Show Summary Details
Page of

Diffusion Tensor Microimaging and Its Applications 

Diffusion Tensor Microimaging and Its Applications
Chapter:
Diffusion Tensor Microimaging and Its Applications
Author(s):

Jiangyang Zhang

, Hao Huang

, Manisha Aggarwal

, and Susumu Mori

DOI:
10.1093/med/9780195369779.003.0025
Page of

PRINTED FROM OXFORD MEDICINE ONLINE (www.oxfordmedicine.com). © Oxford University Press, 2015. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy).

Subscriber: null; date: 20 January 2018

Summary Box

  • This chapter presents the principles and challenges of diffusion tensor microimaging (DTMI).

  • Several practical techniques for achieving satisfactory DTMI results are described.

  • Diffusion tensor imaging provides superior contrast for structural delineation in the brain.

  • Various biomedical applications of DTMI are also discussed, including study of the development and maturation in the rat brain and brain connectivity in the macaque brain.

Microimaging, or microscopy, often refers to imaging with spatial resolution finer than 0.1 mm, which is the smallest separation that the naked human eye can resolve. Many imaging modalities can achieve a resolution finer than 0.1 mm. Figure 25.1 shows the range of imaging resolution of several common imaging modalities. Electron micrography provides the highest resolution and is capable of imaging cellular organelles. With sub-micron resolution, optical imaging is well suited for imaging tissue at the cellular level. Magnetic resonance imaging (MRI) and computed tomography (CT) cover the other, rather coarse, end of the spectrum, mainly targeting the tissue at organ levels. It is clear from Figure 25.1 that MRI cannot compete with optical imaging in terms of imaging resolution or in terms of sensitivity and specificity. The numerous immunohistochemical stainings available with optical imaging have been fine-tuned over decades to detect cellular and molecular events in biological tissue with high sensitivity and specificity (for a description of some of these staining methods, with which MRI competes, please refer to Chapter 3by Axer). The fine resolution, sensitivity, and specificity provided by optical imaging make it the preferred imaging modality in biomedical research.

Figure 25.1 The resolution ranges of electron microscopy (EM), optical microscopy, MRI, and CT ranges from virus to the entire organ. Each imaging modality is suited for imaging structures in its range of resolution. Overlaps between different imaging modalities exist and indicate that multiple imaging modalities are applicable.

Figure 25.1
The resolution ranges of electron microscopy (EM), optical microscopy, MRI, and CT ranges from virus to the entire organ. Each imaging modality is suited for imaging structures in its range of resolution. Overlaps between different imaging modalities exist and indicate that multiple imaging modalities are applicable.

It does, however, have its own limitations. First, it is often difficult to perform optical imaging-based studies on live animals longitudinally. Second, the requisite preparation procedures, including sectioning and staining, are time consuming and make it difficult to acquire distortion free three-dimensional (3D) information for morphological studies. MRI therefore provides complementary information that is difficult to obtain with optical imaging, as it is noninvasive, can be used to monitor live animals longitudinally, and can generate 3D digitized image data with sub-millimeter resolution. These are the main rationales for the development and use of MR microimaging.

One advantage of MRI over other imaging modalities providing a similar range of resolution, such as CT, is that it can provide versatile imaging contrasts for probing the inner structures of biological tissues. Imaging contrast is a critical issue because the diagnostic value of an image depends on both its spatial resolution and the tissue contrasts that it provides. Conventional MR contrasts include proton density, T1, T2, magnetization transfer, and diffusion. These contrasts reflect the physical and chemical microenvironments of tissue water molecules, for example, tissue water and myelin content. Based on diffusion MRI, diffusion tensor imaging (DTI) estimates the extent of water molecule diffusion, the degree of diffusion anisotropy, and the primary orientation of water diffusion (Basser et al., 1994; Le Bihan, 2003; Mori and Zhang, 2006). The orientation and anisotropy information can be encoded to produce unique tissue contrasts, which reflect tissue microstructures, for example, the organization of axons in the white matter. DTI contrasts are relatively independent of tissue water content and are modulated by tissue myelin content (Beaulieu, 2002). The rich information provided by DTI makes it a useful tool in many studies. As described in other chapters, DTI has been used to study brain anatomy and connectivity (Pierpaoli et al., 1996; Conturo et al., 1999; Mori et al., 1999, 2002; Xue et al., 1999; Basser et al., 2000; Catani et al., 2002; Mori and Van Zijl, 2002), white matter injury or degeneration in several neurological diseases, such as stroke and multiple sclerosis (Werring et al., 2000; Pierpaoli et al., 2001; Song et al., 2003; Sun et al., 2006b; Trip et al., 2006; Budde et al., 2007; Poonawalla et al., 2008), and the growth and maturation of white matter structures in embryonic and neonatal brains (Hüppi et al., 1998; Neil et al., 1998; Hüppi and Inder 2001; Mori et al., 2001).

Figure 25.2 shows several high-resolution MR images of a neonatal mouse cerebellum. The T2 and diffusion tensor images were acquired with a resolution of 0.05 mm × 0.05 mm × 0.05 mm. The contrast in the T2-weighted image is poor because neonatal mouse cerebellum lacks myelin. Diffusion tensor images, in comparison, show interesting contrasts, which will be explained later in this chapter. Although DTI can also be used to study skeletal and cardiac muscles (see Chapter 41 by Strijkers, Drost, and Niccolay) most DTI-based studies have been on neuronal tissues, e.g., the brain and spinal cord. This chapter will therefore focus only on the applications of diffusion tensor microimaging on neuronal tissue.

Figure 25.2 MR images of a postnatal day 8 (P8) mouse cerebellum acquired at 0.05 mm × 0.05 mm × 0.05 mm resolution. The 3D reconstruction of the cerebellum is shown in the upper left. The T2-weighted image does not show good white matter gray matter contrast because myelination starts at P8 in the mouse cerebellum. DTI-derived fractional anisotropy (FA) and direction-encoded color map (COLOR, V1*FA) images show good white matter contrast.

Figure 25.2
MR images of a postnatal day 8 (P8) mouse cerebellum acquired at 0.05 mm × 0.05 mm × 0.05 mm resolution. The 3D reconstruction of the cerebellum is shown in the upper left. The T2-weighted image does not show good white matter gray matter contrast because myelination starts at P8 in the mouse cerebellum. DTI-derived fractional anisotropy (FA) and direction-encoded color map (COLOR, V1*FA) images show good white matter contrast.

Soon after the introduction of DTI, studies that used diffusion tensor microimaging (DTMI) to examine tissue microstructures emerged, and DTMI has become a well-accepted technique to examine rodent brain anatomy. Ahrens et al. (1998) used ex vivo MR microscopy, including DTMI, to study white matter injuries in a mouse model of multiple sclerosis. They demonstrated that DTMI can visualize white matter organization in the mouse spinal cord, and diffusion anisotropy is sensitive to white matter lesion and thus can be used to evaluate white matter integrity. Hsu and Setton (1999) used ex vivo DTMI to examine the microstructure of excised porcine intervertebral disc. They showed that DTMI can visualize the layered organization of the annulus fibrosus and measure the orientation of the collagen fibers. Xue et al. (1999) performed in vivo DTMI of the rat brain and the first noninvasive 3D reconstruction of major white matter pathways in the rodent brain. From 2006 to 2007, PubMed listed more than 60 articles that used DTMI to study rodent brain and spinal cord.

In this chapter, we briefly present several technical aspects of DTMI and its applications in biomedical research. We have divided this chapter into three sections. In the first section, we present several technical challenges often encountered when implementing diffusion tensor microimaging. In the second section, we discuss the tissue contrasts in the acquired images and how to use them to examine complex neuroanatomy. In the third section, we present several interesting applications of DTMI.

