Diffusion Tensor Imaging of WhiteMatter Tracts in the Dog Brain

OLIVIER JACQMOT,1* BERT VAN THIELEN,1 YVES FIERENS,2

MARTHA HAMMOND,3 INNEKE WILLEKENS,2 PETER VAN SCHUERBEEK,2

FILIP VERHELLE,1,2 PETER GOOSSENS,1,2 FILIP DE RIDDER,2

JAN PIETER CLARYS,4 ANNE VANBINST,2 AND JOHAN DE MEY1,2

1Department of Radiology, University Hospital Brussels, Morpho – Veterinary – Imaging(MOVE – IM BRUSSELS A.S.B.L.), UZ Brussel, Brussels, Belgium

2Department of Radiology, University Hospital Brussels (UZ Brussel), Vrije UniversiteitBrussel (VUB), Laarbeeklaan 101, 1090 Brussels, Belgium

3Department of Basic Veterinary Sciences, Service of Anatomy, Royal Veterinary College,London, United Kingdom

4Department of Anatomy, Faculty of Medicine, Vrije Universiteit Brussel (VUB),Brussels Laarbeeklaan 103, 1090 Brussels, Belgium

ABSTRACTDiffusion weighted imaging sequences are now widely available on

Magnetic Resonance Imaging (MRI) scanners. Diffusion Tensor Imaging(DTI) of the brain is able to show white matter tracts and is now com-monly used in human medicine to study brain anatomy, tumors, struc-tural pathways,. . . The purpose of this study was to show the interest ofDTI to reveal the white matter fibers in the dogs’ brain. DTI MR Imagesfor this study were obtained with a 3 T system of 4 dogs euthanized forother reasons than neurological disorders. Combined fractional anisotropic(FA) and directional maps were obtained in the first 2 hours after death.The heads were amputated immediately after scanning and stored in 10%formalin until preparation for dissection. An experienced anatomisttracked white matter tracts with clinical relevance using the scanner soft-ware. The selected tracts were volume rendered and correlated with grossdissection. Using DTI we were able to track relevant neurological connec-tions, such as the corticospinal tract, the optic and the cerebellar tract.The three dimensional anatomy is better presented using modern visual-ization techniques. DTI seems to be a valuable tool in order to presentclinically relevant white matter tracts to neurological clinicians andresearchers. Anat Rec, 296:340–349, 2013. VC 2013 Wiley Periodicals, Inc.

Key words: brain; diffusion tensor imaging; dog; anatomy

Since a few years, Magnetic Resonance Imaging is areference technique for imaging the brain in differentplanes (sagittal, transversal, coronal) and the use of1,5T to 7T MRI allows more and more accurate anddetailed visualization of white matter localization thanconventional CT-Scan and X-Ray (Van Thielen et al.,2010). However, an atlas of all the white matter tractswould be particularly useful for providing detailedanatomical data that is not available in studies basedon conventional MRI data (Lawes et al., 2008). So we

*Correspondence to: Dr. Olivier Jacqmot, Morpho – Veterinary– Imaging (MOVE – IM BRUSSELS A.S.B.L.), Brussels(Belgium), Department of Radiology, University HospitalBrussels (UZ Brussel), Laarbeeklaan 101, 1090 Brussels,Belgium. Fax: 0032 (0)10 844398. E-mail: jacqmot@skynet.be

Received 31 March 2012; Accepted 27 October 2012.

DOI 10.1002/ar.22638Published online 15 January 2013 in Wiley Online Library(wileyonlinelibrary.com).

