RESEARCHARTICLE
QuantificationofAbdominalFatinObeseand
HealthyAdolescentsUsing3TeslaMagnetic
ResonanceImagingandFreeSoftwarefor
ImageAnalysis
JulianaCristinaEloi
1
,MatiasEpifanio
1
,MarõÂliaMaiadeGoncËalves
2
,AugustoPellicioli
2
,
PatriciaFroelichGioraVieira
2
,HenriqueBregolinDias
2
,NeideBruscato
3
,Ricardo
BernardiSoder
4
,João CarlosBatistaSantana
5
,MarialenaMouzaki
6
,
MatteoBaldisserotto
7
*
1PediatricGastroenterologist,PediatricGastroenterologyService,HospitalSãoLucasdaPontifõÂcia
UniversidadeCatoÂlicadoRioGrandedoSul(PUCRS),PortoAlegre,RioGrandedoSul,Brazil,2PUCRS,
PortoAlegre,RioGrandedoSul,Brazil,3VeranoÂpolis,RioGrandedoSul,Brazil,4BrainInstitute(InsCer),
PUCRS,PortoAlegre,RioGrandedoSul,Brazil,5SchoolofMedicine,UniversidadeFederaldoRioGrande
doSul(UFRGS),PortoAlegre,RioGrandedoSul,Brazil,6DepartmentofPediatrics,Divisionof
Gastroenterology,HepatologyandNutrition,TheHospitalforSickChildren,Toronto,Canada,7Imaging
CenterCoordinator,BrainInstitute(InsCer),PUCRS,PortoAlegre,RioGrandedoSul,Brazil
*matteob@terra.com.br
Abstract
BackgroundandAims
Computedtomography,whichusesionizingradiationandexpensivesoftwarepackagesfor
analysisofscans,canbeusedtoquantifyabdominalfat.Theobjectiveofthisstudyisto
measureabdominalfatwith3TMRIusingfreesoftwareforimageanalysisandtocorrelate
thesefindingswithanthropometricandlaboratoryparametersinadolescents.
Methods
Thisprospectiveobservationalstudyincluded24overweight/obeseand33healthyado-
lescents(meanage16.55years).AllparticipantsunderwentabdominalMRIexams.Vis-
ceralandsubcutaneousfatareaandpercentagewerecorrelatedwithanthropometric
parameters,lipidprofile,glucosemetabolism,andinsulinresistance.Student'sttestand
Mann-Whitney'stestwasapplied.Pearson'schi-squaretestwasusedtocomparepropor-
tions.TodetermineassociationsPearson'slinearcorrelationorSpearman'scorrelation
wereused.
Results
Inbothgroups,waistcircumference(WC)wasassociatedwithvisceralfatarea(P=0.001
andP=0.01respectively),andtriglycerideswereassociatedwithfatpercentage(P=0.046
andP=0.071respectively).Inobeseindividuals,totalcholesterol/HDLratiowasassociated 
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	1/12
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111  
OPENACCESS
Citation:EloiJC,EpifanioM,deGoncËalvesMM,
PellicioliA,VieiraPFG,DiasHB,etal.(2017)
QuantificationofAbdominalFatinObeseand
HealthyAdolescentsUsing3TeslaMagnetic
ResonanceImagingandFreeSoftwareforImage
Analysis.PLoSONE12(1):e0167625.doi:10.1371/
journal.pone.0167625
Editor:RaffaellaBuzzetti,UniversitadegliStudidi
RomaLaSapienza,ITALY
Received:April22,2016
Accepted:November17,2016
Published:January27,2017
Copyright:©2017Eloietal.Thisisanopenaccess
articledistributedunderthetermsoftheCreative
CommonsAttributionLicense, whichpermits
unrestricteduse,distribution,andreproductionin
anymedium,providedtheoriginalauthorand
sourcearecredited.
DataAvailabilityStatement:Allrelevantdataare
withinthepaperanditsSupportingInformation
files.
Funding:Thecurrentstudywasfundedby
ConselhoNacionaldeDesenvolvimentoCientõÂficoe
TecnoloÂgico(CNPq)grant,resultedofNotice
Universal14/2012withgrantequivalenttoUSD:
13,000.
CompetingInterests:Theauthorshavedeclared
thatnocompetinginterestsexist.

