●

c(●

●

n(wcmswDKsae●

ttfstai●

dcmg4

A

BO(oG

Fe

0d

Detection of Abnormal Visual Cortexin Children With Amblyopia by

Voxel-Based Morphometry

JIANG XI XIAO, MD, SHENG XIE, MD, JIN TANG YE, MD, HAI HUA LIU, MD,

XIAO LING GAN, MD, GAO LANG GONG, PHD, AND XUE XIANG JIANG, MD

TstcpcstsrftgaabcsViiastyatpoiyo

●

PURPOSE: To detect the abnormalities of gray matter inhildren with amblyopia by voxel-based morphometryVBM).

DESIGN: Prospective, nonrandomized clinical trial.METHODS: Thirteen children with amblyopia and 14

ormally sighted children underwent magnetic resonanceMR) examination. The two groups were age-matchedith a mean age of 5.8 years. In the amblyopia group, five

hildren had strabismus amblyopia, and eight had aniso-etropic amblyopia. We analyzed the original 3-dimen-

ional T1 brain images using the VBM module within theidely used analysis software package SPM2 (Welcomeepartment of Cognitive Neurology, London, Unitedingdom). After normalization, segmentation, and

moothing of the images, comparison between amblyopicnd control groups was derived for the gray matter of thentire brain using parametric statistics.RESULTS: The results of VBM analysis indicated that

he amblyopic group had decreased gray matter density inhe middle frontal gyrus, parahippocampal gyrus, fusi-orm gyrus, inferior temporal gyrus of the left hemi-phere, and the bilateral calcarine cortices. The radii ofhese regions ranged from 12 to 36 voxels. Thesebnormalities were consistent with morphologic changesn brain regions related to visual function.

CONCLUSIONS: Using MR and VBM analysis, weetected morphologic changes in the visual cortex ofhildren with amblyopia, which may indicate develop-ental abnormalities of visual cortex during the critical

rowth period. (Am J Ophthalmol 2007;143:89–493. © 2007 by Elsevier Inc. All rights reserved.)

ccepted for publication Nov 11, 2006.From the Department of Radiology, Peking University First Hospital,

eijing, China (J.X.X., S.X., J.T.Y., X.X.J.); Department of Pediatricphthalmology, Peking University First Hospital, Beijing, China

H.H.L.); and the National Laboratory of Pattern Recognition, Institutef Automation, Chinese Academy of Sciences, Beijing, China (X.L.G.,.L.G.).Inquiries to Sheng Xie, MD, Radiology Department, Peking University

(irst Hospital, 8 Xishiku Street, Xicheng District, Beijing 100034 China;-mail: ouyangxu2004@vip.sina.com

© 2007 BY ELSEVIER INC. A002-9394/07/$32.00oi:10.1016/j.ajo.2006.11.039

HE NEURAL BASIS OF HUMAN AMBLYOPIA REPRE-

sents the role of early experience on the structureand function of the human brain. Neurophysiologic

tudies have provided a great deal of evidence regardinghe functional effects of vision deprivation on the visualortex.1–3 Primary visual cortex was considered as therincipal site of vision deficit, whereas some extrastriateortex was also found to be responsible for abnormalities ofpecial visual function.4–6 However, neuroanatomic inves-igations for these abnormalities remain limited in humanubjects, especially in children with amblyopia. Untilecently, a voxel-based morphometry (VBM) study thatocused on the cortical change was addressed, indicatinghat adults and children with amblyopia have decreasedray matter volume in visual cortical regions.7 VBM is anpproach that gives a even-handed and comprehensivessessment of anatomic differences throughout the wholerain. It involves a voxel-wise comparison of the localoncentration of gray matter between two groups ofubjects.8 Because it is an automated analysis technique,BM has been widely used to identify abnormal anatomy

n some neurodegenerative diseases.9 However, one caveats that the VBM tools use an adult brain template to reachsolution during spatial normalization, so this could be a

ource of error for comparing measurements in children. Inhe VBM study of children with amblyopia, no subjectounger than 7 years of age was included, and the meange of the subjects was approximately 10 years.7 However,hey were relatively older in comparison to the criticaleriod of the visual system, which is around the onset agef stereopsis. In this study, we tried to use VBM tonvestigate the morphologic change of gray matter inoung children with amblyopia and to identify the devel-pmental abnormalities related to amblyopia.