THE ANATOMICAL RECORD 296:340–349 (2013)

VVC 2013 WILEY PERIODICALS, INC.

propose the use of Diffusion Tensor Imaging (DTI) tech-nique to study the white matter tracts of the dog brain.The imaging of the white matter tracts is obtained bypost processing of diffusion weighted data obtained withMagnetic Resonance Imaging (MRI). In this article wepresent gross dissection photographs correlated withdirectional DTI color maps to review the anatomy of thenervous tracts. However several advanced DTI postprocessing techniques are reviewed and published in theliterature, for our purposes we have used a commercialDTI software package available on the MRI scanner.Moreover a brief review of the basic principles underly-ing DTI is included in this article and several morecomprehensive reviews are available for the reader whowishes to strengthen knowledge about the technicalaspects of DTI (Mori and van Zijl, 2002). Diffusion ten-sor imaging is a relatively new technique which useswater diffusion anisotropy in axonal fibers to provide atool for analyzing and tracking those fibers in brainwhite matter. In human medicine, DTI is commonlyused in order to study the normal anatomy of normalbrain maturation and ageing (Dong et al., 2004), butalso as an aid in diagnosis and prognosis of cerebral is-chemia, multiple sclerosis, epilepsy, metabolic disordersand brain tumors (Dong et al., 2004). Normal white mat-ter brain anatomy can be studied in detail using DTI; onone side it shows a complete anatomical and statisticalfiber atlas of the white matter (Wakana et al., 2004,Mori S. et al., 2008, Oishi et al., 2008), and on the otherside it can explain, in combination with functional MRI,some anatomical and functional connectivity betweendifferent parts of the brain (Hagmann et al., 2003).During the early development from fetus to two years ofage, there is a marked increase in brain volume andalterations in image contrast on routine MRI imaging,such as an increase in WM T1 signal as the concentra-tion of cholesterol, galactocerebrosides, and proteins inmyelin are increasing, and a decrease in WM T2 signaldue to a reduction in the water content as the myelinsheaths compact with maturation (Barkovisch, 2000). Italso has proven value in the diagnosis and follow up ofmany human congenital brain malformations such ashorizontal gaze palsy with progressive scoliosis, pontinetegmental cap dysplasia, holoprosencephaly and agene-sis of the corpus callosum (Wahl et al., 2010). Thediagnosis and work-up of cerebral ischemia and braintumors can also be significantly increased using DTI(Dong et al., 2004). Both pathologies have a direct andindirect effect on the water restriction in the whitematter. Where several tumors are reflecting a morerestricted diffusion from increasing cellularity, there is areduced diffusion anisotropy and increased ADC in WDby using the MRI diffusion technique (Dong et al., 2004).DTI is also helpful in the set up of individual neuronavi-gation during brain tumor surgery (Volker et al., 2001).By using DTI as a basic model for neuronavigation, theneurosurgeon can minimize the damage to white mattertracts. For multiple sclerosis it is the quantitativeinformation of DTI that offers an added value to themorphological MRI sequences, whereas for epilepsy anincreased diffusivity and reduced anisotropy in the scle-rotic hippocampi can suggest the loss of structuralorganization and expansion of the extracellular space(Dong et al., 2004). By using DTI of the white mattertracts in the dog brain as in human medicine, it may

have the potential for use in studying several patholo-gies by allowing correlation of DTI findings and clinicalsymptoms. This technique provides a better 3 dimen-sional understanding of the inner anatomicalorganization of the dog brain.

