, Miller et al., 2011) and has now, with the improved resolution of in vivo diffusion imaging, been identified in vivo by several groups (Anwander et al., 2010, Heidemann et al., 2010, McNab et al., 2013).

While postmortem diffusion imaging of whole, fixed human brains benefits from long scan times and the absence of motion, it is hindered by decreased T2 relaxation times (due to the fixation process) (D'Arceuil et al., 2007, McNab et al., 2009) and reduced apparent diffusion coefficients (ADCs). Each of these limitations puts competing demands on the diffusion sensitization. Specifically, the diffusion sensitization must be both stronger and executed in less time. As such the gradients of the MGH–UCLA connectome scanner have the potential to dramatically improve the quality of diffusion imaging in whole, fixed, human brain specimens.

Section snippets

Methods

All data were acquired using the MGH–UCLA connectome scanner which consists of a novel AS302 gradient system that is part of a new 3 T system (MAGNETOM Skyra CONNECTOM, Siemens Healthcare) capable of up to 300 mT/m and a slew rate of 200 T/m/s. The slew rate was de-rated to 62.5 T/m/s during the diffusion encoding to prevent physiological stimulation. A custom-made 64-channel phased array coil was used for signal reception (Keil et al., 2012). All studies were performed with the approval of the

Diffusion spectrum imaging of traumatic coma recovery

In both the deterministic and the probabilistic analyses, streamlines were identified between the PPN and the CEM/Pf, CL, and Ret in all four control subjects, consistent with a recent deterministic tractography analysis of ARAS connectivity in the ex vivo and in vivo human brain (Edlow et al., 2012). For the four controls, PSCI values (mean +/− SD) for PPN-CEM/Pf, PPN-CL, and PPN-Ret, were 0.026 +/− 0.006, 0.032 +/− 0.02, and 0.066 +/− 0.01, respectively. For the patient, streamlines were also

Diffusion spectrum imaging of traumatic coma recovery

Over the past several decades, numerous studies in animal models (Gennarelli et al., 1982, Ommaya and Gennarelli, 1974, Smith et al., 2000) and human patients (Adams et al., 1989, Strich, 1961) have implicated lesions of the ARAS arousal network in the pathogenesis of traumatic coma. Yet, identification of disrupted pathways within the ARAS is not currently part of the standard clinical evaluation for patients with traumatic coma. Despite the direct relevance of ARAS connectivity to disorders

Acknowledgments

This work was funded by: a CIHR Fellowship and an NIH Blueprint for Neuroscience Research Grant: U01MH093765, as well as NIH funding from NCRR P41RR14075, NIBIB RO1EB006847 and NINDS R25NS065743 and the Center for Integration of Medicine and Innovative Technology (Boston, MA). We are grateful to the patient who volunteered for this study and whose recovery from traumatic coma continues to inspire this work. Thank you to Louis Vinke for assistance with processing postmortem brain specimens.

References (78)

  • K.L. Miller et al.

    Diffusion imaging of whole, post-mortem human brains on a clinical MRI scanner

    NeuroImage

    (2011)
  • H.H. Ong et al.

    Quantifying axon diameter and intra-cellular volume fraction in excised mouse spinal cord with q-space imaging

    NeuroImage

    (2010)
  • N.B. Reese et al.

    The pedunculopontine nucleus—auditory input, arousal and pathophysiology

    Prog. Neurobiol.

    (1995)
  • J.D. Schmahmann et al.

    Disconnection syndromes of basal ganglia, thalamus, and cerebrocerebellar systems

    Cortex

    (2008)
  • K. Setsompop et al.

    Pushing the limits of in vivo diffusion MRI for the Human Connectome Project

    Neuroimage

    (2013)
  • S.J. Strich

    Shearing of nerve fibers as a cause of brain damage due to head injury: a pathological study of twenty cases

    Lancet

    (1961)
  • J.D. Tournier et al.

    Direct estimation of the fiber orientation density function from diffusion-weighted MRI data using spherical deconvolution

    NeuroImage

    (2004)
  • V.J. Wedeen et al.

    Diffusion spectrum magnetic resonance imaging (DSI) tractography of crossing fibers

    NeuroImage

    (2008)
  • H. Zhang et al.

    Axon diameter mapping in the presence of orientation dispersion with diffusion MRI

    NeuroImage

    (2011)
  • J.H. Adams et al.

    Diffuse axonal injury in head injury: definition, diagnosis and grading

    Histopathology

    (1989)
  • A. Anwander et al.

    In vivo measurement of cortical anisotropy by diffusion-weighted imaging correlates with cortex type

    Proc. Int. Soc. Magn. Reson. Med.

    (2010)
  • Y. Assaf et al.

    AxCaliber: a method for measuring axon diameter distribution from diffusion MRI

    Magn. Reson. Med.

    (2008)
  • D. Barazany et al.

    In vivo measurement of axon diameter distribution in the corpus callosum of rat brain

    Brain

    (2009)
  • T.E.J. Behrens et al.

    Characterization and propagation of uncertainty in diffusion-weighted MR imaging

    Magn. Reson. Med.

    (2003)
  • R.S. Bell et al.

    Military traumatic brain and spinal column injury: a 5-year study of the impact blast and other military grade weaponry on the central nervous system

    J. Trauma

    (2009)
  • N. Bodammer et al.

    Eddy current correction in diffusion-weighted imaging using pairs of images acquired with opposite diffusion gradient polarity

    Magn. Reson. Med.

    (2004)
  • J. Bruns et al.

    The epidemiology of traumatic brain injury: a review

    Epilepsia

    (2003)
  • P.T. Callaghan

    Principles of Nuclear Magnetic Resonance Microscopy

    (1991)
  • P.T. Callaghan et al.

    NMR microscopy of dynamic displacements — k-space and q-space imaging

    J. Phys. E: Sci. Instrum.

    (1988)
  • K. Chung et al.

    Structural and molecular interrogation of intact biological systems

    Nature

    (2013)
  • S. Cluskey et al.

    Mechanisms of neurodegeneration in amyotrophic lateral sclerosis

    Mol. Pathol.

    (2001)
  • D.G. Cory et al.

    Measurement of translational displacement probabilities by NMR—an indicator of compartmentation

    Magn. Reson. Med.

    (1990)
  • J.J. DuBose et al.

    Isolated severe traumatic brain injuries sustained during combat operations: demographics, mortality outcomes, and lessons to be learned from contrasts to civilian counterparts

    J. Trauma

    (2011)
  • T.B. Dyrby et al.

    Contrast and stability of the axon diameter index from microstructure imaging with diffusion MRI

    Magn. Reson. Med.

    (2012)
  • B. Edlow et al.

    Neuroanatomic connectivity of the human ascending arousal system critical to consciousness and its disorders

    J. Neuropathol. Exp. Neurol.

    (2012)
  • B.L. Edlow et al.

    Disconnection of the ascending arousal system in traumatic coma

    J. Neuropathol. Exp. Neurol.

    (2013)
  • M. Faul et al.

    Traumatic Brain Injury in the United States: Emergency Department Visits, Hospitalizations, and Deaths

    (2010)
  • D. Fernandez-Espejo et al.

    A role for the default mode network in the bases of disorders of consciousness

    Ann. Neurol.

    (2012)
  • P. Fuller et al.

    Reassessment of the structural basis of the ascending arousal system

    J. Comp. Neurol.

    (2011)
  • Cited by (285)

    View all citing articles on Scopus
    1

    Jennifer A. McNab and Brian L. Edlow are co-first authors of this manuscript.

    View full text