withvisceralfatarea(P=0.03)andpercentage(P=0.09),andinsulinandHOMA-IRwere
associatedwithvisceralfatarea(P=0.001)andpercentage(P=0.005).
Conclusions
3TMRIcanprovidereliableandgoodqualityimagesforquantificationofvisceralandsubcu-
taneousfatbyusingafreesoftwarepackage.TheresultsdemonstratethatWCisagood
predictorofvisceralfatinobeseadolescentsandvisceralfatareaisassociatedwithtotal
cholesterol/HDLratio,insulinandHOMA-IR.
Introduction
Theprevalenceofchildhoodobesityhasbecomeamajorpublichealthissuearoundtheworld
[1,2]Obesityisoftenassociatedwithmetabolicsyndrome,whichconfersanincreasedriskof
cardiovasculareventsinadulthood[3±5].Previousstudieshaveshownthatcentralobesity,an
indicatorofvisceraladiposity,iscorrelatedwithallthecomponentsofmetabolicsyndrome,
namelyinsulinresistance,dyslipidemia,andhypertension[6,7].
WC(waistcircumference)isagoodpredictorofabdominaladiposity;however,itdoesnot
allowforquantificationofadiposetissuenorcanitdistinguishbetweenvisceralandsubcuta-
neousfat.Theaccurateassessmentofvisceralfatisofutmostclinicalimportance,givenits
associationwithmetabolicsyndromecomponents,whichinturn,contributetoincreased
morbidityandmortality.Severaltechniquesareavailabletomeasurecentraladiposity[8±10].
Whilecomputedtomography(CT)isthemostcommonlyusedimagingmodalitytomeasure
abdominalfat,Magneticresonanceimaging(MRI)hasasimilaraccuracy[11].Anadvantage
ofMRIistheabsenceofexposuretoionizingradiation,alimitationthatrestrictstheuseofCT
inchildrenandadolescents.Inaddition,theMRIapproachtoquantifyingabdominaladipos-
ityisefficient,allowingforimageaqcuisitionwithin5minutes.
OneaspectthathaspreventedtheuseofbothCTandMRIistheneedforexpensiveimage
analysissoftwareinthequantificationofabdominalfat[12,13]. However,Irvingetal[14]have
shownthatafreesoftware,NIHImageJ,canreliablymeasureadiposetissue.Eventhoughthat
studywasfocusedonCT,tonecanexpectsimilarresultsfromtheanalysisofmagneticreso-
nanceimagesobtainedusingthesamesoftware.
RegardingMRIstudiesofabdominalfat,mosthaveemployedequipmentwithfield
strengthof1.5Tesla(T)[11,12,15]. StudiesusingMRI3Tinadultshaveshowngoodaccuracy
forthequantificationofabdominalfat[11].However,itisimportanttodeterminewhether3T
MRIhasgoodperformanceforevaluationofabdominalfatofobeseadolescents.
Thus,theaimofthisstudywastoquantifysubcutaneousandvisceralabdominalfatofado-
lescentsusing3TMRIwiththefreesoftwareandtocorrelatethesefindingswithanthropomet-
ricvariablesandlaboratoryparametersthatarereflectiveofmetabolicdysregulation.
MaterialsandMethods
DesignandParticipants
Thisprospective,cross-sectionalstudywascarriedoutbetweenMarch2013andOctober2014
andincluded57Caucasianadolescentsaged16±18years.Thestudywasapprovedthe
ResearchEthicsCommitteeatouruniversityhospital.Samplesizewascalculatedas23partici-
pantsineachgroupusingthePEPI4.0softwareforasignificancelevelof5%,powerof90%, 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	2/12

andaminimumofcorrelationcoefficient0.5inBMIassociationwithvisceralfatasKellyetal
study.
Exclusioncriteriawerechronicdiseases,hepatorenaldisease,anduseofhepatotoxicdrugs,
corticosteroids,orimmunesuppressantsthatcouldpromotefatstorageintheliver.
Theadolescentswererandomlyselectedfromacohortparticipatinginapopulationstudy
assessinglongevity,whichwasperformedincomputersystem.Aninformedconsentformwas
signedbyallparticipantsorbytheirguardians.Includedsubjectsweresubsequentlydivided
intotwogroups:GroupAincluded33healthyparticipantsandGroupBincluded24over-
weight/obeseparticipants(2overweight,22obese).Allunderwentanthropometricmeasure-
ments,laboratorytesting,andabdominalMRImeasurements.
AnthropometricData
ParticipantswereclassifiedaccordingtoBMIforageashealthy(Zscore-2and<1),over-
weight(Zscore1and<2),orobese(Zscore2),followingWorldHealthOrganization
(WHO)guidelines[16].Participantswereweighedwearinglightclothesandnoshoes,inan
uprightposition,usingananthropometercoupledtoascale(Filizola
1
)certifiedbythe
NationalInstituteforMetrology,Quality,andTechnology(INMETRO).Bodymassindex
(BMI)wascalculatedalongwithZscoresandpercentiles,usingWHOsoftwareAnthroPlus
(http://www.who.int/growthref/tools/en/).
WCwasmeasuredusingnon-stretchableplastictapeatthemidpointbetweentheiliaccrest
andthelowestrib.Waisttoheightratio(WHR)wascalculated,witharatioof0.5usedascut-
pointtoindicatecardiovascularrisk[17,18]. BodysurfaceareawascalculatedusingtheDuBois
method.[19]Tannerpubertalstagewasdeterminedaswell[20,21].
ArterialPressure
Bloodpressurewasmeasuredonthedayofanthropometricassessment.Twomeasurements
wereperformedwiththesubjectsinthesittingpositionafter1and5-minuterestperiodsfol-
lowingtheirarrivalatthemedicaloffice.AnINMETRO-certifiedaneroidsphygmomanometer
wasused.Maximumsystolicanddiastolicpressureswererecordedandcategorizedaccording
tointernationalguidelinesforage,sex,andheightasnormal,upperlimitofnormal,and
hypertension[22].
LaboratoryTests
Laboratorytestswereperformedfollowinga12-hourfastonthesamedayofMRIexamina-
tions.Lipidprofilewasdeterminedbasedontotalcholesterol,low-densitylipoprotein(LDL)
andhigh-densitylipoprotein(HDL)cholesterol,andtriglyceridelevelsusingacolorimetric
enzymaticmethod(Mindray-BS380ChemistryAnalyzer).Fastingglucosewasdetermined
usingaglucose-oxidaseenzymaticmethodandaMindray-BS380ChemistryAnalyzer.Fasting
insulinwasdeterminedbychemiluminescence.Insulinresistance(IR)wasquantifiedby
homeostasismodelassessment(HOMA)usingtheformula:HOMA-IR=fastinginsulin
(?UI/mL)xfastingglucose(mmol/L)/22.5.
AbdominalMagneticResonanceImaging
AllMRIexamswereperformedatBrainInstituteofPUCRS.ImageswereacquiredinaSigna
HDxt3.0TRMscanner(GeneralElectric,Milwaukee,USA)andaneight-elementphased
arrayabdominalcoil(8-channelcoil).Patientswereimagedinthesupinepositionandaxial
T1-weightedfast-spinechoimages(FOV440mm,matrix512x512,TR230,TE4.40,slice 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	3/12