THE PHYSICS OF DTI

Large white matter tracts are composed of numerousaxons organized in a parallel fashion. Water diffusion ingeneral, can be explained through random molecularmotion (Brownian motion), but such motion is con-strained by densely packed axonal membranes andmyelin sheaths. White Matter (WM) fiber tracts, axonalmembranes, and myelin sheaths present barriers to themotion of water molecules in no parallel directions totheir own orientation (Jellison et al., 2004). So, watermovement perpendicular to the white matter fiber tractsis restricted while the movement parallel with the tractis not. By using a pair of strong gradients1 MRI is capa-ble of visualizing the motion of water molecules. Forstatic (bound) water molecules, these gradients will haveno influence on their measured signal, while for movingmolecules, signal loss will be induced. The sensitivity tomotion can be represented by the “b-factor”. This factoris dependent on strength, duration and time lapsebetween the gradients. The direction of maximum diffu-sivity has been shown to coincide with the WM fibertract orientation (Moseley et al., 1990). The mean diffu-sivity is represented by the Apparent DiffusionCoefficient (ADC) for isotrope diffusion. For anisotropediffusion, one ADC value is not sufficient to describe itcompletely and we have to use the diffusion tensor: amathematical model of diffusion in three-dimensionalspace. In general, a tensor is a rather abstract mathe-matic entity having specific properties that enablecomplex physical phenomena to be quantified. In thepresent context, the tensor is simply a matrix of num-bers derived from diffusion measurements in severaldifferent directions, from which one can estimate the dif-fusivity in any arbitrary direction or determine thedirection of maximum diffusivity. With the use of DTI,both the degree of anisotropy and the local fiber direc-tion can be mapped voxel by voxel, providing a new andunique opportunity for studying WM architecture invivo. Whole-brain images are acquired with the diffusiongradients applied in a number of independent directions(at least six); a reference image with no applied gra-dients is also acquired. Using more than six encodingdirections will improve the accuracy of the tensor mea-surement for any arbitrary orientation (Mori and vanZijl, 2002). Information from all images, reference anddifferent diffusion gradients can then be combined toevaluate the three-dimensional profile of diffusion ineach voxel. This information is finally represented in atensor which is subjected to a linear algebric procedure(diagonalization), the result of which is a set of threeeigenvectors representing the major, medium and minorprinciple axes of the diffusion ellipsoid and the corre-sponding eigenvalues (lambda 1, lambda 2, lambda 3).These eigenvalues represent the diffusivity along thedirection of their corresponding eigenvector. Finally, the

1Gradients 5 added small magnetic fields besides the principalfield of the MRI.

DIFFUSION TENSOR IMAGING 341

fraction of the tensor that can be assigned to a certainanisotropic diffusion is represented in the FractionalAnisotropy (FA) on a map, which is often also colorcoded. The color refers then to the principal diffusiondirection. The principal eigenvector is assumed to beco-linear with the dominant fiber orientation within thevoxel (Fig. 1).

FIBER RECONSTRUCTION TECHNIQUES

Assuming that the orientation of the largest compo-nent of the diffusion tensor represents the orientationof dominant axonal tracts, DTI can provide a 3D vectorfield, in which each vector represents the (average)fiber orientation. Currently, there are several differentapproaches to reconstruct white matter tracts, whichcan be divided into two types. Techniques classified inthe first category are based on line propagationalgorithms that reconstruct a 3D trajectory from a 3Dvector field to propagate a line from a seed point byfollowing the local vector orientation. This technique isfollowing a given seed point, the most probable orienta-tion. The second type of approach is based on global

energy minimization to find the energetically mostfavorable path between two predetermined pixels orregions of interest. In summary, this kind of techniqueis looking to connect two arbitrary points in the mostfavorable way.

MATERIALS AND METHODS

Animal Sampling

Four male dogs without neurological disorders of thesame breed (American Staffordshire terriers), agedbetween 4 and 6 years, with an average weight of 30 kg,where euthanized between September and December2011, for different known clinical pathologies. Immedi-ately after euthanasia the dogs were transported to theMRI unit to perform MRI acquisition.

DTI MRI Acquisition and Directional Mapping

DTI MRI images for this article were obtained with a3 T system (Philips Healthcare, The Netherlands) byusing a neurovascular 16-channel head coil, single-shotecho planar imaging sequence (7855/55/2) (TR/TE/excita-tions), 220 mm field of view, 2.2 mm sections, 1 stack, 50slices, matrix 112 3 109 zero-filled to 224 3 224, acqui-sition voxel slice 1.96 3 2.10 3 2.20 mm interpolated to0.98 mm isotropic, diffusion encoding in 46 directionswith b 5 2,800 s/mm2, post processing with FiberTraksoftware PhilipsTM. After this DTI MRI images wereobtained, in addition to T2–weighted Turbo Echo Spinimages (3,000/80/32) (TR/TE/excitations), 230 mm fieldof view, 1 mm sections, matrix 400 3 255 mm2 zerofilled to 512 3 512 matrix, acquisition voxel size 0.57 30.72 3 1 mm interpolated to 0.45 mm isotropic voxels.