DTMI: The Technical Aspects

The Challenges Faced by DTMI

DTMI is basically a fusion of DTI and MR microimaging, and therefore it faces the challenges that are associated with both techniques. The primary technical challenge in DTMI is to achieve satisfactory signal-to-noise ratio (SNR) at high spatial resolution. SNR places the theoretical limitation on the spatial resolution that we can achieve with MR microimaging. The diffusion of water molecules also limits the spatial resolution because of the considerable signal loss resulting from diffusion of spatially encoded spins out of their original voxel before acquisition at high spatial resolution. Callaghan (1991) argued that MR microimaging may achieve a spatial resolution of around 0.002 mm per pixel. For diffusion MR and DTI, the highest spatial resolution achievable may be around 0.01 mm per pixel, given the diffusion of water molecules during the long echo time required for the diffusion-sensitizing gradients. The volume of a mouse brain is approximately 3000 times smaller than that of a human brain, and the typical voxel dimension in current state-of-the-art human brain DTI acquisitions is around 2 mm per pixel. In order to achieve the same relative resolution as that of clinical DTI in the rodent brains, we need to achieve a voxel dimension of 0.1–0.2 mm, an order of magnitude smaller than in human brain DTI. This translates into a 3 orders of magnitude loss in MR signal strength (as the signal is proportional to the volume of the voxel). In addition, DTI is inherently a noisy technique, because the signal magnitude in diffusion-weighted images is attenuated by diffusion-sensitizing gradients, posing further challenges on the front of SNR.

In order to achieve satisfactory SNR in DTMI, several approaches can be taken, namely, use of more sensitive imaging probes, stronger magnetic field, and more signal averages. Increasing the magnetic field strength and the sensitivity of the imaging probes are the most direct ways to improve SNR and achieve high spatial resolution. Specimens in DTMI are often quite small (dimensions 〈20 mm), so they can fit into smaller, more sensitive, custom-designed radiofrequency (RF) coils than coils used in clinical scanners. The strength of magnetic field also has a direct impact on the SNR. SNR is in general proportional to the strength of magnetic field, and high field magnets (e.g., 9.4 Tesla, 11.7 Tesla, or even 17 Tesla) are necessary for DTMI. The disadvantage of a strong magnetic field is that it lengthens tissue T1 while shortening tissue T2. The former is detrimental because it means that it takes longer for the longitudinal magnetization of spins to fully recover after each excitation, and this will lead to either a reduction in the SNR or an increase in the scan time. Shorter T2 is also detrimental because the transverse magnetization decays faster, and this also limits the use of long diffusion time in diffusion-weighted spin echo–based experiments. High-field systems also tend to have more severe field inhomogeneity problems that can cause severe artifacts. For example, the presence of air bubbles or blood clots in ex vivo specimens often causes severe artifacts (due to differences in magnetic susceptibility) and reduces the diagnostic value of the images. The shorter T2 and severe field inhomogeneity also make echo-planar imaging (EPI) types of acquisition difficult on high-field systems, which puts high-field DTI acquisitions at a severe speed disadvantage compared with DT-MRI at lower magnetic fields (e.g., 1.5T or 3T). Signal averaging is less attractive compared to the above approaches because the SNR is proportional only to the square root of the number of averages. More signal averages will result in prolonged imaging time, which is often not an option even for ex vivo specimens because of problems with the stability of the sample or the instrument, for example, drifting of the magnetic field over time, and imaging cost.

In addition to the SNR challenge, the diagnostic value of DTMI is often degraded by artifacts caused by subject motion and gradient eddy currents. Motion artifacts only occur in in vivo experiments and can be minimized by careful animal constraints and respiratory triggering. Eddy current artifacts can be seen in both in vivo and ex vivo experiments and are caused by rapid switching of the diffusion sensitizing gradients. The severity of eddy current artifacts depends on many factors, including the geometry of the coil, shielding of the gradient coil, and the magnitude and direction of the diffusion gradients. Proper setting of the gradient pre-emphasis can significantly reduce the amount of eddy currents, but often cannot completely remove them. Eddy currents are especially troubling for fast spin echo–based DTMI experiments because the eddy current will result in phase differences between even and odd echoes in the echo train, which leads to phase discontinuity in the k-space and ghosting in the reconstructed images (Mori and van Zijl, 1998).

Even with all these challenges, there have been many advances in DTMI on high-field systems in recent years. Table 25.1 lists several DTMI applications and basic imaging parameters, for example, the field strength and imaging resolution achieved. As in clinical DTI, DTMI has been mainly used to examine white matter injury or abnormalities in the brain and spinal cord. The best resolution achieved for in vivo DTMI is approximately 0.1 mm × 0.1 mm × 0.5 mm (Budde et al., 2007), and the best resolution achieved for ex vivo DTMI is 0.02 mm × 0.02 mm × 0.3 mm (Ahrens et al., 1998).

Table 25.1 Selected DTMI Studies on the Central Nervous System in Mouse and Rat

Study and Applications

Resolution and Imaging Parameters

Budde et al. (2007) Axon and myelin injury in the mouse spinal cords in experimental autoimmune encephalomyelitis

0.078 mm × 0.078 mm × 1 mm, in vivo, 4.7 Tesla spectrometer, spin echo, Δ‎ = 25 ms, δ‎ = 10 ms, b = 785 s/mm2. Total imaging time = 2 hours

Sun et al. (2006b) Axon and myelin degeneration in the mouse brains

0.117 mm × 0.117 mm × 0.5 mm, in vivo, 4.7 Tesla spectrometer, spin echo, Δ‎ = 25 ms, δ‎ = 10 ms, b = 768 s/mm2. Total imaging time = 3 hours

Sizonenko et al. (2007) Early cortical injury in neonatal rat after hypoxic ischemic injury

0.125 mm × 0.125 mm × 0.5 mm, in vivo, 4.7 Tesla spectrometer, spin echo, Δ‎ = 25 ms, δ‎ = 10 ms, b = 768 s/mm2. Total imaging time = 4 hours

Ahrens et al. (1998) Axon and myelin pathology in mouse spinal cords in spontaneously acquired experimental allergic encephalomyelitis

0.02 mm × 0.02 mm × 0.3 mm, ex vivo, 11.7 Tesla spectrometer, spin echo, Δ‎ = 7.4 ms, δ‎ = 2 ms, b = 2000 s/mm2

Tyszka et al. (2006) White matter abnormalities in myelin-deficit shiverer mouse brains

0.08 mm × 0.08 mm × 0.08 mm, ex vivo, 11.7 Tesla spectrometer, spin echo, Δ‎ = 5 ms, δ‎ = 3 ms, b = 1450 s/mm2

Mori et al. (2001); Zhang et al. (2003) Cortical and white matter development in embryonic mouse brains

0.08 mm × 0.08 mm × 0.08 mm, ex vivo, 9.4 Tesla spectrometer, spin echo, Δ‎ = 12 ms, δ‎ = 5 ms, b = 1200 s/mm2. Total imaging time = 24 hours

Choices of Pulse Sequences for DTMI

The pulse sequences used in DTMI experiments consist of diffusion preparation, which uses diffusion-sensitizing gradients to tag diffusing spins, and signal acquisition, which samples the k-space data. For the preparation part, both spin echo and stimulated echo diffusion-weighted preparation can be used. Most DTMI experiments use spin echo preparation because stimulated echo preparation reduces the SNR by 50% if ignoring the effect of T2 decay. However, if a long diffusion time (e.g., 80 ms) is necessary, stimulated echo preparation should be used because the short T2 on high-field systems will severely degrade the signals after spin echo preparation with long diffusion time. To reduce eddy current–related imaging artifacts, double spin-echo bipolar diffusion gradients (also known as the “twice refocused” [Reese et al., 2003]) can be used. The scheme is shown in Figure 25.3. Two refocusing pulses and two pairs of diffusion gradients with opposite polarities are used in this scheme. The assumption here is that the eddy current generated by the first diffusion gradient and the eddy current generated by the second diffusion gradient with opposite polarity will cancel each other. However, the additional refocusing pulse in this scheme reduces the SNR in the case of an imperfect refocusing pulse and increases the complexity of the coherence pathway selection.

Figure 25.3 Spin echo (a), stimulated echo (b), and double refocusing bipolar gradient (c) diffusion preparation. In the bipolar gradient preparation, two refocusing pulses follow the initial excitation pulse. Pairs of diffusion gradients (G) with opposite polarity are positioned around each refocusing pulse to reduce diffusion gradient–induced eddy current. gro, read-out gradient; RF, radiofrequency; TE, echo time.

