Diffusion tensor magnetic resonance imaging reveals visual pathway damage that correlates with clinical severity in glaucoma

Original Article

Diffusion tensor magnetic resonance imagingreveals visual pathway damage that correlateswith clinical severity in glaucomaZhiqi Chen MD,1* Fuchun Lin PhD,3* Junming Wang MD,1 Zheng Li PhD,1 Hui Dai MD,2 Ketao Mu MD,2

Jian Ge MD PhD4 and Hong Zhang MD PhD1

1Department of Ophthalmology, Tongji Hospital, Tongji Medical College, 2Radiology Department, Tongji Hospital, Tongji MedicalCollege, Huazhong University of Science and Technology, 3Wuhan Center for Magnetic Resonance, State Key Laboratory of MagneticResonance and Atomic and Molecular Physics, Wuhan Institute of Physics and Mathematics, Chinese Academic of Sciences, Wuhanand 4State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangzhou, China

ABSTRACT

Background: To investigate nerve fibre damage of thevisual pathway in patients with primary open-angleglaucoma using tract-based spatial statistical analysisof diffusion tensor imaging and correlate these mea-sures with the clinical severity of glaucoma.

Design: Cross-sectional study.

Participants: Twenty-five individuals with primaryopen-angle glaucoma and 24 healthy controls wererecruited.

Methods: All subjects underwent detailed ophthal-mological examinations, including the cup-to-discratio, retinal nerve fibre layer thickness and visualfields test. Diffusion tensor imaging of the visualpathway was performed using a 3.0-T magnetic reso-nance scanner.

Main Outcome Measures: Diffusivity changes of thenerve fibres in the visual pathway were calculatedthrough tract-based spatial statistical analysis. Themean diffusivity and fractional anisotropy were

assessed and compared with ophthalmologicalmeasurements.

Results: Compared with controls, bilateral optic tractsand optic radiations in primary open-angle glaucomapatients showed significantly decreased fractionalanisotropy and increased mean diffusivity (P < 0.05).In the glaucoma group, the fractional anisotropy ofthe optic tracts and optic radiations varied consis-tently with the cup-to-disc ratio, retinal nerve fibrelayer thickness and visual function analysis, respec-tively (P < 0.05). The mean diffusivity of the optictracts correlated with these ophthalmological mea-surements (P < 0.05). However, no significant corre-lation was observed between the mean diffusivityof the optic radiations and the ophthalmologicalmeasurements (P > 0.05).

Conclusions: The optic tracts and optic radiationsof primary open-angle glaucoma patients demon-strated radiological evidence of neurodegeneration.This varied with damage to the optic disc and withthe loss of visual function. Tract-based spatial

� Correspondence: Dr Hong Zhang, Department of Ophthalmology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and

Technology, Wuhan 430030, China. Email: [email protected]; and Dr Jian Ge, State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic

Center, Sun Yat-sen University, Guangzhou 510060, China. Email: [email protected]

Received 26 December 2011; accepted 19 April 2012.

*Zhiqi Chen and Fuchun Lin contributed equally to this work.

Competing/conflicts of interest: No stated conflict of interest.

Funding sources: The work was supported by the Open Project Grant at State Key Laboratory of Ophthalmology, Zhongshan Ophthalmic

Center, Sun Yat-sen University, Guangzhou, China; Natural Science Foundation of China (Nos. 81170842, 30800252 and 20921004) and the

Knowledge Innovation Program of Chinese Academy of Sciences.

bs_bs_banner

Clinical and Experimental Ophthalmology 2013; 41: 43–49 doi: 10.1111/j.1442-9071.2012.02832.x

© 2012 The AuthorsClinical and Experimental Ophthalmology © 2012 Royal Australian and New Zealand College of Ophthalmologists

statistical analysis of diffusion tensor imaging is anobjective and effective tool for detecting the lossof cortical nerve fibres in primary open-angleglaucoma.

Key words: diffusion tensor imaging, neurodegenera-tive disease, primary open-angle glaucoma, tract-basedspatial statistics.