thickness5.0mm,gap1.0mm,NEX1)wereobtained.Eachscanlastedapproximately5min-
utes.Accordingtopreviousstudies,a5mmthicknesssliceatthelevelofL3-L4discswas
selectedforthequantificationoffat,asitisthoughttorepresentthelimitoftheupperabdo-
menandisnotinfluencedbyliveroradiposetissuefromthebuttocks[23,24]. Theselected
imagewassavedin.TIFFformat.
ImagingAnalysis
TheTIFFimages(matrix512x512)wereanalyzedusingImageJsoftware(rsbweb.nih.gov/ij)
withautothresholdpluginwhichconvertsautomaticallygrayscalepixelsintobinaryimages,
basedinaglobalhistogram-derivedmethod.Blackpixelsrepresentadiposetissueandwhite
pixelstheremainingsofttissue(muscle,solidorgans,intestinalloops,andvessels)[25].Adi-
posetissuewassubsequentlycategorizedintovisceralandsubcutaneousfatthroughmanual
division,whichwasaccomplishedbydrawingalinefollowingtheabdominalwalltoseparate
intraandextraabdominalcompartments.Visceralandsubcutaneousfatareas(cm
2
)were
measuredseparately(Fig1)[26].
StatisticalAnalysis
Quantitativevariableswereexpressedasmeansandstandarddeviationormedianandinter-
quartilerange.Qualitativevariableswereexpressedasabsoluteandrelativefrequencies.
Student'sttestwasusedtocomparegroupmeansexceptinthepresenceofasymmetricdis-
tribution,inwhichcaseMann-Whitney'stestwasapplied.Pearson'schi-squaretestwasused
tocompareproportions.Todetermineassociationsbetweenvariablesineachgroup,Pearson's
linearcorrelation(symmetricdistribution)orSpearman'scorrelation(asymmetricdistribu-
tion)wereused.
Thelevelofsignificancewassetat5%(P0.05).AllanalyseswerecarriedoutinSPSSv.
21.0. 
Fig1.Magneticresonanceimageofobesemale(BMI32.59).A).jpgMRIimagefromL3-L4.B)Binary
ImageJimageformeasurementoffat:fatappearsinblack.Imageshowsintra-abdominal(visceral)and
subcutaneousfat.
doi:10.1371/journal.pone.0167625.g001 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	4/12

Results
GroupAincluded16girls(48.5%)and17boys(51.5%),vs.13girls(54.2%)and11boys
(45.8%)inGroupB(Table1).Meanagewas16.8±0.7 and16.3±0.7 inGroupsAandBrespec-
tively.Therewerenosignificantdifferencesbetweenthegroupsintermsofgenderandage
distribution.
WCwassignificantlyhigherinGroupB(96,4vs75,2cmP<0.001).WHRwasalsosignifi-
cantlyhigherinGroupB(P<0.001).WHRwashigherthan0.5inonlyoneparticipantin
GroupA(3%),vs.22(91.7%)participantsinGroupB.BMIandbodysurfaceareaweresignifi-
cantlyhigherinGroupB(P<0.001).AllstudyparticipantswereclassifiedasTannerstageV.
Lipidprofileandfastingglucoselevelsweresimilarbetweenthetwogroups.However,fast-
inginsulinandHOMA-IRwerehigheringroupBthanGroupA(P<0.001).Transaminases
andalkalinephosphataselevelswerenotdifferentbetweenthegroups.Elevatedtransaminase
levels(above22forgrilsand26forboys)wereseeninonlyonesubjectinGroupB(Table2).
Asexpected,visceralandsubcutaneousfatareaandpercentage,measuredbyMRI,were
significantlyhigherinGroupB(Table2).Femaleshadhighersubcutaneousfatpercentage
thanmalesinbothgroups;however,therewerenogenderdifferencesintermsofvisceralfat
percentage.Subcutaneousfatareawasalsohigheringirls.Totalabdominalareaandvisceral
fatareaweresignificantlyhigherinGroupAboysvs.girls.BoysandgirlsinGroupBdiffered
onlyinregardingsubcutaneousfatpercentage,whichwashigheringirls(P=0.006)(Table3).
InGroupB,bothWCandWHRcorrelatedwithsubcutaneousandvisceralfatarea
(P<0.001andP<0.01respectively).InGroupA,onlyWCcorrelatedwithvisceralfatarea
(P<0.01);WHRwascorrelatedwithsubcutaneousfatareainthisgroup(P<0.01).BMIdid
notcorrelatewithvisceralfat(areaorpercentage)ineithergroup.However,BMIcorrelated
withsubcutaneousfatareainGroupB(P<0.001)(Table4)andtotalabdominalareainboth
groups(P<0.01forGroupAandP<0.001forGroupB).
Total/HDLcholesterolratiowasassociatedwithvisceralfatareaandpercentageinGroup
B,butnotinGroupA.Similarly,fastinginsulinandHOMA-IRwereassociatedwithvisceral
fatareaandpercentageinGroupB.Inbothgroups,therewasanassociationbetween
Table1.Characteristicsofthesample.
Variable(mean±SDunlessindicated)	Healthy(n=33) Overweight/obese(n=24)	P
Age(years)	16.8±0.7	16.3±0.7	0.013*
Gender
a
0.877* *
Male	17(51.5)	11(45.8)
Female	16(48.5)	13(54.2)
Systolicpressure(mmHg)	116.9±10. 6	121.3±12 .8	0.170*
Diastolicpressure(mmHg)	78.7±11. 4	84.0±13. 1	0.105*
WHR	0.57±0.0 6	0.44±0. 04	<0.001*
WHR0.5
a
1(3.0)	22(91.7)	<0.001*
Waistcircumference(cm)	75.2±6.5	96.4±13. 1	<0.001*
BMI(Zscore)	-0.11±0.53	2.45±0. 54	<0.001*
Bodysurface(m
2
)	1.72±0.1 6	1.99±0. 21	<0.001*
BMI,bodymassindex;WHR,waisttoheightratio.
a
n(%).
*Student'sttest
**Pearson'schi-squaretest.
Healthy:Zscore-2and<1;Obese:Zscore2(groupincludestwooverweightparticipants,Zscore1and<2).
doi:10.1371/journal.pone.0167625.t001 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	5/12