Figure 25.3
Spin echo (a), stimulated echo (b), and double refocusing bipolar gradient (c) diffusion preparation. In the bipolar gradient preparation, two refocusing pulses follow the initial excitation pulse. Pairs of diffusion gradients (G) with opposite polarity are positioned around each refocusing pulse to reduce diffusion gradient–induced eddy current. gro, read-out gradient; RF, radiofrequency; TE, echo time.

Various combinations of diffusion gradient duration (δ‎), diffusion time (Δ‎ − δ‎/3), and b-values have been used in DTMI experiments. Under the narrow pulse approximation, δ‎ should be much shorter than the diffusion time (δ‎ « Δ‎ so that diffusion of spins during δ‎ can be ignored. Practically, δ‎ is often limited by the performance of gradient hardware, i.e., the maximum gradient strength, gradient linearity, and eddy current concerns. In most DTMI studies, δ‎ ranged from 3 to 25 ms (Table 25.1). Diffusion time is a key parameter. Longer diffusion time allows water molecules to fully explore their microenvironment. The dependence of measured diffusivity on diffusion time has been shown in muscle fibers (Kim et al., 2005). In white matter structures, because the diameter of each axon is much smaller than that of muscle fibers, such dependence disappears when the diffusion time is longer than 5 ms (Beaulieu, 2002). For example, the mean axon diameter in rodent corpus callosum is approximately 0.11 ± 0.2 µm (Olivares et al., 2001), whereas the root mean-squared distance traveled by a water molecule during 5 ms in an unbounded environment is approximately 5–10 µm. This means that at 5 ms diffusion time, assuming slow exchange between different compartments, water molecules can fully explore their microenvironment. Clark et al. (2001) have shown that even in the human brain, where mean axon diameter is greater than in rodent brains (Bush and Allman, 2003), there is no apparent diffusion time dependence for diffusion time ranges from 8 to 80 ms. Most DTMI studies have used a diffusion time from 10 to 20 ms.

The b-value is another key parameter, and the optimal b-value for accurate tensor estimation depends on the apparent diffusion coefficient of the tissue (Kingsley and Monahan, 2004). Higher b-values produce more diffusion-weighted contrast but also reduce the SNR. A general rule of thumb is that the product of the b-value and the apparent diffusion coefficient of the target tissue should range from 1 to 2. Most in vivo studies have used b-values from 700 s/mm2 to 1000 s/mm2 for imaging mature brain and spinal cord. For ex vivo studies, because the diffusivity in postmortem samples is lower than in vivo (Sun et al., 2006a; Kim et al., 2007), probably due to changes in tissue microstructure since neither formalin fixation nor change in specimen temperature can fully explain the reduction in diffusivity, it is often necessary to increase the b-value to 1500–2000 s/mm2. For imaging immature brain and spinal cord, because the diffusivity is higher than in mature brain and spinal cord, lower b-values should be used (Kingsley and Monahan, 2004).

For the acquisition part, most current DTMI experiments use spin echo acquisition instead of EPI acquisition to avoid the artifacts associated with EPI-based acquisition on high-field systems. A diagram of a diffusion-weighted spin echo sequence is shown in Figure 25.4. To achieve better resolution and SNR, users can choose from two spin echo–based approaches: multiple spin echo (MSE) and fast spin echo (FSE). In both MSE and FSE acquisitions, there are multiple refocusing pulses and echoes after the initial excitation pulse. In the FSE acquisition, the multiple spin echoes sample different lines in the k-space, whereas in MSE acquisition, the multiple spin echoes sample the same line in the k-space. In both cases, the number of echoes that can be acquired is limited by the T2 decay. From the MSE experiments, multiple images will be obtained and added together to increase the SNR.

Figure 25.4 Diagrams of multiple spin echo (MSE) and fast spin echo (FSE). In the diagrams, phase-encoding gradients are not shown. Abbreviations are Dx, Dy, and Dz, diffusion-sensitizing gradients along the x-, y-, and z-axes; gro, the read-out gradient; RF, radiofrequency; Gx, Gy, and Gz, x-, y-, and z-gradient axes; gx, gy, and gz, crusher gradients around the refocusing pulses along the x-, y-, and z-axes. Inside the rectangular box of broken lines is the standard diffusion-weighted spin echo sequence. In the MSE sequence, additional echoes are acquired after the first echo, with the same phase-encoding step as in the first echo. In the FSE sequence, additional echoes are acquired at different phase-encoding steps. The part inside the orange box illustrates the twin-navigator echoes scheme. The two navigator echoes are acquired at the end of the echo train and always sample the center of the k-space. Diffusion-weighted mouse brain images with and without navigator correction are shown at the bottom. White arrows in the image without navigator correction indicate the artifacts caused by phase incoherence between echoes. These artifacts are removed using the twin-navigator correction scheme. Scale bars = 1 mm.

Figure 25.4
Diagrams of multiple spin echo (MSE) and fast spin echo (FSE). In the diagrams, phase-encoding gradients are not shown. Abbreviations are Dx, Dy, and Dz, diffusion-sensitizing gradients along the x-, y-, and z-axes; gro, the read-out gradient; RF, radiofrequency; Gx, Gy, and Gz, x-, y-, and z-gradient axes; gx, gy, and gz, crusher gradients around the refocusing pulses along the x-, y-, and z-axes. Inside the rectangular box of broken lines is the standard diffusion-weighted spin echo sequence. In the MSE sequence, additional echoes are acquired after the first echo, with the same phase-encoding step as in the first echo. In the FSE sequence, additional echoes are acquired at different phase-encoding steps. The part inside the orange box illustrates the twin-navigator echoes scheme. The two navigator echoes are acquired at the end of the echo train and always sample the center of the k-space. Diffusion-weighted mouse brain images with and without navigator correction are shown at the bottom. White arrows in the image without navigator correction indicate the artifacts caused by phase incoherence between echoes. These artifacts are removed using the twin-navigator correction scheme. Scale bars = 1 mm.

The choice of FSE or MSE acquisition depends on imaging time, resolution, and SNR requirements. FSE is more time efficient and is well suited for in vivo experiments. MSE requires the same amount of time as the conventional spin echo sequence but produces better SNR due to more signal averaging. Because both FSE and MSE acquisitions employ multiple refocusing pulses, unwanted coherence pathways can arise from imperfection in the refocusing pulses. The unwanted coherence pathways are not encoded properly by diffusion and phase-encoding gradients and can cause artifacts in the reconstructed images. It is necessary to combine phase cycling with crusher gradients around the refocusing pulses to remove these unwanted coherence pathways. For FSE acquisition, phase differences between even and odd echoes can cause severe artifacts in the reconstructed images (Fig. 25.4). However, if the crusher gradients are selected properly to remove all unwanted coherence pathways, the phase differences between odd and even echoes are constant. This enables the use of twin-navigator echo correction scheme (Fig. 25.4) to remove the phase differences (Mori and van Zijl, 1998). In this scheme, two additional navigator echoes are positioned at the end of each echo train. One navigator echo records the phase of the odd echoes and the other one the phase of the even echoes. During image reconstruction, the phase difference information recorded by the navigator echoes can be used to remove phase incoherence in the k-space and related image artifacts (Fig. 25.4).

DTMI: The Image Contrast

DTMI Contrasts for White Matter Structures

For white matter, DTMI provides similar tissue contrasts to those in clinical DTI. In DTMI data, white matter tracts can be distinguished from gray matter by their high diffusion anisotropy and by coherent orientation that is tangential to the trajectories of white matter tracts. Table 25.2 shows the definitions of several frequently used anisotropy indices. These anisotropy indices can be combined with eigenvectors to form color images that visualize both anisotropy information (as intensity) and orientation (as color) information (Pajevic and Pierpaoli, 1999). In addition, we have designed an index called the secondary eigenvector index (SI) as

SI=min{λ1λ2Dav,λ2λ3Dav}
where min is the function that returns the smallest input value, and Dav is the average of the three eigenvalues. SI can be used in combination with the secondary eigenvector (V2) to visualize the orientation of the secondary axon bundles, if they present and cross the primary bundles orthogonally, in a voxel (Zhang et al., 2006). It is important to note that these indices are all sensitive to sorting bias, as described by Basser and Pajevic (2000).