DTI Tractograms

The White Matter (WM) tracts were estimated withtractography by using the previously described algo-rithm (PhilipsTM). Tracking was initiated from a startlocation or a seed point—in this case a seed region ofinterest (ROI) in both forward and backward directions,defined by the major eigenvector at the seed point. Thepropagation was terminated when the tract trajectoryreached a voxel with an FA less than 0.15 mm (the esti-mated eigenvector direction becomes less accurate as FAdecreases and becomes very sensitive to image noise forFA less than 0.15 mm), or when the tract was less than10 mm, or when the angle between two consecutivesteps was greater than 45�. A complete set of fiber tra-jectories was obtained by placing seeds in all points orROIs where the first author (O.J.) estimated to localizewhite matter tracts. To help in localizing where to setthe seed points, the T2-weighted high-resolution imageswere superposed on the ADC-map to assist the firstauthor to recognize the connections between the differ-ent anatomical structures. Fiber trajectories were finallydisplayed with colors and volume rendered or overlaidonto the T2-weighted images. Different anatomical fiberswhere recognized and inventoried for each of the fourcanine brains. The white matter tracts were recognizedand correlated by using anatomical textbooks (Evansand Evans, 1993; Barone, 2004; Schuenke, 2010), humanDTI atlas (Dong et al., 2004; Wakana et al., 2004; Con-cha et al., 2005; Mori et al., 2008) and gross dissections.

Fig. 1. A: FA map: image of the brain seen centrally bordered bythe temporal muscle. B: Image of a tensor tracking algorithm which isapplied on the FA map at the level of the spinal cord.

342 JACQMOT ET AL.

Gross Dissections

Heads were stored immediately after scanning in a10% formalin bath until the brain specimens were man-ually removed. Holes were placed in the skull with an

oscillating saw so the formalin solution could diffusethroughout rapidly to preserve and harden the cerebraltissue. Fifteen days later, the skull bones were carefullylifted with a bone rongeur and the encephalon was pre-served in a 10% formalin solution.

Fig. 3. A: Superior and inferior occipitofrontal fasciculi in sagittalview. B: Superior and inferior occipitofrontal fasciculi in coronal view.C: Superior and inferior occipitofrontal fasciculi in transversal view.

Fig. 2. A: Lateral view of gross dissection (telencephalon has beenremoved). B: Corpus callosum and cingulum in sagittal view. C: Cor-pus callosum and cingulum in transversal view.

DIFFUSION TENSOR IMAGING 343

RESULTS

White Matter (WM) tracts traditionally have beenclassified as follows: association fibers, projection fibers,commissural fibers, other tracts such as the optic tract,fornix, tapetum, and many fibers of the brainstem andcerebellum. ROIs were used for the fiber tracking andwere chosen to enclose tract cross sections in the axial,sagittal, or coronal directional color maps. The corpuscallosum and the corticospinal tract in the brainstemwere revealed in the mid-sagittal plane, the anteriorcommissure in the sagittal plane, the cingulum in theaxial plane, the superior occipitofrontal and inferior occi-pitofrontal fasciculi in the axial and coronal planes, theuncinate fasciculus and the occipitotemporal fasciculusin the coronal and axial planes and the arcuate fibers inthe axial plane.

Association Fibers

These fibers interconnect cortical areas in each hemi-sphere. Fibers of this type typically identified on DTIcolor maps include: cingulum, superior and inferioroccipitofrontal fasciculi, uncinate fasciculus, superiorlongitudinalis fasciculus, and inferior longitudinalis(occipitotemporal) fasciculus.

The cingulum begins below the rostrum of the corpuscallosum in the parolfactory area and arches around thedorsal part of the corpus callosum in two cords extend-ing forward into the parahippocampal gyrus. It connectsthe enthorinal cortex and the cingulate gyrus, and is oneof the two most visible white matter connections of thelimbic system. It is connected to the frontal, parietal,temporal, and occipital lobes (Fig. 2).

The superior occipitofrontal fasciculus connects thefrontal and occipital lobes, and follows its course in theangle of the lateral ventricle between the caudate nucleusand the corpus callosum. The inferior occipitofrontal fasci-culus connects the frontal and occipital lobes also, but islocated beneath the superior fasciculus and follows theinferolateral edge of the claustrum (Fig. 3).