INTRODUCTION

Glaucoma affects over 70 million people and is thesecond leading cause of blindness in the world.1 Itis a multifactorial, progressive and neurodegenera-tive optic disease characterized by the loss of retinalganglion cells (RGCs) and consequent visual fielddefects.2,3 Because of the asymptomatic courseand progression of primary open-angle glaucoma(POAG), the disease is often diagnosed at very latestages.

Glaucoma is generally considered to be a diseaseof the eye. However, recent studies have shown thatit may be a neurodegenerative disease like Alzhe-imer’s or Parkinson’s disease.4,5 Neuronal damagecaused by glaucoma is present not only in the retinabut also in the central visual system. The shrink-age and loss of lateral geniculate nucleus (LGN)neurons,6–8 as well as reductions in the dendrite com-plexity and the length of LGN neurons, are welldocumented in the glaucoma literature.9,10 Astrocyteand microglial activation have also been observed inthe LGN and visual cortex.11,12 Furthermore, a rela-tive decrease in the metabolic activity of the visualcortex has been detected in studies employing cyto-chrome oxidase.13,14 Reduced cortical thickness hasalso been demonstrated in the visual cortex.8 Thisinformation has undoubtedly contributed to thebetter understanding of glaucoma pathogenesis.However, most of these studies are examinations ofexperimentally induced glaucoma. Supporting datafrom glaucoma patients are limited because of thedifficulty of assembling a sample and the shortage ofexamination techniques. Modern magnetic reso-nance (MR) neuroimaging is a non-invasive tool thathas been applied to assess the extent of glauco-matous progression and spread within the centralnervous system.

Diffusion tensor imaging (DTI) is a sensitive MRtechnique that can detect microstructural changes inwhite matter (WM).15 The most commonly assessedDTI parameters include fractional anisotropy (FA,which reflects the degree of cellular structural align-ment within fibre tracts and the structural integrityof the fibre tracts) and mean diffusivity (MD, whichmeasures the average motion of water moleculesindependently of fibre directionality). Decreased FAand increased MD values reflect axonal disruption.16

This approach has also been applied to assess neu-rodegeneration in glaucoma patients.17–20 To date,however, the diffusivity of the visual pathway inPOAG patients has not been fully characterized. Thecorrelation between these DTI measurements withoptic disc damage and the loss of visual function arealso of particular interest.

Therefore, in this study, tract-based spatial statis-tical (TBSS) analysis of DTI was applied to assess themicrostructural WM integrity of the visual pathwayin patients with POAG. The cup-to-disc ratio (CDR),retinal nerve fibre layer thickness (RNFLT) aroundthe optic disc and loss of visual field were also evalu-ated for comparison with visual pathway damagedetected by DTI.

METHODS

Subjects

Twenty-five patients with POAG (21–54 years, meanage of 34.5 years, 19 men and 6 women) and 24 age-and sex-matched non-glaucomatous controls (21–55 years, mean age of 33.6 years, 18 men and 6women) were recruited for this study by the Depart-ment of Ophthalmology, Tongji Hospital, TongjiMedical College, Huazhong University of Science andTechnology. All of the subjects underwent a compre-hensive ophthalmological examination. POAG wasdiagnosed on the basis of glaucomatous-type opticdisc cupping (in particular, the thinning of the infe-rior and/or superior rim with a CDR asymmetry of>0.2), matching glaucomatous visual field defects andthe presence of a normal anterior chamber with anopen angle.21 The CDR was evaluated throughdetailed fundoscopic examinations by two experi-enced ophthalmologists. The RNFLT around the opticdisc was assessed using stratus optical coherencetomography (Carl Zeiss Meditec, Dublin, CA, USA;scan type: fast RNFLT 3.4, scan length: 10.87 mm).Visual function was tested using the Octopus 101automated perimeter (Haag-Streit, Bern, Switzerland;Program G2). The control subjects were required tohave normal eye exams and visual fields.

Based on the Octopus assessment of the visual fielddefects, the glaucomatous eyes were stratified into sixgroups22,23 from stage 0 to stage 5. Stage 0, which wasobserved in one eye of one patient, was characterizedby increased intraocular pressure and a mean defectless than -0.8 dB. Stage 1, which was observed innine eyes of eight patients, was characterized by amean defect of -0.7 to +4.4 dB. Stage 2 was defined asa mean defect of +4.5 to +9.5 dB and was observed infour eyes of four patients. Stage 3 was defined as amean defect of +9.6 to +15.3 dB and was observedin 10 eyes of nine patients. Stage 4 was defined as amean defect of +15.4 to +23.1 dB and was observed in

44 Chen et al.


13 eyes of 10 patients. Stage 5, which was observed in13 eyes of 11 patients, was characterized by a meandefect greater than +23.2 dB.