Table 25.2 Definition of several frequently used indices of diffusion anisotropy, their definition using eigenvalues, and references

Anisotropy indices

Definition

Reference

Fractional anisotropy (FA)

FA=(λ1λ2)2+(λ1λ3)2+(λ2λ3)22(λ12+λ22+λ32)

Basser and Pajevic, 2000

Relative anisotropy (RA)

RA=(λ1λ2)2+(λ1λ3)2+(λ2λ3)2(λ1+λ2+λ3)2

Alexander et al., 2000

Linear, planar, and spherical anisotropy indices (CL, CP, CS)

CL=λ1λ2λ1+λ2+λ3,CP=2(λ2λ3)λ1+λ2+λ3,CS=3λ3λ1+λ2+λ3,

Westin et al., 2002

The orientation and anisotropy contrasts can be used to reconstruct the 3D trajectory of a white matter tract and to study the connectivity of animal brains. Figure 25.5 shows examples of reconstructed white matter tracts in an adult mouse brain. In this figure, white matter tracts that are close to each other but have different orientations and can be resolved in the DTMI color-map images. For example, the cerebral peduncle (cp) and optic tract (opt) are both myelinated fiber tracts and located next to each other. These two tracts can be distinguished in the color-map image by difference in their orientation (color). The diffusivity and anisotropy contrasts have been used to study the integrity of white matter structures in various pathological conditions. Recently, several reports have demonstrated in several animal models that parallel diffusivity is sensitive to axonal injury and perpendicular diffusivity is sensitive to myelin injury (Song et al., 2003; Sun et al., 2006b; Kozlowski et al., 2008).

Figure 25.5 Diffusion tensor microscopy of an adult mouse brain. a) Nissl-stained histology (left) and diffusion tensor images of adult mouse brains. b) Reconstructed white matter tracts from the DTI results. Abbreviations are 2n, optic nerve; ac, anterior commissure; cc, corpus callosum; cp, cerebral peduncle; DG, dentate gyrus; ec, external capsule; f, fornix; fi, fimbria; H, hippocampus; ml, medial lemniscus; opt, optic tract; py, pyramidal tract; sm, stria medularis. The scale bar represents 1 mm. The color arrows illustrate our color scheme. Red represents rostral–caudal; green, medial–lateral; and blue, dorsal–ventral.

Figure 25.5
Diffusion tensor microscopy of an adult mouse brain. a) Nissl-stained histology (left) and diffusion tensor images of adult mouse brains. b) Reconstructed white matter tracts from the DTI results. Abbreviations are 2n, optic nerve; ac, anterior commissure; cc, corpus callosum; cp, cerebral peduncle; DG, dentate gyrus; ec, external capsule; f, fornix; fi, fimbria; H, hippocampus; ml, medial lemniscus; opt, optic tract; py, pyramidal tract; sm, stria medularis. The scale bar represents 1 mm. The color arrows illustrate our color scheme. Red represents rostral–caudal; green, medial–lateral; and blue, dorsal–ventral.

DTMI contrast for gray matter structures

DTMI can reveal interesting contrasts in several gray matter structures. In clinical DTI, it is often difficult to examine gray matter structures because of considerable partial volume effect. Recent DTMI studies have shown that there exist reproducible contrast patterns in the gray matter, and changes in these contrast patterns may reflect changes in gray matter microstructures (Zhang et al., 2002, 2003; Sizonenko et al., 2007). However, the complex arrangement of axons, neurons, and dendrites in the gray matter creates difficulties in understanding the basis of DTMI contrasts in the gray matter, and we should interpret DTMI data carefully. In this section, we will present contrast patterns in mouse cerebellum, and embryonic cortex.

DTMI contrast in the rodent cerebellum

The adult mouse cerebellum contains a large network of axons. Histological studies have shown that the cerebellar cortex contains millions of densely packed unmyelinated parallel fibers, which run parallel to the cerebellar surface, and Purkinje fibers, which fan out in a plane perpendicular to the parallel fibers. The number of Purkinje fibers is much less than the number of parallel fibers. Figure 25.6 shows mid-sagittal DTMI data from an adult mouse cerebellum. In the mid-sagittal color-map image (Fig. 25.6b, V1*FA), the cerebellar cortex has significantly higher diffusion anisotropy (FA values of 0.346 ± 0.036) than that of the mouse sensory cortex (FA values 0.140 ± 0.023). The V1s in the cerebellar cortex point to the medial–lateral orientation (red), along the orientation of parallel fibers as previously reported in the human cerebellum (Pierpaoli et al., 2002). These patterns are visualized using an enlarged vector plot (Fig. 25.6b).

Figure 25.6 DTMI data of the adult mouse cerebellum. a) Mid-sagittal diffusion-weighted image of a mouse brain. The green box indicates the location of cerebellar region shown in b) and c). b) Mid-sagittal DTMI color-map image of the mouse cerebellum. In the color map, red represents medial–lateral axis, green represents anterior–posterior axis, and blue represents superior–infereior axis. Schematic diagrams of possible axon arrangement in the orange box (cerebellar cortex) and blue box (cerebellar white matter) are shown on the right. In the diagram, red tubes represent parallel fibers, green tubes represent Purkinje fibers, and blue tubes represent axons running underneath the cerebellar cortex. c) Maps of fractional anisotropy (FA), secondary eigenvector index (SI), planar index (CP), and their corresponding color-map images.

Figure 25.6
DTMI data of the adult mouse cerebellum. a) Mid-sagittal diffusion-weighted image of a mouse brain. The green box indicates the location of cerebellar region shown in b) and c). b) Mid-sagittal DTMI color-map image of the mouse cerebellum. In the color map, red represents medial–lateral axis, green represents anterior–posterior axis, and blue represents superior–infereior axis. Schematic diagrams of possible axon arrangement in the orange box (cerebellar cortex) and blue box (cerebellar white matter) are shown on the right. In the diagram, red tubes represent parallel fibers, green tubes represent Purkinje fibers, and blue tubes represent axons running underneath the cerebellar cortex. c) Maps of fractional anisotropy (FA), secondary eigenvector index (SI), planar index (CP), and their corresponding color-map images.

In the V2*SI images (Fig. 25.6c), V2s in the cerebellar cortical regions are along the rostral–caudal orientation (green), the same orientation as that of the Purkinje fibers. The values of SI in the cerebellar cortex (0.266 ± 0.021) are significantly higher than the SI values in the cerebral cortex (0.106 ± 0.029 for the motor cortex and 0.086 ± 0.022 for the sensory cortex) and the optic nerve (0.154 ± 0.018), suggesting that a significant amount of underlying axon population consists of Purkinje fibers. The cerebellar white matter runs in the superior–inferior orientation in this area and can be identified as blue fibers in the V1*FA images. The V2*SI image clearly indicates a rostral–caudal fiber orientation (green), probably due to contributions of fibers entering to and from the cortex. Note that the cerebellar cortical region and the cerebellar white matter have different orientations in their V1 (red and blue for medial–lateral and dorsal–ventral orientations, respectively) but share the same orientation in V2 (green for rostral–caudal orientation). These results using a system with known anatomical structures confirmed a hypothesis that V2*SI can reveal underlying anatomical information that can’t be appreciated by V1-based contrasts.

DTMI contrast in the rodent cerebral cortex

The cerebral cortex of the mouse has a layered, columnar structure. In the mammalian embryonic brain, neurons are born in the ventricular zone near the lateral ventricles. They migrate along the radial glial cells, which form the scaffold of the embryonic cortex, from the ventricular zone to their destination. Later neurons end up in more superficial locations in the cortex. As the brain matures, radial glia gradually degenerate, but the columnar structure is preserved.

Diffusion anisotropy in the developing cortex was first described by Thornton et al. (1997) in the neonatal piglet brain and later observed in several mammalian species (Zhang et al., 2003; Sizonenko et al., 2007; Huang et al., 2009; Kroenke et al., 2009). In embryonic and neonatal cortex, it is hypothesized that the presence of radial glia is the main cause of diffusion anisotropy and the radial appearance of tissue orientation. As the brain matures, the radial glia gradually disappear (Bayer and Altman, 1991), and as the structure in the cortex becomes more complex, the degree of anisotropy in the cortex gradually decreases. However, even in adult mouse brains, it is still possible to observe this radial organization.