METHODS

SUBJECTS: Fourteen pediatric patients with amblyopia

six males, eight females; age range, 4 to 8 years; mean age,

LL RIGHTS RESERVED. 489

5frotcfscovw(kdsaTufwbosng

●

i3Tpr(

[[�wwBrndi

●

ssnascfnpmsigdmuvpgTd

motio

4

.8 years) and 14 normally sighted children (10 males, fouremales; age range, 3.5 to 9 years; mean age, 5.8 years) wereecruited in this study. Informed consent for participation wasbtained from every subject’s parents, and the ethics commit-ee at Beijing University approved all study protocols. Thehildren with amblyopia were recruited by physician referralrom the pediatric ophthalmology service at Peking Univer-ity First Hospital. First, the children with vision problemsompleted an ophthalmologic exam that included tests ofcular motility, dilation, fundus exam, autorefraction, andisual-evoked potentials. After the diagnosis of amblyopiaas confirmed, they were referred to magnetic resonance

MR) scanning prior to amblyopia treatment. Patients with anown organic brain disorder or with specific clinical evi-ence of neurologic dysfunction were excluded from thiseries. Five children had strabismus amblyopia and nine hadnisometropic amblyopia. Their results were summarized inable 1. Controls were recruited from the children whonderwent magnetic resonance imaging (MRI) examinationor other purposes unrelated to vision problems. Four of themere volunteers, five of them were referred to MR scanningecause of headache, two children were scanned for trauma,ne child was scanned for dwarfism, and the other two werecanned for febrile convulsions. They were confirmed to haveormal visual acuity and no neurologic conditions. The tworoups were matched for age.

MAGNETIC RESONANCE DATA ACQUISITION: All MRmaging was performed on a 3.0-Tesla MR scanner (GE signa.0T HD, Milwaukee, Wisconsin, USA). Conventional axial2-weighted images were obtained previously to rule out theresence of any detectable brain lesions. After that, high-esolution 3-dimensional Spoiled Gradient Recalled Echo

TABLE 1. Informatio

Subject Amblyopia

Distance Acui

(logMAR)

Left Rig

Child 1 Anisometropia 0.6 1.

Child 2 Strabismic 1.2 0.

Child 3 Strabismic 0.1 0.

Child 4 Strabismic 0.2 1.

Child 5 Anisometropia 0.6 1.

Child 6 Anisometropia 0.4 0.

Child 7 Anisometropia 1.0 0.

Child 8 Anisometropia 0.7 0.

Child 9 Strabismic 0.4 1.

Child 10 Anisometropia 0.4 0.

Child 11 Anisometropia 1.0 0.

Child 12* Anisometropia 0.6 0.

Child 13 Strabismic 0.3 1.

Child 14 Anisometropia 0.1 1.

*This child’s data were excluded because of

SPGR) images (Time repetition [TR] � 7.8 ms, Time echo t

AMERICAN JOURNAL OF90

TE] � 3.2 ms, Time inversion [TI] � 450 ms, Field of viewFOV] � 22 � 22 cm2, matrix � 256 � 256, slice thickness

1.6 mm, Number of exitation [NEX] � 1) covering thehole brain were obtained. Head motion was minimizedith restraining foam pads offered by the manufacturer.efore the data processing, MR images were reviewed by a

adiologist to conform that all data sets were uncontami-ated by head motion artifacts. Consequently, the MRata of one child with anisometropic amblyopia (child 12n Table 1) were excluded.

DATA PROCESSING: For gray matter analysis, 3-dimen-ional SPGR images were processed with the method oftandard VBM using SPM2 (Welcome Department of Cog-itive Neurology, London, United Kingdom). Because thedult brain template provided by SPM2 is different from ourubjects in the aspects of brain configuration and imageontrast, we built our own template based on the data setsrom all subjects for spatial normalization. An affine and aonlinear spatial normalization of the raw MRI sets wereerformed, and they were segmented to extract the grayatter tissue. The extracted gray matter �set was then

moothed with a 12-mm full width at half-maximum Gauss-an filter to reduce confounding by individual variation inyral anatomy and to increase the signal-to-noise ratio. Theata (without normalization for the total amount of grayatter) of the two groups of subjects were then compared

sing the Student t test. The threshold used for selection ofoxels was 40% of the grand mean value. In addition, theroportional scaling routine was used, which normalized theray matter sets across subjects to the overall grand mean.his ancillary analysis was meant to assess the relativeistribution of gray matter reduction in amblyopes compared