The superior longitudinal fasciculus is an associationof fibers connecting the frontal lobe cortex with the pari-etal, temporal, and occipital lobes cortices, and is morelaterally situated than the inferior occipitofrontal fasci-culus. This fasciculus is also known as the “arcuatefibers” which run between two adjacent gyri (Fig. 4).

Uncinate fibers run from the frontal lobe cortex to thetemporal lobe cortex and parahippocampic gyrus. Thesefibers lie parallel to, and underneath, the inferior occipi-tofrontal fasciculus (Fig. 5).

Fig. 4. A: Lateral view of gross dissection. B: Arcuate fibers in transversal view. C: Arcuate fibers insagittal view. D: Arcuate fibers in coronal view.

344 JACQMOT ET AL.

The inferior longitudinal fasciculus (occipitotemporalfascicuclus) joins the temporal and occipital lobes, and isinvolved in audio visual associations (Fig. 5).

Projection Fibers

These fibers interconnect cortical areas with deepnuclei, the brainstem, cerebellum, and the spinal cord.There are both efferent (corticofugal) and afferent (corti-copetal) projection fibers. Fibers of this type typicallyidentified on DTI color maps include: the corticospinal,corticobulbar, and corticopontine tracts, as well as thegeniculocalcarine tracts (optic radiations). The corticospi-nal, corticopontine, and corticobulbar tracts are majorefferent projection fibers connecting motor cortex to thebrainstem and spinal cord. The bundles of fibers radiateinto the corona radiata and continue through the inter-nal capsule to the cerebral peduncle. The program usedwas unable to discern between the different bundles(Fig. 6).

The corona radiata is easily identified on directionalDTI color maps. Its fibers pass into the superolateralpart of the internal capsule and radiate out at the fron-tal and prefrontal lobes.

Commissural Fibers

These fibers interconnect similar cortical areasbetween opposite hemispheres. Fibers of this type typi-cally identified on DTI color maps include the corpuscallosum and the anterior commissure. The corpus cal-losum is a massive accumulation of commissural fibers,transversely oriented between the two hemispheres andconnecting corresponding areas of frontal lobe corticesby the genu, parietal lobe cortices by the trunk,and temporal and occipital lobe cotices by the splenium.(Fig. 7)

The anterior commissure is made of anterior fibersconnecting olfactory bulbs and nuclei, and posteriorfibers connecting inferior temporal lobes (Fig. 8).

Other Fibers (Brainstem)

The brainstem, cerebellum, and other tracts such asthe fornix of the limbic system and optic tract are alsoidentified. The brainstem is composed of a large numberof tracts and nuclei with multiple decussations many ofwhich many can be resolved on directional DTI colormaps.

Fig. 5. A: Uncinate and occipitotemporal fasciculi in sagittal view. B: uncinate fasciculus in transversalview. C: Uncinate and occipitotemporal fasciculi in transversal view. D: Uncinate and occipitotemporalfasciculi in transversal view.

DIFFUSION TENSOR IMAGING 345

The fornix is a bundle of associated intrahemisphericand commissural fibers connecting the two hippocampi.It is part of the limbic system lying on the side of thehippocampus and surrounding the thalamus. The fornixand the cingulum are the two most visible white matterconnections. (Fig. 9)

DISCUSSION

In comparison with other anatomical techniques (suchas classical dissection and cryo-dissection) and withother imaging techniques (such as scanner, classicalMRI, and scintigraphy), DTI technique allows a non

Fig. 6. A: Lateral view of gross dissection (telencephalon has been removed). B: Corticospinal tract insagittal view. C: Corticospinal tract in coronal view. D: Corticospinal tract in dorsal view. E: Corticospinaltract in lateral view. F: Corticospinal tract in coronal view.