increasedvisceralfatpercentageandelevatedtriglyceridelevels;however,thisreachedsignifi-
canceonlyinGroupB(P=0.046)(Table5).
Discussion
InthisstudywewereabletoshowthattheuseofMRI3.0Teslawiththefreesoftwarepackage
ImageJallowsforsimple,efficientandsemi-automaticquantificationofabdominal
Table2.LaboratoryandMRIfindingsinhealthyandobeseadolescents.
Variable(mean±SDunlessindicated)	Healthy(n=33) Overweight/obese(n=24)	P
Lipidpro®le
Totalcholesterol(mg/dL)	159.2±28. 0	153.5±31 .6	0.478*
HDLcholesterol(mg/dL)	57.9±10. 1	50.3±8.0	0.003*
Totalcholesterol/HDLratio	2.84±0.4 9	3.16±0. 52	0.022*
Triglycerides(mg/dL)	70.8±27. 4	80.0±44. 5	0.339*
Glycemicpro®le
Glucose(mg/dL)	80.5±6.1	81.3±7.8	0.653*
Insulin(?UI/mL)
a
5.1(4.5±7.0)	9.7(5.7±12.4)	<0.001* *
HOMA-IR
a
0.94(0.79±1.31)	1.73(1.03±2.16)	<0.001* *
Magneticresonanceimaging
Totalabdominalarea(cm
2
)	415±63.5	692±152	<0.001*
Visceralfat(cm
2
)
a
16.5(12.9±21)	57.8(40±84.5)	<0.001* *
Subcutaneousfat(cm
2
)
a
54.4(42±88)	250(174±347)	<0.001* *
%Abdominalfat(%)	20.9±9.0	44.6±9.5	<0.001*
%Visceralfat(%)	4.10±1.3 6	9.11±4. 05	<0.001*
%Subcutaneousfat(%)	16.8±8.5	35.4±8.3	<0.001*
HDL,high-densitylipoproteincholesterol;HOMA-IR,homeostaticmodelassessmentÐinsulinresistanceindex.
a
Median(P25±P75).
*Student'sttest;
**Mann-Whitney'stest.
Healthy:Zscore-2and<1;Obese:Zscore2(groupincludestwooverweightparticipants,Zscore1and<2).
doi:10.1371/journal.pone.0167625.t002
Table3.MRIfindingsinadolescentboysandgirls.
MRIvariables(mean±SDunlessindicated) Healthy(n=33) P Obese/overweight(n=24) P
Boys Girls	Boys	Girls
Totalabdominalarea(cm
2
)	450±55.8 379±50.3 0.001 741±166	650±131 0.148*
Visceralfatarea(cm
2
)
a
17(15±21) 15(12±19) 0.053 60(42±95) 56(38±83) 0.776**
Subcutaneousfatarea(cm
2
)
a
44(33±55) 80(54±109) 0.004 219(146±351) 254(182±360) 0.424**
%Abdominalfat	16.9±9.4 25.0±6.2 0.007 39.3±9.1	49.1±7.5 0.008*
%Visceralfat	4.10±1.4 4.07±1.3 0.952 8.6±3.4	9.6±4.6 0.545*
%Subcutaneousfat	12.8±8.6 20.9±6.1 0.004 30.7±7.3	39.5±6.9 0.006*
a
Median(25±75percentile).
*Student'sttest;
**Mann-Whitney'stest.
Healthy:Zscore-2and<1;Obese:Zscore2(groupincludestwooverweightparticipants,Zscore1and<2).
doi:10.1371/journal.pone.0167625.t003 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	6/12