Amblyopic Subjects

Refractive Error

Eye Deviation, ArcLeft Right

�3.50 �2.00 Orthotropia 0

�1.50 �1.50 Esotropic �15

�9.00 �7.50 Esotropic �25

�6.50 �2.0 Exotropic �40

�1.00 0 Orthotropia 0

�5.00 �4.25 Orthotropia 0

�6.00 �7.50 Orthotropia 0

�2.50 �2.25 Orthotropia 0

�3.50 �2.75 Esotropic �20

�0.50 �0.50 Orthotropia 0

�1.50 �5.00 Orthotropia 0

�0.50 �0.50 Orthotropia 0

�5.25 �1.25 Esotropic �15

�5.00 �1.00 Orthotropia 0

n artifact contamination.

n of

ty

ht

0

1

5

0

0

5

7

7

0

4

1

6

0

0

o controls. Implicit masking was used to ignore zeros, and

OPHTHALMOLOGY MARCH 2007

gVd(S

T

t

ocea

comnghTct(

N

pthttgstw

Ffwpg

Fsc

V

lobal calculation was based on the mean voxel value.10

oxel-by-voxel between-group comparisons of gray matterensity were investigated at a statistical threshold of T �3.42P � .001, uncorrected for multiple comparisons) using thetudent t test.

RESULTS

HERE WAS NO STATISTICAL DIFFERENCE IN AGE BETWEEN

IGURE 1. Voxel-based morphometry (VBM) revealed theoci of reduced gray matter volume in visual cortex of childrenith amblyopia relative to normal subjects. Left parahippocam-al gyrus and fusiform gyrus (arrowhead), left middle frontalyrus (black arrow), left inferior temporal gyrus (red arrow).

TABLE 2. Regions of Significantly Reduced Gray MatterAge-Matched Controls Revea

Brain Region Number of Voxels

Bilateral calcarine cortices 14

Left parahippocampal gyrus and fusiform gyrus 36

Left inferior temporal gyrus 12

Left middle frontal gyrus 16

Children with amblyopia, n � 13; age-matched controls, n � 14.

he two groups (5.8 years � 1.4 for children with ambly- s

VOXEL-BASED MORPHOMOL. 143, NO. 3

pia, 5.8 years � 1.9 for controls). The amblyopic grouponsisted of five children with strabismus amblyopia andight children with anisometropic amblyopia (clinical datare shown in Table 1).

The statistical parametric map showed several voxellusters of reductions in gray matter volume in the ambly-pia group relative to controls. Areas of significant grayatter reductions were seen in the following distincteocortical regions: middle frontal gyrus, parahippocampalyrus, fusiform gyrus, inferior temporal gyrus of the leftemisphere, and the bilateral calcarine cortices (Table 2).hey were all significant at the uncorrected P � .001luster level. After correction for multiple comparisons,hese clusters remained significant at the level of P � .05Figures 1 and 2).

DISCUSSION

EUROPHYSIOLOGIC STUDIES IN THESE YEARS HAVE

roved that primary visual cortex is the principal site forhe visual loss, yet there are definitely additional deficits atigher levels of the visual pathways. As an automatic wayo evaluate differences in brain morphology between pa-ient groups, VBM provides voxel-wise estimates of inter-roup differences in gray matter volume in a standardizedpace.8 With this method, we detected significant reduc-ion of gray matter density in regions of visual cortex,hich is consistent with previous studies.7,11 Histologic

IGURE 2. Voxel-based morphometry (VBM) revealed theignificant cluster of reduced gray matter volume in bilateralalcarine cortices of children with amblyopia (crossed line).