Figure 25.7 shows DTMI color-map images of the embryonic mouse cortex. Embryonic mouse cortical development has been extensively studied and well documented in the past several decades using histochemical methods. As shown in previous reports (Mori et al., 2001; Zhang et al., 2003), the V1*FA images carry a wealth of anatomical information on the embryonic cortical development. In the DTMI results (Fig. 25.7), the neuroepithelium (orange arrows), which surrounds the ventricles, can be identified in the V1*FA images during embryonic days 13–17 (E13–E17), with high anisotropy and unique orientation that is perpendicular to the ventricular surface. The neuroepithlium has high intensity in the V1*CL (linear index) and low intensity in the V3*CP (planar index) images, suggesting highly tubular anisotropy, which is consistent with its known columnar structure. The cortical plate (blue arrows) also has highly tubular diffusion similar to that in the neuroepithelium. During E14–E16, the intermediate zone (green arrows) is suppressed in the V1*CL images compared to the V1*FA images. The low intensity of the intermediate zone in the CL-based contrast is due to its planar-type anisotropy that can be clearly identified in the V3*CP images. The planar property of the anisotropy is expected in this period because of the contributions from radial glia and axons (phase II). In later stages (E17 and later), as the axons become the dominant component and the neuroepithelium and radial glia start to diminish (phase III), anisotropy of the intermediate zone becomes more tubular and less planar. At E17, the shape of the newly formed corpus callosum (yellow arrows) can be appreciated in the V1*FA and V1*CL images, but not in CP, suggesting a highly tubular nature of the diffusion.

Figure 25.7 DTMI color maps of embryonic mouse cortex. Three-dimensional schematic diagrams of embryonic cortical cytoarchitecture are shown on the left for mouse brain images at embryonic day 13 (E13) (phase I), E13–E15 (phase II), and E15–P0 (phase III). In the diagrams, the orange and blue cylinders represent neurons in the neuroepithelium (in the ventricular zone) and the cortical plate (precursor of cerebral cortex), respectively. The light blue cylinders represent radial glials, and the green cylinders represent early axons. Coronal sections from the average diffusion-weighted images (DWI) and DTMI color images are shown from E13 to E17 (corresponding to the regions within the boxes in the DWI images). Important embryonic cortical structures, including the neuroepithelium (orange arrows), cortical plate (blue arrows), intermediate zone (green arrows), and early corpus callosum (yellow arrows), are presented in these images. Abbreviations are CxP, cortical plate; IZ, intermediate zone; NE, neuroepithelium; VS, ventricular surface. At the bottom, profiles of FA, CL, and CP in E13, E15, and E17 mouse cortex are shown. The scale bars = 2 mm.

Figure 25.7
DTMI color maps of embryonic mouse cortex. Three-dimensional schematic diagrams of embryonic cortical cytoarchitecture are shown on the left for mouse brain images at embryonic day 13 (E13) (phase I), E13–E15 (phase II), and E15–P0 (phase III). In the diagrams, the orange and blue cylinders represent neurons in the neuroepithelium (in the ventricular zone) and the cortical plate (precursor of cerebral cortex), respectively. The light blue cylinders represent radial glials, and the green cylinders represent early axons. Coronal sections from the average diffusion-weighted images (DWI) and DTMI color images are shown from E13 to E17 (corresponding to the regions within the boxes in the DWI images). Important embryonic cortical structures, including the neuroepithelium (orange arrows), cortical plate (blue arrows), intermediate zone (green arrows), and early corpus callosum (yellow arrows), are presented in these images. Abbreviations are CxP, cortical plate; IZ, intermediate zone; NE, neuroepithelium; VS, ventricular surface. At the bottom, profiles of FA, CL, and CP in E13, E15, and E17 mouse cortex are shown. The scale bars = 2 mm.

The V1*CL and V3*CP images clearly demonstrate their usefulness to differentiate various tissue compartments. For example, leading edges of the growing cortical plate (indicated by red arrows) and the intermediate zone are ambiguous in V1*FA maps (E14–E16). This ambiguity makes segmentation of these structures in the cingulate cortex (white arrows) especially difficult at E15 and E16 in the V1*FA maps. In comparison, the V1*CL and V3*CP images have clear advantages in segmentation of these structures. For example, they demonstrate that both the cortical plate and the axonal layer are present in the cingulate cortex at E16. Similarly, the morphology of the neuroepithelium (orange arrows) during development can be described most clearly by the V1*CL images.

DTMI: The Applications

In this section, we will discuss the applications of DTMI in studying brain development, from rat to human.

DTMI of Rat Postnatal Cortical Development

The cortical development in the rat has been well studied using histological methods (Altman and Bayer, 1977; Bayer and Altman, 1991; Bayer et al., 1991, 1994). In the rat cortex, dendritic growth occurs mostly at about postnatal day 1 (P1) to P7, followed by synapse formation at P7–P14 (Ince-Dunn et al., 2006). As described in the previous section, embryonic cortex has a high degree of diffusion anisotropy. During postnatal development, diffusion anisotropy in the cortex gradually decreases. It is hypothesized that the drop in cortical FA is caused by the gradual degeneration of the radial glia and increased dendritic density (Sidman and Rakic, 1973; McAllister, 2000). In a recent study, Huang et al. (2008) examined postnatal changes in cortical FA in different regions of the rat cortex using ex vivo DTMI. The DTMI data were acquired using the diffusion-weighted MSE sequence on a 9.4 Tesla spectrometer with a b-value of 1000 s/mm2, diffusion time of 15 ms, and imaging resolution around 0.1 mm × 0.1 mm × 0.1 mm. The study covered five developmental time points, P0, P3, P7, P11, and P19. The FA values were measured in various cortical areas as a quantitative marker of anatomical changes, and heterogeneous changes in FA in different cortical areas were discovered.

Figures 25.8a and 25.8b show two coronal slices of FA maps and DTI color maps (V1*FA) of postnatal rat brains at the five selected time points (P0–P19). The cortical parcellation was achieved by overlaying the Paxinos atlas (Paxinos and Watson, 2005) on the DTMI images. Figure 25.8a and 25.8b also show the measurement points for cingulate, prelimbic cortex, and somatosensory cortex. The definitions of the inner, middle, and outer layers of the cortical plate are shown in the inset of Figure 25.8a.

Figure 25.8 a, b) FA maps and DTI color maps of developing rat brain at two mid-coronal slices. FA maps and DTI color maps of the right and left hemispheres of the P0, P3, P7, P11, and P19 rat brains are shown. The FA map of a P19 brain is enlarged and the anatomical definition from the Paxinos atlas is superimposed. Yellow dots indicate the locations of sampling points for cortical FA measurements. Sampling points C1–C3 are located in the cingulate (C1, C2) and prelimbic cortex (C3), M1–M2 are located in the motor cortex, S1–S4 are located in the sensory cortex, and I1 and I2 are located in the auditory cortex. The inset in a) illustrates the three evenly divided segments in the cortical plate. For the color maps, red, green, and blue indicate structures aligning along left–right, anterior–posterior, and superior–inferior orientations, respectively. Abbreviations used in the Paxinos atlas for the structures near the sampling points are as follows: Cg, cingulate cortex; DP, dorsal peduncular cortex; DTT, dorsal tenia tecta; Ins, insular cortex; M, motor cortex; M1–M2, primary and secondary motor cortex; PrL, prelimbic cortex; RSA, retrosplenial agranular cortex; RSG, retrosplenial granular cortex; S1BF, primary somatosensory cortex, barrel field; S1ULp, primary somatosensory cortex, upper lip region; S2, secondary somatosensory cortex. c) FA measurement results from various cortical regions during the P0–P19 period. The FA values averaged over all measurement points are shown as reference curves with error bars indicating standard deviations.

