Elsevier

NeuroImage

Volume 53, Issue 4, December 2010, Pages 1233-1243
NeuroImage

Track-density imaging (TDI): Super-resolution white matter imaging using whole-brain track-density mapping

https://doi.org/10.1016/j.neuroimage.2010.07.024Get rights and content

Abstract

Neuroimaging advances have given rise to major progress in neurosciences and neurology, as ever more subtle and specific imaging methods reveal new aspects of the brain. One major limitation of current methods is the spatial scale of the information available. We present an approach to gain spatial resolution using post-processing methods based on diffusion MRI fiber-tracking, to reveal structures beyond the resolution of the acquired imaging voxel; we term such a method as super-resolution track-density imaging (TDI). A major unmet challenge in imaging is the identification of abnormalities in white matter as a cause of illness; super-resolution TDI is shown to produce high-quality white matter images, with high spatial resolution and outstanding anatomical contrast. A unique property of these maps is demonstrated: their spatial resolution and signal-to-noise ratio can be tailored depending on the chosen image resolution and total number of fiber-tracks generated. Super-resolution TDI should greatly enhance the study of white matter in disorders of the brain and mind.

Research Highlights

►A new super-resolution method based on fiber-tracking MRI is presented. ►It produces high-quality white matter images. ►It produces high spatial resolution and outstanding anatomical contrast. ►The spatial resolution and signal-to-noise ratio can be tailored. ►Novel contrast mechanism for tissues of mixed grey matter and white matter.

Introduction

The advent of neuroimaging has undoubtedly been one of the major contributors to advances in the neurosciences and clinical neurology. Each subsequent major advance in brain imaging, such as those in MRI of functional MRI (fMRI) for cognition (Kwong et al., 1992, Bandettini, 2009, Matthews et al., 2006) and diffusion-weighted imaging (DWI) for stroke (Moseley et al., 1990, Thomas et al., 2000), has been the precursor to new understanding of brain function and the basis of disease, as well as provided impetus for new clinical applications (Matthews et al., 2006, Gillard et al., 2005). Even though it can be argued that much of this knowledge was available from histopathological studies in post-mortem brains and from other techniques, advanced neuroimaging methods link in vivo the immediate clinical problem directly with the underlying brain structure and function in individual subjects. Many of the major advances in understanding the basis of brain abnormalities in disease have arisen from the ability to image the whole-brain and detect underlying pathology, with ever increasing sophistication. Importantly, each new neuroimaging technical advance has led to new insights, as ever more subtle and specific imaging and post-processing methods reveal new information about the brain that can be clinically applied.

One major limitation of current neuroimaging methods, compared to traditional histopathology, is the spatial scale of the information available. Many structures of biological interest are small compared to the scale of the imaging voxel. Increased signal-to-noise ratio (SNR) can allow improved spatial resolution to be achieved (i.e. smaller voxels), and is part of the impetus for ultra-high field MRI at 7 T and beyond (Duyn et al., 2007). Our proposed approach is to gain spatial resolution using post-processing methods, to reveal structures beyond the resolution of the acquired imaging voxel by using additional information obtained from outside that voxel. We propose such a super-resolution method (which we term as super-resolution track-density imaging, or super-resolution TDI) in the current study.

The aims of this study are to:

  • 1.

    Propose a method to generate higher resolution (i.e. super-resolution) anatomical images with high anatomical contrast, by incorporating extra information from diffusion tractography modeling.

  • 2.

    Demonstrate that these super-resolution images allow a direct visualization of important sub-structures of the functional nodes of the brain network (e.g. the thalamic nuclei and the cerebellar peduncles) not easily identified on high-resolution conventional MRI.

  • 3.

    Use the super-resolution images to demonstrate structural connections from the thalamus to the cortex, as well as the cerebellar sub-networks.

Such an advance in neuroimaging, which allows high-quality brain images to be generated at a resolution not feasible before with standard MRI technology, should have far reaching consequences on neuroscience.

Section snippets

Data acquisition

DWI data were acquired from 5 healthy volunteers on a 3 T Siemens Trio system, using a twice-refocused SE–EPI sequence (Reese et al., 2003) (b = 3000 s/mm2). Three subjects (subjects S1S3) were scanned with 150 DWI-directions, 54 contiguous slices, voxel size 2.3 × 2.3 × 2.3 mm3, and acquisition time: 20.5 min. A further subject (S4) was scanned with 60 DWI-directions, 60 contiguous slices, voxel size 2.5 × 2.5 × 2.5 mm3, and acquisition time: 9.4 min. The final subject (S5) was scanned with 20

Results

To illustrate the super-resolution properties of this new imaging method, Fig. 1 shows axial maps generated for subject S1 with increasing spatial resolution, i.e. with decreasing grid-size: isotropic grids of 2.3 mm (the original source resolution of the data for this subject), 1.0 mm, 250 μm, and 125 μm. For comparison, fractional anisotropy (FA) (Basser, 1995) and anatomical 3D T1-weighted images (a current state-of-the-art MRI conventional high-resolution image (MPRAGE), in this case at 1 mm

Discussion

This study has shown that the novel super-resolution TDI technique provides a method to produce high-quality white matter images, with high spatial resolution, and exquisite anatomical contrast not achievable with other MRI modalities. A unique property of these maps is their super-resolution nature: their spatial resolution and SNR can be tailored depending on the chosen grid-size and the total number of tracks generated. For example, Fig. 2 showed that high-quality 125 μm isotropic resolution

Acknowledgments

We are grateful to the National Health and Medical Research Council (NHMRC) of Australia and the Austin Health for the support. We thank Dr. Heath Pardoe (Brain Research Institute, Australia) for the helpful discussions, and Mr. Robert E. Smith (Brain Research Institute, Australia) for his help in implementing the directionally-color encoded TDI maps.

References (29)

  • T.E. Behrens et al.

    Non-invasive mapping of connections between human thalamus and cortex using diffusion imaging

    Nature Neurosci.

    (2003)
  • F. Calamante et al.

    Whole-brain track-density mapping as a tool for fiber tractography

  • J.H. Duyn et al.

    High-field MRI of brain cortical substructure based on signal phase

    Proc. Natl Acad. Sci. USA

    (2007)
  • K.V. Embleton et al.

    Anatomical connectivity mapping

  • Cited by (328)

    View all citing articles on Scopus
    1

    These authors contributed equally to this work.

    View full text