me in the Brains of Children With Amblyopia Relative toy Voxel-Based Morphometry

P value of Cluster Level (Corrected) Coordinates X, Y, Z Peak Z-Score

0.042 -1, -37, 8 3.24

0.035 -25, -2, -24 3.80

0.027 -60, -25, -24 3.31

0.029 -45, 33, 23 3.68

Voluled b

tudies conducted on animal models have revealed the

ETRY OF AMBLYOPIA 491

esetdvr

cavpavscdncvamcv

gcAavtatmartirptap

molimVitobes

alrtghot

atfccoytrsos

wicogVds

Tn(trX

4

ffects of visual deprivation on the ocular dominancetripes in layer IVc of the striated cortex.12 Yet directvidence from human amblyopia concerning neuroana-omic changes remains unavailable. By using VBM, ourata revealed the morphologic changes of the primaryisual cortex in children with amblyopia, which might beelated to the underlying histologic changes.

The fusiform gyrus and inferior temporal gyrus mayorrespond to the visual areas V4, which have beenssociated with analysis of form and color.13 The higherisual functions, such as shape discrimination, patternerception and feature detection, were proved to bebnormal in patients with amblyopia.14,15 However, theseisual deficits cannot be simply attributed to neural under-ampling or disarray in primary visual cortex; higher visualortical areas are likely to be involved in the neuralysfunction. A recent study of functional magnetic reso-ance imaging (fMRI) showed that the ventral temporalortex is very poorly activated when adults with amblyopiaiew face stimuli with their weak eye, despite normalctivation in other conditions.16 The prominent grayatter loss in the ventral temporal cortex of amblyopic

hildren may provide neuroanatomic evidence of ventralisual pathway impairment.An enigmatic finding in our results is the reduction of

ray matter density in the middle frontal gyrus, which isonsistent with the location of frontal eye fields (FEFs).mong the visual output of primary visual cortex, the FEFs

re also reached. The FEFs triggers intentional saccades toisual targets in the environment, to target locations, or tohe location where it is predicted that the target willppear.17 The abnormality of FEFs seemed to be related tohe presence of strabismic amblyopia; however, the grayatter volume in the FEFs was found to be reduced in our

mblyopic group, whereas it was increased in the previouseport of Chan and associates.11 The investigators arguedhat the greater gray matter volume in FEFs was compat-ble with a hypothesis of plasticity in the oculomotoregions to compensate for the cortical deficits in the visualrocessing areas. Perhaps the age difference between thewo studies induced the emergence of significant discrep-ncy. It will be interesting to further investigate therocess of neuroplasticity with respect to age.It is perhaps of note that our significant results wereostly gathered in the left hemisphere, except for regions

f calcarine cortex. The cluster of calcarine cortex wasocated in the middle of two hemispheres and extendednto both sides, so we believed that it represented abnor-alities of bilateral calcarine cortices. In the previousBM study, the gray matter loss of various visual cortices

n the amblyopes was almost symmetrical.7 It is possiblehat the small sample in our study hindered the detectionf abnormalities in the right hemisphere. Another possi-ility is that laterality exists in the development ofxtrastriate regions of visual cortex. Data from a previous

tudy revealed that age-related differences in sylvian fissure

AMERICAN JOURNAL OF92

symmetry were significant.18 Age-related increases inocal gray matter proportion bilaterally in the temporopa-ietal cortices are anatomically and temporally related tohe sulcal asymmetries. These extrastriate regions may alsorow faster in the left hemisphere than in the rightemisphere, so the anatomic difference between ambly-pic children and normal children was more prominent inhe left side.

Although voxel-based analysis can automatically revealbnormalities of the entire brain, it is not without limita-ions and problems.8 The templates provided by SPM arerom normal adults and are potentially less suitable forhildren. For this reason, we built our own template toonduct spatial normalization. By using this template, webtained reasonable results in our study, while our subjects’oung age seemed to cause no major problem of misregis-ration. The extension of clusters in our results waselatively local in comparison to that of the previoustudy,7 however. It may be attributed to the low sensitivityf our method, or the young age of our groups, or the mildeverity of amblyopia.

In conclusion, our results show gray matter reductionithin visual cortices in the brains of amblyopic children,

ncluding primary visual cortex and some extrastriateortices. These findings may indicate developmental changesf visual cortex during the critical period, as well as theenesis of visual dysfunction in amblyopic children. TheBM method can be reliably used in the analysis of MRIata sets of children, and it will be helpful in providingupplemental information to clinical evaluation.