subcutaneousandvisceralfatinacohortofleanandoverweightadolescents.Onceagainit
wasshownthatBMIdidnotcorrelatewithmeasuresofabdominaladiposity,whereasWCcor-
relatedwithbothvisceralandsubcutaneousfattissue.Theimportanceofquantifyingvisceral
fatparticularlyofoverweightandobesesubjectswasshownagaininthisstudy,asvisceraladi-
positycorrelatedwithmarkersofinsulinresistanceanddyslipidemia.
ToourknowledgetherehaveonlybeentwootherstudiesreportingontheuseofMRI3.0
Teslaintheevaluationofabdominaladiposetissue,bothinadults[11,15]. Klopfensteinetal.
comparedimagesobtainedbyMRI3.0TeslatoimagesobtainedusingCT,whichwasconsid-
eredthegoldstandard[11].ParticipantswereyoungadultswithameanBMIof37kg/m
2
.
ThisstudydemonstratedthatMRIprovidesaccuratemeasurementsofvisceralandsubcutane-
ousadiposetissue.Lietal.reportedsimilarresults[15].Inthepresentstudywewereableto
showthat3TMRIallowsclinicianstoobtaingoodqualityimagesinobeseadolescents.
Table4.CorrelationbetweenanthropometricdataandMRIfindings.
MRI	Healthy(n=33)	Overweight/obese(n=24)
WC BMI(Zscore)Bodysurface WHR WC BMI(Zscore)Bodysurface WHR
Totalabdominalarea(cm
2
) 0.474** 0.451** 0.712*** 0.133 0.907** * 0.875*** 0.791** * 0.862***
Visceralfat(cm
2
) 0.456** -0.067 0.268 0.294 0.426* 0.387 0.326 0.602**
Subcutaneousfat(cm
2
) 0.14 0.344 -0.211 0.455** 0.709** * 0.821*** 0.490* 0.849***
%Abdominalfat 0.145 0.173 -0.188 0.332 0.287 0.464*	0.031 0.457*
%Visceralfat(%) 0.234 -0.256 0.006 0.205 0.099 0.097 0.018 0.118
%Subcutaneousfat(%) 0.115 0.223 -0.198 0.316 0.282 0.486*	0.027 0.468*
BMI,bodymassindex;WC,waistcircumference;WHR,waisttoheightratio.
*P<0.05;
**P<0.01;
***P <0.001
doi:10.1371/journal.pone.0167625.t004
Table5.Associationbetweenmetabolicvariablesandvisceralfat.
Variable	Visceralfatarea(cm
2
) %Visceralfat(%)
Totalcholesterol/HDLratio
Healthy	r=-0.019;P=0.918 r=-0.004;P=0.981
Obese/overweight	r=0.586;P=0.003 r=0.522;P=0.009
Insulinlevels(?U/mL)
Healthy	r
s
=0.019;P=0.915 r
s
=0.051;P=0.780
Obese/overweight	r
s
=0.625;P=0.001 r
s
=0.553;P=0.005
HOMA
Healthy	r
s
=0.100;P=0.581 r
s
=0.065;P=0.720
Obese/overweight	r
s
=0.625;P=0.001 r
s
=0.556;P=0.005
Triglyceridelevels
Healthy	r
s
=0.054;P=0.767 r
s
=0.318;P=0.071
Obese/overweight	r
s
=0.264;P=0.213 r
s
=0.412;P=0.046
HOMA,homeostasismodelassessment;r,Pearsoncorrelationcoef®cient;r
s
,Spearmancorrelation
coef®cient.
Healthy:Zscore-2and<1;Obese:Zscore2(groupincludestwooverweightparticipants,Zscore1
and<2
doi:10.1371/journal.pone.0167625.t005 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	7/12