Eyes that exhibited signs of non-glaucomatousretinal disease (such as optic neuritis and diabeticretinopathy), opaque media or other ocular diseasesthat affect the optic nerve (such as retinal vasculardiseases and ocular tumours) were excluded. Sub-jects with neurological diseases, such as Alzheimer’sdisease, Parkinson’s disease and stroke were alsoexcluded. Subjects unable to undergo MR imaging(MRI) scanning because of the implantation of apacemaker or other metal in the body, a history ofclaustrophobia or other psychological disorders werenot included in the study.

This study adhered to the tenets of the Declarationof Helsinki and was approved by the Research EthicsBoard of Tongji Hospital, Tongji Medical College,Huazhong University of Science and Technology.Written informed consent was obtained from eachsubject before participation in the study.

MRI acquisition

DTI was performed using a 3.0T General Electricscanner with an eight-channel head coil. A single-shot, echo planar, diffusion-weighted image wasacquired with bmax = 800 s/mm2, and 30 non-collineargradient encoding directions and one volume withoutdiffusion gradients. The following acquisition param-eters were also applied: repetition time = 13 000 ms;echo time = 69 ms; matrix = 128 ¥ 128; zero-filledto 256 ¥ 256; field of view = 24 ¥ 24 cm2; slicethickness = 2 mm with no gap, and number of excita-tions = 2. Axial slices were aligned along the plane ofthe anterior-posterior commissures to encompass thewhole brain, including the cerebellum.

Data processing

All of the DTI data were processed using theFunctional MRI of the Brain’s diffusion toolboxwithin Functional MRI of the Brain Software Library(http://www.fmrib.ox.ac.uk/fsl). The diffusion-weighted volumes were aligned to the corres-ponding non-diffusion-weighted (b0) image, andnon-brain tissue was removed from the b0 image.After these steps, the diffusion tensor for each voxelwas estimated using the multivariate linear fittingalgorithm. The tensor matrix was diagonalized toobtain three pairs of eigenvalues (l1, l2, l3) andeigenvectors.24 In addition, voxel-wise values of FAand MD were calculated.

Statistical analysis

Voxel-wise statistical analysis of FA volumes wasperformed using TBSS.25 In brief, FA maps were first

normalized to the Montreal Neurological Institutestandard space. Then, the registered FA images wereaveraged to generate a cross-subject mean FA image.This mean FA image was applied to create a mean FAskeleton image that represented the main fibre tracts.The mean FA skeleton was further thresholded at a FAvalue of 0.2 to exclude non-WM tissue. Following thethresholding, the realigned FA volumes were pro-jected onto the skeleton to create a skeletonized FAmap. After these steps, the skeletonized FA data werefed into the voxel-wise statistics analysis to identifyFA differences between groups. Two contrasts wereestimated: POAG patients greater than in controls andcontrols greater than in POAG patients. The resultsincluded maps corrected at the cluster level (t = 3)threshold of P < 0.05 and were fully corrected formultiple comparisons by controlling for the family-wise error rate. The results were annotated using theJohns Hopkins University (International Consortiumfor Brain Mapping) ICBM-DTI-81 WM labels atlas26

in the Montreal Neurological Institute space.Volume of interest (VOI) analysis was performed

to investigate the changes in FA and MD. The VOImasks were created based on the voxels where dif-ferences in FA were detected. Then, these VOI maskswere back-projected onto the original images of eachsubject, and the mean values of MD were calculated.

Next, using the Student’s t-test and Pearson’s cor-relation, a within-group analysis was performed toassess the correlation between the FA of regions inwhich significant differences between groups werefound and the clinical features of the POAG patients.Neural fibres of the nasal half of retina decussate andjoin the uncrossed temporal fibres of the oppositenerve to form the optic tracts, so the cumulativeclinical measurements (CDR, RNFLT and glaucomastages) of both eyes were calculated for correla-tion analysis. A P-value of <0.05 was consideredstatistically significant (Statistical Package for theSocial Sciences, Version 13.0; SPSS, Inc., Chicago,IL, USA).