HE AUTHORS INDICATE NO FINANCIAL SUPPORT OR FI-ancial conflict of interest. Involved in design and conduct of studyS.X.); data collection (S.X., J.T.X.); collection, analysis and interpreta-ion and preparation of manuscript (S.X., J.X.X.); and involved in theeviewing and approval of manuscript (J.X.X., S.X., J.T.X., H.H.L.,.L.G., G.L.G., X.X.J.).

REFERENCES

1. Garey LJ, de Courten C. Structural development of thelateral geniculate nucleus and visual cortex in monkey andman. Behav Brain Res 1983;10:3–13.

2. Von Noorden GK. Amblyopia: a multidisciplinary approach.Proctor lecture. Invest Ophthalmol Vis Sci 1985;26:1704–1716.

3. Goodyear BG, Nicolle DA, Humphrey GK, Menon RS.BOLD fMRI response of early visual areas to perceivedcontrast in human amblyopia. J Neurophysiol 2000;84:1907–1913.

4. Barnes GR, Hess RF, Dumoulin SO, Achtman RL, Pike GB.The cortical deficit in humans with strabismus amblyopia.J Physiol 2001;533:281–297.

5. Anderson SJ, Swettenham JB. Neuroimaging in human

amblyopia. Strabismus 2006;14:21–35.

OPHTHALMOLOGY MARCH 2007

1

1

1

1

1

1

1

1

1

V

6. Lee KM, Lee SH, Kim NY, et al. Binocularity and spatialfrequency dependence of calcarine activation in two types ofamblyopia. Neurosci Res 2001;40:147–153.

7. Janine D. Mendola, Ian P. Conner, Anjali Roy, et al.Voxel-based analysis of MRI detects abnormal visual cortexin children and adult with amblyopia. Hum Brain Mapp2005;25:222–236.

8. Ashburner J, Friston KJ. Voxel-based morphometry—themethods. Neuroimage 2000;11:805–821.

9. Ludolph AG, Juengling FD, Libal G, Ludolph AC, FegertJM, Kassubek J. Grey-matter abnormalities in boys withTourette syndrome: magnetic resonance imaging study usingoptimized voxel-based morphometry. Br J Psychiatry 2006;188:484–485.

0. Juottonen K, Laakso MP, Partanen K, Soininen H. Compar-ative MR analysis of the entorhinal cortex and hippocampusin diagnosing Alzheimer disease. Am J Neuroradiol 1999;20:139–144.

1. Chan ST, Tang KW, Lam KC, Chan LK, Mendola JD,Kwong KK. Neuroanatomy of adult strabismus: a voxel basedmorphometric analysis of magnetic resonance structural

scans. Neuroimage 2004;22:986–994.

VOXEL-BASED MORPHOMOL. 143, NO. 3

2. Swindale NV, Vital-Durand F, Blakemore C. Recovery frommonocular deprivation in the monkey. III. Reversal ofanatomical effects in the visual cortex. Proc R Soc Lond BBiol Sci 1981;213:435–450.

3. Schoenfeld MA, Noesselt T, Poggel D, et al. Analysis ofpathways mediating preserved vision after striate cortexlesions. Ann Neurol 2002;52:814–824.

4. Hess RF, Pointer JS, Simmers A, Bex P. Border distinctnessin amblyopia. Vision Res 2003;43:2255–2264.

5. Kovacs I, Polat U, Pennefather PM, Chandna A, NorciaAM. A new test of contour integration deficits in patientswith a history of disrupted binocular experience during visualdevelopment. Vision Res 2000;40:1775–1783.

6. Lerner Y, Pianka P, Azmon B, et al. Area-specific amblyopiceffects in human occipitotemporal object representations.Neuron 2003;40:1023–1029.

7. Xiao Q, Barborica A, Ferrera VP. Radial motion bias inmacaque frontal eye field. Vis Neurosci 2006;23:49–60.

8. Sowell ER, Thompson PM, Rex D, et al. Mapping sulcalpattern asymmetry and local cortical surface gray matterdistribution in vivo: maturation in perisylvian cortices.

Cereb Cortex 2002;12:17–26.

ETRY OF AMBLYOPIA 493