TheavailabilityoffreesoftwareImageJsuggeststhattheuseofthistechnologyisgeneraliz-
able.ImageJhasbeenpreviouslyshowntoprovidereliablemeasurementsofadiposetissue,
withsimilaraccuracyasSlice-O-Maticversion4.3software(Tomovision)[14].Inaddition,
ImageJfeaturesanªeraserºtoolthatallowsfordeletionofbowelcontents,whichcanother-
wiseintroduceanoverestimateoffatmeasurements[10,11].
Incomparisontootheranthropometricparameters,WCcorrelatedbestwithvisceraladi-
posityareainbothgroups.UsingMRI,Brambillaetal[27].previouslyshowedthatWCisa
goodpredictorofvisceraladiposity,whereasBMIpredictssubcutaneousadiposity.Inour
studyBMIdidnotcorrelatewithvisceralorsubcutaneousfat.Otherstudieshaveunderscored
thesuperiorityofWCtoBMIinreflectingvisceraladiposity[28±30]. Thelimitationsofusing
BMIinthisclinicalsettingarenumerous.Forexample,dependingonthedefinitionofobesity
used,theprevalenceofoverweightandobesityusingthesameBMIvaluescanvarywidely
[31].Furthermore,ithasbeenshownthatBMIfailstoidentifyexcessadiposityinoverone
quarterofchildren[32],whichinturnmeansthatcliniciansmayfailtoidentifytheneedto
screenpatientsatriskformetabolicdysregulation.Thesedata,alongwiththefactthatbecause
ofitsassociationwithvisceraladiposity,anelevatedWCisassociatedwithincreasedfuture
cardiovascularrisk,supporttheinclusionofWCmeasurementstotheroutinemedicalassess-
mentofadolescents.
WhileWCcorrelateswithmarkersofabdominaladiposity,itislimitedbythefactthatit
cannotdistinguishbetweenvisceralandsubcutaneousadiposetissue.Thisisakeydistinction
whendeterminingthecardiometabolicriskofpatients[33].Accesstoanefficientandcheap
imagingmodality,suchastheonedescribedinthisstudy,thatcandistinguishbetweenvisceral
andsubcutaneousfatcan,hence,becomplementarytothebaselineassessmentofpatientswho
maybefoundtohaveanelevatedWC.Theadditionalbenefitofthistechnologyisthataccu-
rateimagescanbeobtainedwithouttheriskofexposingchildrentoionizingradiation.Laslty,
evidenceofincreasingvisceraladipositycanbeusedasanadditionalclinicaltooltoconveyto
thefamiliestheneedtobecompliantwithlifestylechangesaimedatimprovingtheirbody
compositionandultimatelydecreasingthepatients'futurecardiometabolicrisk.
Wedidnotobserveelevationsinthefastingglucoseoftheadolescentsincludedinthis
study.However,fastinginsulin,HOMA-IRandtriglyceridesweresignificantlyhigherinover-
weightandobeseparticipants,suggestingthepresenceofinsulinresistance.Inaddition,
HOMA-IRwasstronglycorrelatedwithvisceraladiposityareainobeseparticipants.Thisfind-
ingisinagreementwithpreviouslypublishedstudiesthatshowacausativerelationship
betweenvisceraladiposityandinsulinresistance[34,35].
Inbothgroups,therewasatrendtowardsincreasedtriglyceridelevelsandincreasedvis-
ceralfatpercentage.Itshouldbenotedthatonlythreehealthy(9%)andfiveobese(20.8%)par-
ticipantshadtriglyceridesabove100mg/dL.WefoundastrongcorrelationoftotaltoHDL
cholesterolwithvisceralfatpercentage.Thisisinaccordancewithotherinvestigators,who
havealsoshownastrongassociationbetweencentralobesityanddyslipidemia[36,37].
LimitationsofthepresentstudyincludethefactthatonlyCaucasianadolescentsof
advancedpubertalstagewereincluded.Anotherlimitationisthatwedidnotassesstheaccu-
racyof3TMRIinmeasuringsubcutaneousandvisceralabdominalfatbutextrapolateddata
fromtheadultliteraturethatsuggeststhatthistechnologyisaccurate.Astudyassessingthe
accuracyofthisMRItechnologyinadolescentswouldhaverequiredexposuretounnecessary
radiation,asCTscansareconsideredthegoldstandardforthesetypesofmeasurements.
Inconclusion,weshowthat3TMRIcanprovidegoodqualityimagesusingafreesoftware
packagethatallowsfastandaccuratequantificationofvisceralandsubcutaneousfatinover-
weightandobeseadolescents.TheabdominalfatsegmentationresultsdemonstratethatWCis 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	8/12

agoodestimateofvisceralandsubcutaneousfatandthevisceralfatareaisassociatedwith
totalcholesterol/HDLcholesterol,insulinandHOMA-IR.
SupportingInformation
S1Data.
(XLSX)
S1Table.Characteristicsofthesample.BMI,bodymassindex;WHR,waisttoheightratio.
an(%).

Student'sttest

Pearson'schi-squaretest.Healthy:Zscore-2and<1;Obese:Z
score2(groupincludestwooverweightparticipants,Zscore1and<2).
(DOCX)
S2Table.LaboratoryandMRIfindingsinhealthyandobeseadolescents.HDL,high-den-
sitylipoproteincholesterol;HOMA-IR,homeostaticmodelassessmentÐinsulinresistance
index.aMedian(P25±P75).

Student'sttest;

Mann-Whitney'stest.Healthy:Zscore-2
and<1;Obese:Zscore2(groupincludestwooverweightparticipants,Zscore1
and<2).
(DOCX)
S3Table.S3Title:MRIfindingsinadolescentboysandgirls.aMedian(25±75percentile).

Student'sttest;

Mann-Whitney'stest.Healthy:Zscore-2and<1;Obese:Zscore2
(groupincludestwooverweightparticipants,Zscore1and<2).
(DOCX)
S4Table.CorrelationbetweenanthropometricdataandMRIfindings.BMI,bodymass
index;WC,waistcircumference;WHR,waisttoheightratio.

P<0.05;

P<0.01;

P<0.001.Healthy:Zscore-2and<1;Obese:Zscore2(groupincludestwoover-
weightparticipants,Zscore1and<2).
(DOCX)
S5Table.Associationbetweenmetabolicvariablesandvisceralfat.HOMA,homeostasis
modelassessment;r,Pearsoncorrelationcoefficient;rs,Spearmancorrelationcoefficient.
Healthy:Zscore-2and<1;Obese:Zscore2(groupincludestwooverweightpartici-
pants,Zscore1and<2.
(DOCX)
AuthorContributions
Conceptualization:MBJCEMEJCBSRBS.
Datacuration:JCEMMGAPPFGVHBDNBMM.
Formalanalysis:MBRBSMEJCE.
Fundingacquisition:MBJCE.
Investigation:JCEMMGAPPFGVHBDNBMM.
Methodology:MBJCEMEJCBSRBS.
Projectadministration:MBJCBSJCE.
Resources:MMGAPPFGVHBDNBMM.
Software:MBRBSJCE. 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	9/12

Supervision:MBJCBS.
Validation:MBRBSJCE.
Visualization:MMGAPPFGVHBDNBMM.
Writing±originaldraft:MBMEMMRBS.
Writing±review&editing:MBMEMMRBS.
References
1.OgdenCL,CarrollMD,KitBK,FlegalKM.PrevalenceofchildhoodandadultobesityintheUnited
States,2011±2012.JAMA2014;311:806±814.doi:10.1001/jama.2014.732PMID:24570244
2.deOnisM,BlossnerM,BorghiE.Globalprevalenceandtrendsofoverweightandobesityamongpre-
schoolchildren.AmJClinNutr2010;92:1257±1264.doi:10.3945/ajcn.2010.29786PMID:
20861173
3.BerensonGS,SrinivasanSR,BaoW,NewmanWP3rd,TracyRE,WattigneyWA.Association
betweenmultiplecardiovascularriskfactorsandatherosclerosisinchildrenandyoungadults.The
BogalusaHeartStudy.NEnglJMed1998;338:1650±1656.doi:10.1056/NEJM199806043382302
PMID:9614255
4.JuonalaM,MagnussenCG,BerensonGS,VennA,BurnsTL,SabinMA,etal.Childhoodadiposity,
adultadiposity,andcardiovascularriskfactors.NEnglJMed2011;365:1876±1885.doi:10.1056/
NEJMoa1010112PMID:22087679
5.TiroshA,ShaiI,AfekA,Dubnov-RazG,AyalonN,GordonB,etal.AdolescentBMItrajectoryandrisk
ofdiabetesversuscoronarydisease.NEnglJMed2011;364:1315±1325.doi:10.1056/
NEJMoa1006992PMID:21470009
6.Agredo-ZuÂñigaRA,Aguilar-dePlataC,SuaÂrez-OrtegoÂnMF.Waist:heightratio,waistcircumference
andmetabolicsyndromeabnormalitiesinColombianschooledadolescents:amultivariateanalysiscon-
sideringlocatedadiposity.BrJNutr.2015;114(5):700±5.doi:10.1017/S0007114515002275PMID:
26279413
7.GuntscheZ,GuntscheEM,SaravõÂFD,GonzalezLM,LopezAvellanedaC,AyubE,etal.Umbilical
waist-to-heightratioandtrunkfatmassindex(DXA)asmarkersofcentraladiposityandinsulinresis-
tanceinArgentineanchildrenwithafamilyhistoryofmetabolicsyndrome.JPediatrEndocrinolMetab.
2010;23(3):245±56.PMID:20480723
8.KullbergJ,BrandbergJ,AngelhedJE,FrimmelH,BergelinE,StridL,etal.Whole-bodyadiposetissue
analysis:comparisonofMRI,CTanddualenergyX-rayabsorptiometry.BrJRadiol2009;82:123±
130.doi:10.1259/bjr/80083156PMID:19168691
9.KarlssonAK,KullbergJ,StoklandE,AllvinK,GronowitzE,SvenssonPA,etal.Measurementsoftotal
andregionalbodycompositioninpreschoolchildren:AcomparisonofMRI,DXA,andanthropometric
data.Obesity(SilverSpring)2013;21:1018±1024.
10.Mook-KanamoriDO,HolzhauerS,HollesteinLM,DurmusB,ManniesingR,KoekM,etal.Abdominal
fatinchildrenmeasuredbyultrasoundandcomputedtomography.UltrasoundMedBiol2009;35:
1938±1946.doi:10.1016/j.ultrasmedbio.2009.07.002PMID:19800165
11.KlopfensteinBJ,KimMS,KriskyCM,SzumowskiJ,RooneyWD,PurnellJQ.Comparisonof3TMRI
andCTforthemeasurementofvisceralandsubcutaneousadiposetissueinhumans.BrJRadiol2012;
85:e826±830.doi:10.1259/bjr/57987644PMID:22514099
12.BenfieldLL,FoxKR,PetersDM,BlakeH,RogersI,GrantC,etal.Magneticresonanceimagingof
abdominaladiposityinalargecohortofBritishchildren.IntJObes(Lond)2008;32:91±99.
13.AbateN,BurnsD,PeshockRM,GargA,GrundySM.Estimationofadiposetissuemassbymagnetic
resonanceimaging:validationagainstdissectioninhumancadavers.JLipidRes1994;35:1490±
1496.PMID:7989873
14.IrvingBA,WeltmanJY,BrockDW,DavisCK,GaesserGa,WeltmanA.NIHImageJandSlice-O-Matic
computedtomographyimagingsoftwaretoquantifysofttissue.Obesity2007;15(2):370±6.doi:10.
1038/oby.2007.573PMID:17299110
15.LiX,YoungrenJF,HyunB,SakkasGK,MulliganK,MajumdarS,etal.Technicalevaluationofinvivo
abdominalfatandIMCLquantificationusingMRIandMRSIat3T.MagnResonImaging2008;26:
188±197.doi:10.1016/j.mri.2007.06.006PMID:17683890 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	10/12

16.deOnisM,OnyangoAW,BorghiE,SiyamA,NishidaC,SiekmannJ.DevelopmentofaWHOgrowth
referenceforschool-agedchildrenandadolescents.BullWorldHealthOrgan2007;85:660±667.doi:
10.2471/BLT.07.043497PMID:18026621
17.NambiarS,TrubyH,DaviesPS,BaxterK.Useofthewaist-heightratiotopredictmetabolicsyndrome
inobesechildrenandadolescents.JPaediatrChildHealth2013;49:E281±287.doi:10.1111/jpc.
12147PMID:23521181
18.SavvaSC,LamnisosD,KafatosAG.Predictingcardiometabolicrisk:waist-to-heightratioorBMI.A
meta-analysis.DiabetesMetabSyndrObes2013;6:403±419.doi:10.2147/DMSO.S34220PMID:
24179379
19.DuBoisD,DuBoisEF.Aformulatoestimatetheapproximatesurfaceareaifheightandweightbe
known.1916.Nutrition1989;5:303±311;discussion312±303.PMID:2520314
20.MarshallWA,TannerJM.Variationsinthepatternofpubertalchangesinboys.ArchDisChild1970;
45:13±23.PMID:5440182
21.MarshallWA,TannerJM.Variationsinpatternofpubertalchangesingirls.ArchDisChild1969;44:
291±303.PMID:5785179
22.Nationalhighbloodpressureeducationprogramworkinggrouponhighbloodpressureinchildrenand
adolescents.Thefourthreportonthediagnosis,evaluation,andtreatmentofhighbloodpressureinchil-
drenandadolescents.Pediatrics2004;114:555±576.PMID:15286277
23.SottierD,PetitJM,GuiuS,HamzaS,BenhamicheH,HillonP,etal.Quantificationofthevisceraland
subcutaneousfatbycomputedtomography:interobservercorrelationofasingleslicetechnique.Diagn
IntervImaging. 2013;94(9):879±84.doi:10.1016/j.diii.2013.04.006PMID:23725783
24.TongY,UdupaJK,TorigianDA.OptimizationofabdominalfatquantificationonCTimagingthrough
useofstandardizedanatomicspace:anovelapproach.MedPhys.2014;41(6):063501.doi:10.1118/
1.4876275PMID:24877839
25.SezginM&SankurB.SurveyoverImageThresholdingTechniquesandQuantitativePerformance
Evaluation.JournalofElectronicImaging13(1):146±165.
26.HuHH,NayakKS,GoranMI.AssessmentofAbdominalAdiposeTissueandOrganFatContentby
MagneticResonanceImaging.ObesRev.2011;12(501):e504±e515.
27.BrambillaP,BedogniG,MorenoLA,GoranMI,GutinB,FoxKR,etal.Crossvalidationofanthropometry
againstmagneticresonanceimagingfortheassessmentofvisceralandsubcutaneousadiposetissue
inchildren.IntJObes(Lond)2006;30:23±30.
28.SchroÈderH,RibasL,KoebnickC,FuntikovaA,GomezSF,FõÂtoM,etal.Prevalenceofabdominalobe-
sityinSpanishchildrenandadolescents.Doweneedwaistcircumferencemeasurementsinpediatric
practice?PLoSOne.2014;9(1):e87549.doi:10.1371/journal.pone.0087549PMID:24475305
29.GroÈber-GraÈtzD,WidhalmK,deZwaanM,ReinehrT,BluÈherS,SchwabKO,etal.Bodymassindexor
waistcircumference:whichisthebetterpredictorforhypertensionanddyslipidemiainoverweight/
obesechildrenandadolescents?Associationofcardiovascularriskrelatedtobodymassindexorwaist
circumference.HormResPaediatr. 2013;80(3):170±8.doi:10.1159/000354224PMID:24021483
30.SavvaSC,TornaritisM,SavvaME,KouridesY,PanagiA,SilikiotouN,etal.Waistcircumferenceand
waist-to-heightratioarebetterpredictorsofcardiovasculardiseaseriskfactorsinchildrenthanbody
massindex.IntJObesRelatMetabDisord. 2000;24(11):1453±8.PMID:11126342
31.Gonzalez-CasanovaI,SarmientoOL,GazmararianJA,CunninghamSA,MartorellR,PrattM,Stein
AD.Comparingthreebodymassindexclassificationsystemstoassessoverweightandobesityinchil-
drenandadolescents.RevPanamSaludPublica. 2013;33(5):349±55.PMID:23764666
32.JavedA,JumeanM,MuradMH,OkoroduduD,KumarS,SomersVK,etal.Diagnosticperformanceof
bodymassindextoidentifyobesityasdefinedbybodyadiposityinchildrenandadolescents:asystem-
aticreviewandmeta-analysis.PediatrObes. 2015;10(3):234±44.doi:10.1111/ijpo.242PMID:
24961794
33.ClementeG,ManciniM,GiaccoR,TornatoreA,RagucciM,RiccardiG.Visceraladiposityandsubclini-
calatherosclerosisinhealthyyoungmen.IntJFoodSciNutr.2015;66(4):466±70.doi:10.3109/
09637486.2015.1042845PMID:26017320
34.FraynKN.Visceralfatandinsulinresistance-causativeorcorrelative?BrJNutr.2000;83Suppl1:
S71±7.
35.KabirM,CatalanoKJ,AnanthnarayanS,KimSP,VanCittersGW,DeaMK,etal.Molecularevidence
supportingtheportaltheory:acausativelinkbetweenvisceraladiposityandhepaticinsulinresistance.
AmJPhysiolEndocrinolMetab. 2005;288(2):E454±61.doi:10.1152/ajpendo.00203.2004PMID:
15522994 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	11/12

36.CaprioS,HymanLD,McCarthyS,LangeR,BronsonM,TamborlaneWV.Fatdistributionandcardio-
vascularriskfactorsinobeseadolescentgirls:importanceoftheintraabdominalfatdepot.AmJClin
Nutr1996;64:12±17.PMID:8669407
37.DanielsSR,MorrisonJA,SprecherDL,KhouryP,KimballTR.Associationofbodyfatdistributionand
cardiovascularriskfactorsinchildrenandadolescents.Circulation1999;99:541±545.PMID:9927401 
3TMRItoMeasureAbdominalFatinAdolescents
PLOSONE|DOI:10.1371/journal.pone.0167625January27,2017	12/12