RESULTS

There was no significant difference in the mean agebetween the glaucoma and control groups (P > 0.05).The CDR, RNFLT and mean visual field defects of theglaucoma group were significantly different fromthose of the control group for both eyes (P < 0.001).

TBSS analyses revealed significantly reduced FAin the bilateral optic tracts and optic radiations of thePOAG patients (Figs 1,2A) (P < 0.001). Further, VOIanalysis found significantly elevated MD in thebilateral optic tracts and optic radiations of thePOAG patients (Fig. 2B) (P < 0.05).

Table 1 shows that the FA and MD of the optictracts as well as the FA of the optic radiations varied

Visual pathway damage in glaucoma patients 45


consistently with glaucoma stage and optic discdamage in the patient group (P < 0.05). The MD ofoptic radiations was not correlated with thesevariables. No significant correlation between the dif-fusion indices and age was observed in the controlgroup (P > 0.05).

The mean difference of the FA values (controlminus glaucoma) and the MD values (glaucomaminus control) was both larger in the optictracts than in the optic radiations (0.124 � 0.087

vs. 0.066 � 0.05, P < 0.001 and 0.199 � 0.249 ¥10-3 mm2/s vs. 0.036 � 0.063 ¥ 10-3 mm2/s, P < 0.001,respectively).

DISCUSSION

In this study, we applied TBSS analysis to assess theWM integrity of the visual pathway in POAGpatients. We found significant differences in thebilateral optic tracts and optic radiations when

Figure 1. Tract-based spatial statistical (TBSS) analysis of fractional anisotropy (FA) images. Areas in red depict regions in which FA wassignificantly lower (P < 0.05) in primary open-angle glaucoma patients relative to healthy subjects. Regions showing reduced FA (red)were thickened using the TBSS fill script for emphasis. The results were overlaid onto the MNI152-T1 template and the mean FA skeleton(green). The left side of the image corresponds to the right hemisphere of the brain.

46 Chen et al.


comparing between the POAG and control groups(Fig. 1). MD was prominently increased, and FA wasdecreased in the POAG group (Fig. 1). We alsoobserved significant correlations between the diffu-sion indices for bilateral optic tracts and optic radia-tions, and the clinical glaucomatous damage to theoptic disc and the loss of visual function (Table 1).This result implies that the neurodegeneration mea-sured using diffusivity analysis can be used as anindicator of glaucoma severity.

Ninety percent of RGCs axons terminate at theLGN, then exchange neurons to the striate cortex.27

Prominent injury of the optic tracts and optic radia-tions were observed in patients with POAG in thepresent study. Glaucoma-related RGC axon losscould have led to diffusivity changes in the optictracts. In addition, in a study examining the DiluteBrown Non-Agouti/2 (DBA/2) glaucoma mousemodel, Crish et al.28 revealed that RGC axonal trans-port deficits progress in a distal-to-proximal manner.

The abnormalities first appear in the superiorcolliculus and then in the optic tract. The distal axon-opathy of RGCs may also contribute to glau-coma pathogenesis. Changes in the optic radiationssuggest that injury from POAG is not limited tothe RGCs but rather spreads trans-synaptically to thetarget central neurons. On one hand, because theoptic radiation consists of the axons of LGN neurons,the shrinkage and loss of LGN neurons observed inexperimental primate glaucoma would lead to opticradiation damage. On the other, POAG patients alsoexhibit decreased middle cerebral artery blood flowvelocity and an absence of vasoreactivity to hyper-oxia compared with controls.29 The mean flow veloc-ity of the middle cerebral arteries is significantlycorrelated with glaucomatous damage,30 and thus,the widespread cerebrovascular insufficiency inPOAG may also contribute to lesions of the WM.Regions affected by this mechanism may include theoptic radiations and optic tracts. Moreover, there was

0.800

(a)

0.600

Mea

n FA

P < 0.001

P < 0.001

P < 0.001

GlaucomaControl

P < 0.001

0.400

Right OT Right OR Left ORLeft OT

0.200

0.000

2.000

(b)

1.500

Mea

n M

D (

× 10

−3 m

m2 /s

)

P < 0.001

P < 0.001

P < 0.001

GlaucomaControl

P < 0.0011.000

Right OT Right OR Left ORLeft OT

0.500

0.000

Figure 2. Mean fractional anisotropy (FA) and mean mean diffusivity (MD) (¥10-3 mm2/s) of the optic tracts (OTs) and optic radiations(ORs) in patients with glaucoma and control subjects.

Table 1. Correlations between the FA and MD of the optic tracts and optic radiations and clinical features (loss of visual function anddamage to the optic disc)

Cumulative RNFLT Cumulative CDR Cumulative glaucoma stages

OT-FA r = 0.609, P < 0.001 r = -0.6, P < 0.001 r = -0.639, P < 0.001OT-MD r = -0.581, P < 0.001 r = 0.504, P < 0.001 r = 0.482, P < 0.001OR-FA r = 0.407, P = 0.003 r = -0.389, P = 0.005 r = -0.386, P = 0.006OR-MD r = -0.161, P = 0.265 r = -0.003, P = 0.981 r = 0.189, P = 0.189

Cumulative refers to right eye plus left eye. CDR, cup-to-disc ratio; FA, fractional anisotropy; MD, mean diffusivity; OR, optic radiations;OT, optic tracts; RNFLT, retinal nerve fibre layer thickness.



another meaningful finding in POAG patients; theoptic tract is more severely injured than the opticradiation in POAG patients, which showed somepathological features of central nervous systeminjury during POAG.

The loss of visual field was significantly correlatedwith the diffusion parameters of the optic tracts andoptic radiations in POAG patients. The visual fielddefect in glaucoma is secondary to RGC axon loss,31 inour study, the more affected the optic radiations, themore affected the visual field, so we speculated thatthe visual field defect may be also directly related tooptic radiation damage. Furthermore, the diffusionparameters of optic tracts and optic radiations inPOAG patients varied consistently with the damageto the optic disc. Thus, transneuronal degenerationseems to be progressive with increasing nerve axonloss in the optic nerve. In glaucoma, a loss of approxi-mately 50–60% of the ganglion cells may already bepresent when visual field defects are detectable byperimetry.32 Degeneration of the optic nerve axon mayprecede the neuron loss.33 In vivo DTI can detect RGCaxon injury at an earlier stage in the disease thanimmunohistochemistry evaluation, as demonstratedin a study of axon abnormalities in a mouse modelof optic nerve crush-induced glaucoma.34 Thus, theexamination of diffusivity changes in the optictracts and optic radiations may offer help for earlierdiagnosis of POAG in the future.

However, according to our data (Table 1), the FA,but not MD, of the optic radiations varied consis-tently with ophthalmological measurements. Thisresult implies that FA may be a more sensitive andreliable biomarker than MD for glaucoma evalua-tion, which is consistent with previous results.17

There was no correlation between the diffusionindices and age in normal controls. This result isinconsistent with previous studies. These DTIstudies have reported that the optic radiation thinswith increasing age in healthy controls19 and thatlower FA values are observed in the WM of olderindividuals.35,36 We speculate that we did not observethis result because the enrolled control subjects wererelatively young (most were under 50 years old)in this study. Researchers reported that this WMdecrease occurs in individuals 70 years of age andolder;37 the possible degeneration in our controlsmay not yet be detectable. Thus, further studies arenecessary to confirm these results.

Several methods have been used to process thedata collected in glaucoma DTI studies. Garaciet al.17 reported diffusivity abnormalities of the opticnerves and optic radiations in individuals withPOAG. However, the region-of-interest analysisthey used was a subjective process that requiresmanual replacement of the regions of interest. Thisanalysis has limited reproducibility and is time-

consuming and not reliable. The associationbetween optic radiation thinning and glaucomatousoptic nerve atrophy has also previously beenreported. The authors applied a semiautomated DTIsegmentation and quantification approach that mayfail to account for some parts of the optic radia-tion.18,19 Further, El-Rafei et al.20 identified local WMdegeneration in the glaucomatous optic radiationusing voxel-based morphometric analysis. Thisapproach is very complicated, does not register theDTI images adequately and includes an arbitrarychoice of smoothing factors during data processing.In the present study, we applied TBSS analysis toour DTI data. This approach is one of the mostwell-established and widely applied techniques inmorphometry analysis. It is completely automatedand observer-independent and does not requiremanual smoothing. It has been proven to result inlower data variability, and the analyses are morerobust and sensitive.25 Thus, TBSS analysis of DTIdata may be a feasible and valuable tool for exam-ining individuals with glaucoma.

In conclusion, the MRI technique developed in thisreport (TBSS-DTI) may be useful for POAG researchand has great potential for clinical applications.Our results, showing correlations between visualpathway damage and clinical severity, also enhanceour understanding of the trans-synaptic degenera-tion in POAG. The entire visual pathway shouldbe considered in both clinical practice and drugdevelopment.

REFERENCES

1. Quigley HA, Broman AT. The number of people withglaucoma worldwide in 2010 and 2020. Br J Ophthalmol2006; 90: 262–7.

2. Barkana Y, Belkin M. Neuro-protection in ophthalmo-logy: a review. Brain Res Bull 2004; 62: 447–53.

3. Weinreb RN, Khaw PT. Primary open angle glaucoma.Lancet 2004; 363: 1711–20.

4. Gupta N, Yücel Y. Glaucoma as a neurodegenerativedisease. Curr Opin Ophthalmol 2007; 18: 110–4.

5. Wostyn P, Audenaert K, De Deyn PP. Alzheimer’sdisease: cerebral glaucoma? Med Hypotheses 2010; 74:973–7.

6. Ito Y, Shimazawa M, Chen YN et al. Morphologicalchanges in the visual pathway induced by experimen-tal glaucoma in Japanese monkeys. Exp Eye Res 2009;89: 246–55.

7. Sasaoka M, Nakamura K, Shimazawa M. Changes invisual fields and lateral geniculate nucleus in monkeylaser-induced high intraocular pressure model. Exp EyeRes 2008; 86: 770–82.

8. Gupta N, Ang LC, Noël de Tilly L et al. Human glau-coma and neural degeneration in intracranial opticnerve, lateral geniculate nucleus, and visual cortex.Br J Ophthalmol 2006; 90: 674–8.

48 Chen et al.


9. Gupta N, Ly T, Zhang Q et al. Chronic ocular hyperten-sion induces dendrite pathology in the lateral genicu-late nucleus of the brain. Exp Eye Res 2007; 84: 176–84.

10. Ly T, Gupta N, Weinreb RN et al. Dendrite plasticity inthe lateral geniculate nucleus in primate glaucoma.Vision Res 2011; 51: 243–50.

11. Lam D, Jim J, To E et al. Astrocyte and microglialactivation in the lateral geniculate nucleus and visualcortex of glaucomatous and optic nerve transectedprimates. Mol Vis 2009; 15: 2217–29.

12. Wang X, Sam-Wah Tay S, Ng YK. Microglial activitiesand neuronal cell death in the lateral geniculatenucleus of glaucomatous rats. Brain Res 2000; 878:136–47.

13. Yücel Y, Zhang Q, Weinreb RN et al. Effects of retinalganglion cell loss on magno-, parvo-, koniocellularpathways in the lateral geniculate nucleus and visualcortex in glaucoma. Prog Retin Eye Res 2003; 22: 465–81.

14. Yücel Y, Gupta N. Glaucoma of the brain: a diseasemodel for the study of transsynaptic neuraldegeneration. Prog Brain Res 2008; 173: 465–78.

15. Basser PJ, Mattiello J, LeBihan D. Estimation of theeffective self-diffusion tensor from the NMR spin echo.J Magn Reson B 1994; 103: 247–54.

16. Wheeler-Kingshott CA, Trip SA, Symms MR et al. Invivo diffusion tensor imaging of the human opticnerve: pilot study in normal controls. Magn Reson Med2006; 56: 446–51.

17. Garaci FG, Bolacchi F, Cerulli A et al. Optic nerve andoptic radiation neurodegeneration in patients withglaucoma: in vivo analysis with 3-T diffusion-tensorMR imaging. Radiology 2009; 252: 496–501.