346 JACQMOT ET AL.

invasive 3D representation of cerebral white mattertracts and provides detailed anatomical data that is notavailable in studies based on conventional MRI data.The classical dissection technique without formaline oralcoholic fixation is not usable as cerebral tissues arevery delicate and liquefy very quickly. After formalinefixation and very fine and delicate dissection or by cryo-

dissection technique, macroscopic bundles of whitematter fibers can be isolated. But several brains arerequested as some bundles are spoiled to isolate others.The aim of our project was to investigate the use of DTIin providing a representation of the cerebral whitematter anatomy of the dog. Multiple examples of whitematter structures are described within this article. Theyare divided into a classification system dealing with theassociation fibers, the projection fibers, commissuralfibers and other tracts such as the fornix, tapetum andmany fibers of the brainstem. Generally our trackedstructures correlated well with gross dissections anddrawings in anatomical textbooks. Considering thesefindings, it can be considered that DTI – reconstructions,and their 3D datasets, could be of considerable aid torepresent in a clear way white matter anatomy to neuro-logical clinicians and researchers. However, all of ourseries of images were obtained on dead animals, and wecame across very little information in the literatureregarding the velocity with which decomposition of whitematter could have an influence on tracking results.Although image acquisition was performed on animalspost mortem, since data sets were acquired within amaximum of two hours post euthanasia, the influence ofdecomposition on track results is presumed to be

Fig. 7. A: Dorsal view of gross dissection. B: Corpus callosum insagittal view. C: Corpus callosum in transversal view.

Fig. 8. A: anterior commisure in transversal view. B: Anterior commi-sure in coronal view.

DIFFUSION TENSOR IMAGING 347

negligible. In human medicine, DTI is not only used asan anatomical atlas, but it also has widespread clinicaland research applications. In veterinary medicinereports of its use are limited to a few papers; a technicalresearch publication using the visual cortex of a cat(Ronen et al., 2003) and two canine based papers, one

validating DTI for the quantitative assessment of tumormicrostructure in canine brain (Zhu et al., 2006) and theother describing a methodology for DTI of fixed dogbrains at 7 T (Wang P. and Zhu. J.M., 2010). As of yetthere is no described or published work on its use forstudying the anatomy of white matter in domestic

Fig. 9. A: Lateral view of gross dissection (telencephalon has been removed). B: Fornix in lateral view.C: Fornix in transversal view. D: Fornix in transversal view. E: Fornix in coronal view.

348 JACQMOT ET AL.

animals. The aim of our project was to study the interestand feasibility that DTI could have to represent the cere-bral white matter anatomy of the dog to clinicians andresearchers for interpreting the neurological examina-tions conducted in clinical situations. For example theaudio-occipital reflex test (slapping the hands near theears, which normally cause a reflex of closing the eyes),testing the integrity of the occipitotemporal fibers, partof the association fibers (Fig. 4). Since visualization ofthe neurological connections is displayed in the DTIvolume rendered figures, it could be advantageous to useDTI datasets in the planning of canine neurologicalsurgeries. However we must consider the challenges toits use in practice: scanning required prior to surgery onthe 3 T MRI systems which are rarely available forveterinary cases, and the need for general templatebased datasets to be produced for image interpretation.But its potential for use as a future tool to both studyand describe canine neurological disorders (hydroce-phaly) should not be overlooked.

CONCLUSIONS

DTI-reconstructions and their 3D datasets could be ofconsiderable aid to represent in a clear way white mat-ter tracts of dog brain. A general race-specific templateof a 3D dataset could enable veterinary surgeons tostudy, and have an improved visualization, of the inter-nal structures. This information can be used to finddifferent and improved surgical access for instances dur-ing surgery of neurological disorders. As far as theauthors are aware, this is the first work that is describ-ing the anatomical white matter atlas of the dog brainvia the use of DTI.

ACKNOWLEDGEMENT

The authors thank Doctor Maryline Peters for providingthe dead dogs.

LITERATURE CITED

Barone R. 2004. Anatomie compar�ee des animaux domestiquesTome 6: Neurologie I Systeme Nerveux Central. Paris, France:Vigot Freres Ed.

Concha L, Gross D, Beaulieu C. 2005. Diffusion tensor tractographyof the limbic system. Am J Neuroradiol 26: 2267–2274.