Figure 25.8
a, b) FA maps and DTI color maps of developing rat brain at two mid-coronal slices. FA maps and DTI color maps of the right and left hemispheres of the P0, P3, P7, P11, and P19 rat brains are shown. The FA map of a P19 brain is enlarged and the anatomical definition from the Paxinos atlas is superimposed. Yellow dots indicate the locations of sampling points for cortical FA measurements. Sampling points C1–C3 are located in the cingulate (C1, C2) and prelimbic cortex (C3), M1–M2 are located in the motor cortex, S1–S4 are located in the sensory cortex, and I1 and I2 are located in the auditory cortex. The inset in a) illustrates the three evenly divided segments in the cortical plate. For the color maps, red, green, and blue indicate structures aligning along left–right, anterior–posterior, and superior–inferior orientations, respectively. Abbreviations used in the Paxinos atlas for the structures near the sampling points are as follows: Cg, cingulate cortex; DP, dorsal peduncular cortex; DTT, dorsal tenia tecta; Ins, insular cortex; M, motor cortex; M1–M2, primary and secondary motor cortex; PrL, prelimbic cortex; RSA, retrosplenial agranular cortex; RSG, retrosplenial granular cortex; S1BF, primary somatosensory cortex, barrel field; S1ULp, primary somatosensory cortex, upper lip region; S2, secondary somatosensory cortex. c) FA measurement results from various cortical regions during the P0–P19 period. The FA values averaged over all measurement points are shown as reference curves with error bars indicating standard deviations.

Figure 25.8c shows the FA values of the outer and inner layers of the cortex for cingulate (C1–C2) / prelimbic (C3) cortex and somatosensory (S1–S4) cortex. For both cortical regions, the FA values of the inner layers lie within a small range (mostly 0.15–0.25). The age-dependent loss of FA can be clearly appreciated in the outer layer, and each cortical area has a unique time course. Dendrite maturation studies (McAllister, 2000; Whitford et al., 2002; Libersat and Duch, 2004) have suggested that the morphogenesis of dendritic trees is regulated by innate genetic factors, neuronal activity, and external molecular cues. During development, neuronal activity induced by the afferent stimulation of the visual and auditory systems occurs later than that of the somatosensory system. The unique pattern of time courses of FA maybe related to the neuronal activity timeline of these systems.

High-Resolution DTI of Adult Macaque Brain

Because of the similarity to the human brain, animal models of primates have been unique and irreplaceable to understand functions of somatosensory, auditory, and visual systems (e.g., Hendry and Yoshioka, 1994; Steinmetz et al., 2000; Bendor and Wang, 2005). Histology-based monkey brain atlases and chemical tracing studies (Martin and Bowden, 2000; Kamper and Bozkurt, 2002; Schmahmann and Pandya, 2006) provide detailed information on monkey brain anatomy. DTMI is noninvasive and DTMI data can be readily used to three-dimensionally reconstruct neural structures and white matter tracts with tractography. Hence DTMI can be used to study the primate brain anatomy.

Figure 25.10 Three-dimensional reconstruction of three association tracts in the macaque brain: the inferior fronto-occipital fasciculus (ifo: yellow fibers) and inferior longitudinal fasciculus (ilf: blue fibers) (a), and the uncinate fasciclus (unc: green fibers) and arcuate fasciclus (arc: red fibers) (b). The dashed yellow line in b) is the Sylvian fissure.

Figure 25.10
Three-dimensional reconstruction of three association tracts in the macaque brain: the inferior fronto-occipital fasciculus (ifo: yellow fibers) and inferior longitudinal fasciculus (ilf: blue fibers) (a), and the uncinate fasciclus (unc: green fibers) and arcuate fasciclus (arc: red fibers) (b). The dashed yellow line in b) is the Sylvian fissure.

After fixing a rhesus macaque (Macaca mulatta) brain with perfusion fixation, Huang and colleagues (2006a) collected DTMI and T2-weighted MRI data at room temperature using the diffusion-weighted MSE sequence on a 4.7 Tesla spectrometer with a b-value of 1000 s/mm2, diffusion time of 15 ms, field of view of 80 mm × 58 mm × 60 mm, and imaging resolution of 0.3 mm × 0.4 mm × 0.5 mm. The data acquisition time was 20 hours for DTMI and 8 hours for T2-weighted images. The high resolution of the data enabled delineation of almost all white matter fiber tracts recorded in literature.

Figure 25.9 shows the annotation of gray and white matter structures in axial and coronal views of DTI color maps (V1*FA) and T2-weighted images. The white matter assignment is based on orientation information in the color maps, and gray matter and gyral assignment is based on the contrasts in the T2-weighted image. In the anterior limb of the internal capsule (alic), the striatopallidal (caudate-putamen) projection can be clearly identified (red streaks in the mostly green alic), which has not been identified clearly in human brain imaging because of limits in resolution. In addition to the major white matter bundles, many small tracts can also be clearly identified, for example, the central tegmental tract and the medial lemniscus fasciculus in the brainstem in Figure 25.9. From the DTMI data, major white matter tracts in the macaque brain can be reconstructed.

Figure 25.9 Axial (left) DTMI color maps (V1 * FA, a–c) and T2-weighted images (d–f) of macaque brain with structural annotations. Structural abbreviations are as follows: ac, anterior commissue; alic, anterior limb of internal capsule; cc, corpus callosum; cg, cingulum; cp, cerebral peduncle; cst, corticospinal tract; ctt, central tagmental tract; dscp, decussation of superior cerebellar peduncle; fx, fornix; gcc, genu of corpus callosum; ic, internal capsule; icp, inferior cerebellar peduncle; ifo, inferior fronto-occipital peduncle; ilf, inferior longitudinal fasciculus; mcp, middle cerebellar peduncle; mlf, medial longitudinal fasciculus; pc, posterior commissure; oc, optical chiasm; or, optical radiation; ot, optical tract; plic, posterior limb of internal capsule; scc, splenium of corpus callosum; scp, superior cerebellar peduncle; scr, superior region of internal capsule; slf, superior longitudinal fasciculus; AcgG, anterior cingulated gyrus; AnG, angular gyrus; BS, brainstem; Caud, caudate nucleus; Cun, cuneus; FOG, fronto-orbital gyrus; FuG, fusiform gyrus; GP, globus pallidus; Hip, hippocampus; HyTha, hypothalamus; IOG, inferior occipital gyrus; Ins, insula; ITG, inferior temporal gyrus; LIG, lingual gyrus; MFG, middle frontal gyrus; MorG, medial orbital gyrus; MTG, middle temporal gyrus; Pcu, precuneus; OG, occipital gyrus; PHG, parahippocampal gyrus; PoG, postcentral gyrus; PrG, precentral gyrus; Put, putamen; SFG, superior frontal gyrus; SMG, supramarginal gyrus; SPL, superior parietal lobule; STG, superior temporal gyrus; SuTha, subthalamus; Tha, thalamus; Ver, vermis.

Figure 25.9
Axial (left) DTMI color maps (V1 * FA, a–c) and T2-weighted images (d–f) of macaque brain with structural annotations. Structural abbreviations are as follows: ac, anterior commissue; alic, anterior limb of internal capsule; cc, corpus callosum; cg, cingulum; cp, cerebral peduncle; cst, corticospinal tract; ctt, central tagmental tract; dscp, decussation of superior cerebellar peduncle; fx, fornix; gcc, genu of corpus callosum; ic, internal capsule; icp, inferior cerebellar peduncle; ifo, inferior fronto-occipital peduncle; ilf, inferior longitudinal fasciculus; mcp, middle cerebellar peduncle; mlf, medial longitudinal fasciculus; pc, posterior commissure; oc, optical chiasm; or, optical radiation; ot, optical tract; plic, posterior limb of internal capsule; scc, splenium of corpus callosum; scp, superior cerebellar peduncle; scr, superior region of internal capsule; slf, superior longitudinal fasciculus; AcgG, anterior cingulated gyrus; AnG, angular gyrus; BS, brainstem; Caud, caudate nucleus; Cun, cuneus; FOG, fronto-orbital gyrus; FuG, fusiform gyrus; GP, globus pallidus; Hip, hippocampus; HyTha, hypothalamus; IOG, inferior occipital gyrus; Ins, insula; ITG, inferior temporal gyrus; LIG, lingual gyrus; MFG, middle frontal gyrus; MorG, medial orbital gyrus; MTG, middle temporal gyrus; Pcu, precuneus; OG, occipital gyrus; PHG, parahippocampal gyrus; PoG, postcentral gyrus; PrG, precentral gyrus; Put, putamen; SFG, superior frontal gyrus; SMG, supramarginal gyrus; SPL, superior parietal lobule; STG, superior temporal gyrus; SuTha, subthalamus; Tha, thalamus; Ver, vermis.