18. Engelhorn T, Haider S, Michelson G, Doerfler A. Anew semi-quantitative approach for analysing 3Tdiffusion tensor imaging of optic fibres and its clini-cal evaluation in glaucoma. Acad Radiol 2010; 17:1313–6.

19. Engelhorn T, Michelson G, Waerntges S et al. Diffusiontensor imaging detects rarefaction of optic radiation inglaucoma patients. Acad Radiol 2011; 18: 764–9.

20. El-Rafei A, Engelhorn T, Wärntges S et al. A frame-work for voxel-based morphometric analysis of theoptic radiation using diffusion tensor imaging inglaucoma. Magn Reson Imaging 2011; 29: 1076–87.

21. European Glaucoma Society. Perimetry. In: Heijl Aand Traverso CE. ed. Terminology and Guidelines forGlaucoma. Savona: Dogma, 2008; 82–7.

22. Hodapp E, Parrish RK, Anderson DR. Clinical Decisionsin Glaucoma. St. Louis, MO: Mosby, 1993.

23. Mills RP, Budenz DL, Lee PP et al. Categorizing thestage of glaucoma from pre-diagnosis to end-stagedisease. Am J Ophthalmol 2006; 141: 24–30.

24. Basser PJ, Pierpaoli C. Microstructural and physiologi-cal features of tissues elucidated by quantitative diffu-sion tensor MRI. J Magn Reson B 1996; 111: 209–19.

25. Smith SM, Jenkinson M, Johansen-Berg H et al. Tract-based spatial statistics: voxelwise analysis of multi-subject diffusion data. Neuroimage 2006; 31: 1487–505.

26. Wakana S. Fiber tract-based atlas of human whitematter anatomy. Radiology 2004; 230: 77–87.

27. Perry VH, Oehler R, Cowey A. Retinal ganglion cellsthat project to the dorsal lateral geniculate nucleus inthe macaque monkey. Neuroscience 1984; 12: 1101–23.

28. Crish SD, Sappington RM, Inman DM, Horner PJ,Calkins DJ. Distal axonopathy with structural persis-tence in glaucomatous neurodegeneration. Proc NatlAcad Sci U S A 2010; 107: 5196–201.

29. Harris A, Zarfati D, Zalish M et al. Reduced cerebrovas-cular blood flow velocities and vasoreactivity in open-angle glaucoma. Am J Ophthalmol 2003; 135: 144–7.

30. Harris A, Siesky B, Zarfati D et al. Relationship of cere-bral blood flow and central visual function in primaryopen-angle glaucoma. J Glaucoma 2007; 16: 159–63.

31. Quigley HA, Dunkelberger GR, Green WR. Retinalganglion cell atrophy correlated with automatedperimetry in human eyes with glaucoma. Am J Ophthal-mol 1989; 107: 453–64.

32. Harwerth RS, Carter-Dawson L, Shen F. Ganglioncell losses underlying visual field defects from experi-mental glaucoma. Invest Ophthalmol Vis Sci 1999; 40:2242–50.

33. Buckingham BP, Inman DM, Lambert W et al. Progres-sive ganglion cell degeneration precedes neuronal lossin a mouse model of glaucoma. J Neurosci 2008; 28:2735–44.

34. Zhang X, Sun P, Wang J et al. Diffusion tensor imagingdetects retinal ganglion cell axon damage in the mousemodel of optic nerve crush. Invest Ophthalmol Vis Sci2011; 52: 7001–6.

35. Stadlbauer A, Salomonowitz E, Strunk G et al. Age-related degradation in the central nervous system:assessment with diffusion tensor imaging and quanti-tative fiber tracking. Radiology 2008; 247: 179–88.

36. Furutani K, Harada M, Minato M et al. Regionalchanges of fractiona anisotropy with normal agingusing statistical parametric mapping (SPM). J MedInvest 2005; 52: 186–90.

37. Courchesne E, Chisum HJ, Townsend J et al. Normalbrain development and aging: quantitative analysis atin vivo MR imaging in healthy volunteers. Radiology2000; 216: 672–82.



Diffusion tensor magnetic resonance imaging reveals visual pathway damage that correlates with clinical severity in glaucoma

Documents