Dong Q, Welsh R, Chenevert T, Carlos R, Maly-Sundgren P, Gomez-Hassan D, Mukherji S. 2004. Clinical applications of diffusiontensor imaging. J Magn Reson Imaging 19:6–18.

Evans SA, Evans HE. 1993. Miller’s anatomy of the dog. 3rd ed.Philadelphia, Pensylvania, USA: Saunders Ed.

Hagmann P, Thiran JP, Jonasson L, Vandergheynst P, Clarke S,Maeder P, Meuli R. 2003. DTI mapping of human brain connectiv-ity: statistical fibre tracking and virtual dissection. NeuroImage19:545–554.

Jellison B, Field A, Medow J, Lzar M, Salamat M, Alexander A.2004. Diffusion tensor imaging of cerebral white matter: a pic-toral review of physics, fiber tract anatomy, and tumor imagingpatterns. Am J Neuroradiol 25:356–369.

Lawes I, Barrick T, Murugam V, Spierings N, Evans D, Song M,Clarck C. 2008. Atlas-based segmentation of white matter tractsof the human brain using diffusion tensor tractography and com-parison with classical dissection. NeuroImage 39:62–79.

Mori S, vanZijl P. 2002. Fiber tracking: principles and strategies—atechnical review. NMR Biomed 15:468–480.

Mori S, Oishi K, Jiang H, Jiang L, Li X, Akhter K, Hua K, Faria A,Mahmood A, Woods R, Toga A, Pike G, Neto P, Evans A, Zhang J,Huang H, Miller M, vanZijl P, Mazziotta J. 2008. Stereotaxicwhite matter atlas based on diffusion tensor imaging in an ICBMtemplate. NeuroImage 40:570–582.

Moseley M, Cohen Y, Kucharczyk J, Mintorovitch J, Asgari HS,Wendland MF, Tsuruda J, Norman D. 1990. Diffusion—weightedMR imaging of anisotropic water diffusion in cat central nervoussystem. Radiology 176:439–446.

Oishi K, Zilles K, Amunts K, Faria A, Jiang H, Li X, Akhter K,Hua K, Woods R, Toga A, Pike B, Rosa-Neto P, Evans A, Zhang J,Huang H, Miller M, vanZijl P, Mazziotta J, Mori S. 2008. Humanbrain white matter atlas: identification and assignment of com-mon anatomical structures in superficial white matter. Neuro-Image 43:447–457.

Ronen I, Kim K, Garwood M, Ugurbil K, Kim D. 2003. ConventionalDTI vs. slow and fast diffusion tensors in cat visual cortex. MagnReson Med 49:785–790.

Schuenke M, Schulte E, Schumacher U. 2010. Head and Neuroan-atomy. In: Georg Thieme, editor. Atlas of Anatomy. Germany: Ver-lag Stuttgard.

Van Thielen B, Visser F, Denolin V, Antoine N, Engelen V, LuijtenP, Jacqmot O. 2010. Comparison of MRI cerebral images obtainedat 7T in a dog to macroscopic and histopathological examination.European Veterinary Diagnostic Imaging Annual Scientific Con-ference, Giessen, Germany July 20–24th.

Wahl M, Barkovich A, Mukherjee P. 2010. Diffusion imaging andtractography of congenital brain malformations. Pediatr Radiol40:59–67.

Wakana S, Jiang H, Nagae Poetscher L, vanZijl P, Mori S. 2004.Fiber tract-based atlas of human white matter anatomy. Radiol-ogy 230:77–87.

Wang P, Zhu JM. 2010. Quantitative diffusion tensor imaging ofwhite matter microstructure in dog brain at 7 T. O. M. I J. 4: 1–5.

Zhu J-M, Wang P, Robertson JL, Rossmeisl JH, Robbins ME, ShawEG. 2006. Quantitative assessment of tumor microstructure incanine brain using high resolution DTI microimaging. Proc IntSoc Magn Resonance Med 14:2924.

DIFFUSION TENSOR IMAGING 349