Among all the tracts, association tracts that connect different parts of the cortical areas are unique for human and high-level mammalian species. The human brain is known to have the most sophisticated association tracts. Tracing the association tracts in the macaque brain provides a window for comparative neuroanatomy studies. Figure 25.10 shows 3D reconstruction of three major association tracts in the macaque brain—the inferior longitudinal fasciculus, the inferior fronto-occipital fasciculus, and the uncinate fasciculus.

DTMI was also applied to study fetal brain development (Huang et al., 2006b, 2009). The details of human fetal imaging can be found in Chapter 30 by Hüppi.

Summary

Although DTI is now routinely performed in clinics, DTMI and its applications to basic research using animal models have just started. This chapter describes the principles, challenges, practical techniques, and potential applications of DTMI. With the recent availability of high-field animal imaging instruments and advances in gradient and shimming systems, successful DTMI-based studies and new applications of DTMI have been increasingly reported. Future advancements in DTMI, in terms of resolution, contrast, and throughput, will certainly benefit biomedical researchers in related fields.

References

Ahrens ET, Laidlaw DH, Readhead C, Brosnan CF, Fraser SE, Jacobs RE (1998). MR microscopy of transgenic mice that spontaneously acquire experimental allergic encephalomyelitis. Magn Reson Med 40(1):119–132.Find this resource:

Alexander AL, Hasan K, Kindlmann G, Parker DL, Tsuruda JS (2000). A geometric analysis of diffusion tensor measurements of the human brain. Magn Reson Med 44:283–291.Find this resource:

Altman J, Bayer SA (1977). Time of origin and distribution of a new cell type in the rat cerebellar cortex. Exp Brain Res 29(2):265–274.Find this resource:

Basser PJ, Mattiello J, LeBihan D (1994). MR diffusion tensor spectroscopy and imaging. Biophys J 66:259–267.Find this resource:

Basser PJ, Pajevic S (2000). Statistical artifacts in diffusion tensor MRI (DT-MRI) caused by background noise. Magn Reson Med 44(1):41–50.Find this resource:

Basser PJ, Pajevic S, Pierpaoli C, Duda J, Aldroubi A (2000). In vivo fiber tractography using DT-MRI data. Magn Reson Med 44:625–632.Find this resource:

Bayer SA, Altman J (1991). Neocortical Development. New York: Raven Press.Find this resource:

    Bayer SA, Altman J, Russo RJ, Dai XF, Simmons JA (1991). Cell migration in the rat embryonic neocortex. J Comp Neurol 307(3):499–516.Find this resource:

    Bayer SA, Zhang X, Russo RJ, Altman J (1994). Three-dimensional reconstructions of the developing forebrain in rat embryos. Neuroimage 1(4):296–307.Find this resource:

    Beaulieu C (2002). The basis of anisotropic water diffusion in the nervous system—a technical review. NMR Biomed 15(7-8):435–455.Find this resource:

    Bendor D, Wang X (2005). The neuronal representation of pitch in primate auditory cortex. Nature 436(7054):1161–1165.Find this resource:

    Budde MD, Kim JH, Liang HF, Schmidt RE, Russell JH, Cross AH, Song SK (2007). Toward accurate diagnosis of white matter pathology using diffusion tensor imaging. Magn Reson Med 57(4):688–695.Find this resource:

    Bush EC, Allman JM (2003). The scaling of white matter to gray matter in cerebellum and neocortex. Brain Behav Evol 61(1):1–5.Find this resource:

    Callaghan PT (1991). Principles of Nuclear Magnetic Resonance Microscopy. London: Oxford University Press.Find this resource:

      Catani M, Howard RJ, Pajevic S, Jones DK (2002). Virtual in vivo interactive dissection of white matter fasciculi in the human brain. Neuroimage 17:77–94.Find this resource:

      Clark CA, Hedehus M, Moseley ME (2001). Diffusion time dependence of the apparent diffusion tensor in healthy human brain and white matter disease. Magn Reson Med 45(6):1126–1129.Find this resource:

      Conturo TE, Lori NF, Cull TS, Akbudak E, Snyder AZ, Shimony JS, McKinstry RC, Burton H, Raichle ME (1999). Tracking neuronal fiber pathways in the living human brain. Proc Natl Acad Sci U S A 96:10422–10427.Find this resource:

      Hendry SH, Yoshioka T (1994). A neurochemically distinct third channel in the macaque dorsal lateral geniculate nucleus. Science 264(5158):575–577.Find this resource:

      Hsu EW, Setton LA (1999). Diffusion tensor microscopy of the intervertebral disc anulus fibrosus. Magn Reson Med 41(5):992–999.Find this resource:

      Huang H, Walker M, Ying S, Zee D, van Zijl PC, Mori S (2006). 3D DT-MRI atlas of macaque brain. International Society of Magnetic Resonance in Medicine 14th Scientific Meeting, Seattle.Find this resource:

        Huang H, Xue R, Zhang J, Ren T, Richards LJ, Yarowsky P, Miller MI, Mori S (2009). Anatomical characterization of human fetal brain development with diffusion tensor magnetic resonance imaging. J Neurosci 29(13):4263–4273.Find this resource:

        Huang H, Yamamoto A, Hossain MA, Younes L, Mori S (2008). Quantitative cortical mapping of fractional anisotropy in developing rat brains. J Neurosci 28(6):1427–1433.Find this resource:

        Huang H, Zhang J, Wakana S, Zhang W, Ren T, Richards LJ, Yarowsky P, Donohue P, Graham E, van Zijl PC, Mori S (2006). White and gray matter development in human fetal, newborn and pediatric brains. Neuroimage 33(1):27–38.Find this resource:

        Hüppi PS, Inder TE (2001). Magnetic resonance techniques in the evaluation of the perinatal brain: recent advances and future directions. Semin Neonatol 6(2):195–210.Find this resource:

        Hüppi PS, Maier S, Peled S, Zientara GP, Barnes PD, Jolesz FA, Volpe JJ (1998). Microstructural development of human newborn cerebral white matter assessed in vivo by diffusion tensor magnetic resonance imaging. Pediatr Res 44:584–590.Find this resource:

        Ince-Dunn G, Hall BJ, Hu SC, Ripley B, Huganir RL, Olson JM, Tapscott SJ, Ghosh A (2006). Regulation of thalamocortical patterning and synaptic maturation by NeuroD2. Neuron 49(5):683–695.Find this resource:

        Kamper L, Bozkurt A (2002). An introduction to CoCoMac-online. The online-interface of the primate connectivity database CoCoMac. In Kotter R (Ed.), Neuroscience Databases—A Practical Guide. Norwell, MA: Kluwer Academic, pp. 155–169.Find this resource:

          Kim JH, Trinkaus K, Ozcan A, Budde MD, Song SK (2007). Postmortem delay does not change regional diffusion anisotropy characteristics in mouse spinal cord white matter. NMR Biomed 20(3):352–359.Find this resource:

          Kim S, Chi-Fishman G, Barnett AS, Pierpaoli C (2005). Dependence on diffusion time of apparent diffusion tensor of ex vivo calf tongue and heart. Magn Reson Med 54(6):1387–1396.Find this resource:

          Kingsley PB, Monahan WG (2004). Selection of the optimum b factor for diffusion-weighted magnetic resonance imaging assessment of ischemic stroke. Magn Reson Med 51(5):996–1001.Find this resource:

          Kozlowski P, Raj D, Liu J, Lam C, Yung AC, Tetzlaff W (2008). Characterizing white matter damage in rat spinal cord with quantitative MRI and histology. J Neurotrauma 25(6):653–676.Find this resource:

          Kroenke CD, Taber EN, Leigland LA, Knutsen AK, Bayly PV (2009). Regional patterns of cerebral cortical differentiation determined by diffusion tensor MRI. Cereb Cortex 19(12):2916–2929.Find this resource:

          Le Bihan D (2003). Looking into the functional architecture of the brain with diffusion MRI. Nat Rev Neurosci 4(6):469–480.Find this resource:

          Libersat F, Duch C (2004). Mechanisms of dendritic maturation. Mol Neurobiol 29(3):303–320.Find this resource:

          Martin R, Bowden DM (2000). Primate Brain Maps: Structure of the Macaque Brain. Amsterdam: Elsevier.Find this resource:

            McAllister AK (2000). Cellular and molecular mechanisms of dendrite growth. Cereb Cortex 10(10):963–973.Find this resource:

            Mori S, Crain BJ, Chacko VP, van Zijl PC (1999). Three dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann. Neurol. 45:265–269.Find this resource:

            Mori S, Itoh R, Kaufmann WE, van Zij lPC, Solaiyappan M, Yarowsky P (2001). Diffusion tensor imaging of the developing mouse brain. Magn Reson Med 46(1):18–23.Find this resource:

            Mori S, Kaufmann WE, Davatzikos C, Stieltjes B, Amodei L, Fredericksen K, Pearlson GD, Melhem ER, Solaiyappan M, Raymond GV, Moser HW, van Zijl PC (2002). Imaging cortical association tracts in human brain. Magn Reson Med 47:215–223.Find this resource:

            Mori S, van Zijl PC (1998). A motion correction scheme by twin-echo navigation for diffusion weighted magnetic resonance imaging with multiple RF echo acquisition. Magn Reson Med 40:511–516.Find this resource:

            Mori S, Van Zijl PC (2002). Fiber tracking: principles and strategies—a technical review. NMR Biomed 15(7-8):468–480.Find this resource:

            Mori S, Zhang J (2006). Principles of diffusion tensor imaging and its applications to basic neuroscience research. Neuron 51(5):527–539.Find this resource:

            Neil J, Shiran S, McKinstry RC, Schefft GL, Snyder AZ, Almli CR, Akbudak E, Aronovitz JA, Miller JP, Lee BC, Conturo TE (1998). Normal brain in human newborns: apparent diffusion coefficient and diffusion anisotropy measured by using diffusion tensor MR imaging. Radiology 209:57–66.Find this resource:

            Olivares R, Montiel J, Aboitiz F (2001). Species differences and similarities in the fine structure of the mammalian corpus callosum. Brain Behav Evol 57(2):98–105.Find this resource:

            Pajevic S, Pierpaoli C (1999). Color schemes to represent the orientation of anisotropic tissues from diffusion tensor data: application to white matter fiber tract mapping in the human brain. Magn Reson Med 42:526–540.Find this resource:

            Paxinos G, Watson C (2005). The Rat Brain in Stereotaxic Coordinates. Burlington, MA: Elsevier Academic Press.Find this resource:

              Pierpaoli C, Barnett A, Pajevic S, Chen R, Penix LR, Virta A, Basser P (2001). Water diffusion changes in Wallerian degeneration and their dependence on white matter architecture. Neuroimage 13(6 Pt 1):1174–1185.Find this resource:

              Pierpaoli C, Guerrini UC, Marenco S, Basser P, Barnett A, Pluta R (2002). DT-MRI reveals different architectural patterns in the cerebral and cerebellar cortex. International Society for Magnetic Resonance in Medicine 10th Scientific Meeting, Honolulu.Find this resource:

                Pierpaoli C, Jezzard P, Basser PJ, Barnett A, Di Chiro G (1996). Diffusion tensor MR imaging of human brain. Radiology 201:637–648.Find this resource:

                Poonawalla AH, Hasan KM, Gupta RK, Ahn CW, Nelson F, Wolinsky JS, Narayana PA (2008). Diffusion-tensor MR imaging of cortical lesions in multiple sclerosis: initial findings. Radiology 246(3):880–886.Find this resource:

                Reese TG, Heid O, Weisskoff RM, Wedeen VJ (2003). Reduction of eddy-current-induced distortion in diffusion MRI using a twice-refocused spin echo. Magn Reson Med 49(1):177–182.Find this resource:

                Schmahmann JD, Pandya DN (2006). Fiber Pathways of the Brain. Cambridge, UK: Oxford University Press.Find this resource:

                Sidman RL, Rakic P (1973). Neuronal migration, with special reference to developing human brain: a review. Brain Res 62(1):1–35.Find this resource:

                Sizonenko SV, Camm EJ, Garbow JR, Maier SE, Inder TE, Williams CE, Neil JJ, Hüppi PS (2007). Developmental changes and injury induced disruption of the radial organization of the cortex in the immature rat brain revealed by in vivo diffusion tensor MRI. Cereb Cortex 17(11):2609–2617.Find this resource:

                Song SK, Sun SW, Ju WK, Lin SJ, Cross AH, Neufeld AH (2003). Diffusion tensor imaging detects and differentiates axon and myelin degeneration in mouse optic nerve after retinal ischemia. Neuroimage 20(3):1714–1722.Find this resource:

                Steinmetz PN, Roy A, Fitzgerald PJ, Hsiao SS, Johnson KO, Niebur E (2000). Attention modulates synchronized neuronal firing in primate somatosensory cortex. Nature 404(6774):187–190.Find this resource:

                Sun SW, Liang HF, Le TQ, Armstrong RC, Cross AH, Song SK (2006a). Differential sensitivity of in vivo and ex vivo diffusion tensor imaging to evolving optic nerve injury in mice with retinal ischemia. Neuroimage 32(3):1195–1204.Find this resource:

                Sun SW, Liang HF, Trinkaus K, Cross AH, Armstrong RC, Song SK (2006b). Noninvasive detection of cuprizone induced axonal damage and demyelination in the mouse corpus callosum. Magn Reson Med 55(2):302–308.Find this resource:

                Thornton JS, Ordidge RJ, Penrice J, Cady EB, Amess PN, Punwani S, Clemence M, Wyatt JS (1997). Anisotropic water diffusion in white and gray matter of the neonatal piglet brain before and after transient hypoxia-ischaemia. Magn Reson Imaging 15(4):433–440.Find this resource:

                Trip SA, Wheeler-Kingshott C, Jones SJ, Li WY, Barker GJ, Thompson AJ, Plant GT, Miller DH (2006). Optic nerve diffusion tensor imaging in optic neuritis. Neuroimage 30(2):498–505.Find this resource:

                Tyszka JM, Readhead C, Bearer EL, Pautler RG, Jacobs RE (2006). Statistical diffusion tensor histology reveals regional dysmyelination effects in the shiverer mouse mutant. Neuroimage 29(4):1058–65.Find this resource:

                Werring DJ, Toosy AT, Clark CA, Parker GJ, Barker GJ, Miller DH, Thompson AJ (2000). Diffusion tensor imaging can detect and quantify corticospinal tract degeneration after stroke. J Neurol Neurosurg Psychiatry 69(2):269–272.Find this resource:

                Westin CF, Maier SE, Mamata H, Nabavi A, Jolesz FA, Kikinis R. (2002). Processing and visualization for diffusion tensor MRI. Med Image Anal 6:93–108.Find this resource:

                Whitford KL, Dijkhuizen P, Polleux F, Ghosh A (2002). Molecular control of cortical dendrite development. Annu Rev Neurosci 25:127–149.Find this resource:

                Xue R, van Zijl PC, Crain BJ, Solaiyappan M, Mori S (1999). In vivo three-dimensional reconstruction of rat brain axonal projections by diffusion tensor imaging. Magn Reson Med 42(6):1123–1127.Find this resource:

                Zhang J, Richards LJ, Yarowsky P, Huang H, van Zijl PC, Mori S (2003). Three-dimensional anatomical characterization of the developing mouse brain by diffusion tensor microimaging. Neuroimage 20(3):1639–1648.Find this resource:

                Zhang J, van Zijl PC, Mori S (2002). Three-dimensional diffusion tensor magnetic resonance microimaging of adult mouse brain and hippocampus. Neuroimage 15(4):892–901.Find this resource:

                Zhang J, van Zijl PC, Mori S (2006). Image contrast using the secondary and tertiary eigenvectors in diffusion tensor imaging. Magn Reson Med 55(2):439–449.Find this resource: