Table Of ContentRESEARCHARTICLE
A new approach for the analysis of facial
growth and age estimation: Iris ratio
CarlosEduardoPalharesMachado1,2,3*,MartaReginaPinheiroFlores4,La´ıseNascimento
CorreiaLima5,6,RachelLimaRibeiroTinoco7,8,AdemirFranco9,AnaCristina
BarretoBezerra2,MartinPaulEvison10,MarcoAure´lioGuimarães3
1 NationalInstituteofCriminalistics,BrazilianFederalPolice,MinistryofJustice,Bras´ılia,DistritoFederal,
Brazil,2 HealthScienceCollege,UniversityofBras´ılia,Bras´ılia,DistritoFederal,Brazil,3 MedicoLegal
Centre,DepartmentofPathologyandLegalMedicine,RibeirãoPretoMedicalSchool,UniversityofSão
Paulo,RibeirãoPreto,SãoPaulo,Brazil,4 SchoolofDentistry,UniversityofSãoPaulo,SãoPaulo,São
Paulo,Brazil,5 SchoolofDentistry,FederalUniversityofMaranhão,SãoLu´ıs,Maranhão,Brazil,
6 PiracicabaDentistrySchool,StateUniversityofCampinas,Piracicaba,SãoPaulo,Brazil,7 Schoolof
Dentistry,UniversitySalgadodeOliveira,Nitero´i,RiodeJaneiro,Brazil,8 DepartmentofAnthropology,
NationalMuseum,FederalUniversityofRiodeJaneiro,RiodeJaneiro,RiodeJaneiro,Brazil,9 Department
a1111111111
ofOralHealthSciences,KatholiekeUniversiteit,Leuven,Belgium,10 DepartmentofAppliedSciences,
a1111111111
CentreforForensicScience,NorthumbriaUniversity,Newcastle,UnitedKingdom
a1111111111
a1111111111 *[email protected]
a1111111111
Abstract
Thestudyoffacialgrowthisexploredinmanyfieldsofscience,includinganatomy,genetics,
OPENACCESS
andforensics.Inthefieldofforensics,itactsasavaluabletoolforcombatingchildpornogra-
Citation:MachadoCEP,FloresMRP,LimaLNC, phy.Thepresentresearchproposesanewmethod,basedonrelativemeasurementsand
TinocoRLR,FrancoA,BezerraACB,etal.(2017)A
fixedreferencesofthehumanface—specificallyconsideringmeasurementsofthediameter
newapproachfortheanalysisoffacialgrowthand
ageestimation:Irisratio.PLoSONE12(7): oftheiris(irisratio)—fortheanalysisoffacialgrowthinassociationwithageinchildrenand
e0180330.https://doi.org/10.1371/journal. sub-adults.Theexperimentalsampleconsistedofdigitalphotographsof1000Braziliansub-
pone.0180330
jects,agedbetween6and22years,distributedequallybysexanddividedintofivespecific
Editor:MohammedE.Elsalanty,Augusta agegroups(6,10,14,18,and22yearolds±onemonth).ThesoftwarepackageSAFF-2D®
University,UNITEDSTATES
(ForensicFacialAnalysisSystem,BrazilianFederalPolice,Brazil)wasusedforpositioning
Received:June9,2016 11landmarksontheimages.Tenmeasurementswerecalculatedandusedasfixedrefer-
Accepted:June14,2017 encestoevaluatethegrowthoftheothermeasurementsforeachagegroup,aswelltheaccu-
mulatedgrowth(6–22yearsold).TheIntraclassCorrelationCoefficient(ICC)wasappliedfor
Published:July7,2017
theevaluationofintra-examinerandinter-examinerreliabilitywithinaspecificsetofimages.
Copyright:©2017Machadoetal.Thisisanopen
Pearson’sCorrelationCoefficientwasusedtoassesstheassociationbetweeneachmea-
accessarticledistributedunderthetermsofthe
CreativeCommonsAttributionLicense,which surementtakenandtherespectiveagegroups.ANOVAandPost-hocTukeytestswereused
permitsunrestricteduse,distribution,and tosearchforstatisticaldifferencesbetweentheagegroups.Theoutcomesindicatedthat
reproductioninanymedium,providedtheoriginal
facialstructuresgrowwithdifferenttiminginchildrenandadolescents.Moreover,thegrowth
authorandsourcearecredited.
allometryexpressedinthisstudymaybeusedtounderstandwhatstructureshavemoreor
DataAvailabilityStatement:Allrelevantdataare
lessproportionalvariationinfunctionfortheagerangesstudied.Thediameteroftheiriswas
withinthepaperanditsSupportingInformation
foundtobethemoststablemeasurementcomparedtotheothersandrepresentedthebest
files.
cephalometricmeasurementasafixedreferenceforfacialgrowthratios(orindices).The
Funding:Theauthorsreceivednospecificfunding
methoddescribedshowspromisingpotentialforforensicapplications,especiallyaspartof
forthiswork.
thearmamentariumagainstcrimesinvolvingchildpornographyandchildabuse.
Competinginterests:Theauthorshavedeclared
thatnocompetinginterestsexist.
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 1/19
Irisratiofortheanalysisoffacialgrowth
Introduction
Thehumanfacedoesnotgrowhomogenouslyovertime.Eachofthemanyfacialstructures
developsindifferentdimensionsanddirections[1,2].Consequently,thefacialanatomy
reachesdifferentproportionsdependingonage[1,3,4].Thisphenomenon,knownasallome-
try,isthereasonwhyachild’sfacedoesnotcorrespondtoasmallerversionofanadult’sface.
Growthandalterationsincraniofacialmorphologyarousesinterestinmanyfieldsofscience,
especiallyphysicalanthropology[5–7]andgenetics[8].Inthesefields,morphologyismost
oftenstudiedinrelationtoevolutionaryprocess[9,10]andmedicaltherapeutics[11–15].
However,nowadays,alternativeapplicationsofcraniofacialmorphology,suchasinthescope
offorensicsciences,hasreceivedmajorattention.Inthisregard,forensicstudiesweredevel-
opedinthelastdecadetoinvestigatefacialgrowthasananthropometrictoolforageestima-
tionproceduresincasesinvolvingchildpornography[16,17].
Traditionalanthropometryisperformedbytakingmeasurementsdirectlyfromthesubjects
usingcalipersormeasuringtapes.Clearly,thisrequiresacontrolledenvironmentaswellas
theconsentandcooperationoftheexaminedsubject—whichcanbeachallengingtaskwhen
childrenareinvolved[18].Analysisoftwo-dimensional(2D)images,suchascephalograms
andphotographs,hasconsequentlyemergedasanalternativemethodforinvestigationsinthis
field.Amongthe2Dtechniquesusedforimageacquisitionandanalysisofthehumanface,
photo-anthropometryremainsapopularapproachforepidemiologicalandforensicstudies.It
consistsoflandmarkingphotographstoenablethemeasurementofdistances,angles,andpro-
portions[5].Thesameprincipleisalsoapplicableinthree-dimensional(3D)images[15,19].
Adirectcomparisonbetweenmeasurementstakenfromdifferentphotographsisonlyreli-
ableiftheimageswereacquiredstandardlyorwithmetricreferences.Ingeneral,photographs
couldhamperanthropometricanalyses.However,anthropometricexamsarefeasibleifthe
photographsaretakenfollowingthesameprotocol.Inthiscontext,ratiosandanglesbetween
facialdistancescouldbecalculatedtoallowananthropometricexam,evenintheabsenceof
metricreferencesintheimage[20].Yet,theratiosobtainedbetweentwolinearfacialdistances
culminateinindicesthatmayplayanimportpartinanthropologyastoolsfortheclassification
offacialtypes[21].
Ingeneral,understandingthegrowthandalterationsincraniofacialmorphologyrequires
long-termlongitudinalstudiesthatregisterhumandevelopmentfromchildhoodtoadulthood
withdirectorindirectmeasurementsoftheface.Usinganinnovativeapproach,Ferrarioetal.
[15]conductedamixedlongitudinalandcross-sectionalstudyonthequantificationofgrowth
alterationsincraniofacialmorphologybyemployinganoninvasive3Dassessmentthatused
camerasandinfra-redsensors.Differencesingrowthtimingandfacialproportionsbetween
malesandfemalescouldbedetectedfollowingthisapproach.Otherauthorshavesimilarly
foundotheranthropologicalapplicationsfortheassessmentofgrowthalterationsincraniofa-
cialmorphology,namelytheageestimationofchildreninvolvedinpornography[17].
Ononehand,allometryenablesageestimationinforensicsciences.Ontheotherhand,it
limitstheusefulnessoffacialindicesforthedetectionofdifferencesingrowthproportions.
Thisisjustifiedsincetheindicesarefoundedontwomeasurementsthataretakenfromfacial
structuresthatdevelopwithindifferenttimings.Inotherwords,theuseofindicestoassess
growthproportionsissimilartoestimatingsomeoneelse’scarspeedwhiledrivingaseparate
car.Inthiscontext,theoutcomeisnottherealcarspeed,butrathertherelativespeed
betweenbothcars.Forthehumanface,theuseofindicesresultsinrelativevariationsof
facialstructures[2].Despitetheirrelativenature,measurementstakenfromthefaceare
closertoreal-lifemeasurementsifmorestablefacialstructuresareconsideredasreferences
intheindexequation.
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 2/19
Irisratiofortheanalysisoffacialgrowth
Thepresentstudythereforeaimedtoanalyzetheallometryofhumangrowthusingasetof
10facialmeasurementstakenfromchildren,adolescents,andyoungadults.Thediameterof
theiriswasusedasreferencetoassessthecraniofacialvariationsintheremainingnine
measurements.
Materialandmethods
Samplingandfacialanalysis
Thesampleconsistedof1000photographs,takenstandardly,ofBraziliansubjectsaged
between6and22years.Thesamplewashomogeneouslydivided(n=200)intofiveagegroups
of6,10,14,18,and22yearolds,withastandarddeviationof1monthineachagegroup.The
photographswereselectedfromadatabaseoftheBrazilianFederalPolice.Forinclusioninthe
database,thephotographsweretakenfollowingtheInternationalCivilAviationOrganization
(ICAO)guidelinesforpassportsandwerestoredin.PNG24-bitformatwithresolutionof
640×480pixels.Inaddition,thesameflashsystemandcameramodelwereusedforallphoto-
graphs,andpositioned1.5metersfromthesubject’sface.Onlyphotographsofsubjectswith
neutralfacialexpression,closedlips,andheadpositionedstraighttowardsthecamerawere
selected.Photographsofsubjectswithheadrotationinthesagittal,axial,orcoronalaxeswere
excluded,aswellasthosewithfacialdeformationsorevidentasymmetries.Subjectswithfacial
hair,adornments,andmakeupwerealsoexcludedduetoincompletevisualizationoftheface.
Acephalometricanalysisofallphotographswasperformedbyasingleexaminer.Toassess
intra-examinerreliability,asetof100imageswereexaminedinduplicatebypositioning11
landmarksonpre-establishedreferencepoints(Table1)[22].Atthisstep,anon-commercial
softwarepackagedevelopedbytheBrazilianFederalPolicefor2Dfacialanalysis(SAFF-2D1–
ForensicFacialAnalysisSystem,DepartmentofFederalPolice,Brazil)wasused.Thesoftware
registersCartesiancoordinatesinthex-andy-axesforeachofthepositionedlandmarks.The
intra-andinter-examinerreliabilityofthelandmarkingmethodologywasalsoassessedprior
tothestudyandconsistedoftheanalysisof10photographsbythreetrainedexaminers,three
times,withinanintervalof15days.
Table1. Definitionofthecephalometriclandmarksusedinthisstudy[2,22].
Landmark Definition
1.Nasion(n) Theinterceptionofthemidsagittalplaneandthelinecrossingthesuperiorpalpebral
creases,abovetheuppereyelids.
2.Subnasale(sn) Thelowestpointofthenoseonthemidsagittalplane.
3.Gnathion(gn) Thelowestpointofthechin,onthemidsagittalplane.
4.Endocanthion Themediallimitoftheeye.
(en)
5.Ectocanthion Thelaterallimitoftheeye.
(ec)
6.Iridionlaterale(il) Themostlateralpointoftherimoftheiris.
7.Iridionmediale Themostmedialpointoftherimoftheiris.
(im)
8.Pupil(pu)a Thecentralpointoftheiris,mathematicallycalculated,betweentheIridionlaterale
andtheIridionmedialeofeacheye.
9.Zygion(zy) Thewidestpointintheregionofthezygomaticboneseeninthefrontalview.
10.Chelion(ch) Thelaterallimitofthemouth.
11.Alare(al) Themostlateralpointonthe“wing”ofthenose.
aNotdescribed,butcalculatedasthearithmeticmeanbetweentheIridions.
https://doi.org/10.1371/journal.pone.0180330.t001
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 3/19
Irisratiofortheanalysisoffacialgrowth
Fig1.Graphicrepresentationofthemeasurementsadoptedinthisstudy.*Theirismeasurementcorrespondedtotheaverage
valueoftherightandleftim-il;theimagewasobtainedandillustrativelyusedwiththeconsentofthesubjectandrelatives.
https://doi.org/10.1371/journal.pone.0180330.g001
Calculationofabsolutemeasurements
Oncelandmarkswereregistered,theircoordinateswereusedtocalculate10measurements
expressedinpixels.Amongthe10measurements,ninewereconsideredtobejustifiedforthe
presentstudyonthebasisthattheyarecommonlyusedtobuildfacialindicesinanthropomet-
ricstudies[17,18,21,22].Theremainingmeasurementcalculatedthediameteroftheirisand
wasincludedasanewapproachforanthropometricfacialanalysis(Fig1).
RelativeMeasurementGroups(RMG)
Inordertoenabletheinvestigationofallometryamongthe10measurements,RelativeMea-
surementGroups(RMG)wereestablished(Table2).Thesegroupsallowedtheobservationof
growingdifferencesfrom10diverseperspectives.Thisapproachwasfoundedondetermining
afixedreferencemeasurementtobuildspecificratios(similartoindices,butwithoutexpress-
ingcentesimalvalues)foreachgroup.Variationsbetweenmeasurementswereobservedasa
functionofage.
Sincethisapproachisfoundedontheanalysisofproportions,thefixedreferencevalue
couldbeselectedrandomlyforeachgroup.However,themeanvaluesforthereferencemea-
surementsfromthetotalsample(n=1000)wereusedtoreachvaluesclosertotheactualfacial
measurements.Oncethereferencevalueswereestablished,theimageswerescaledbasedon
thepercentagedifferencebetweenthereferencevalueandatargetvalueforeachgroup(size
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 4/19
Irisratiofortheanalysisoffacialgrowth
Table2. RelativeMeasurementGroups(RMG)andfixedmeasurements.
RMG FixedMeasurement n
RMG1 zy-zy 1000
RMG2 ec-ec 1000
RMG3 en-en 1000
RMG4 pu-pu 1000
RMG5 iris* 1000
RMG6 al-al 1000
RMG7 ch-ch 1000
RMG8 n-sn 1000
RMG9 n-gn 1000
RMG10 sn-gn 1000
*Theiriswascalculatedbythearithmeticmeanofthedistancesbetweenthepointsilandimforeacheye.
https://doi.org/10.1371/journal.pone.0180330.t002
normalization);e.g.iftheoriginalreferencemeasurementrequiredenlargementbyafactorof
1.5toreachthetargetvalue,thesameenlargementfactorwasappliedtotheremainingmea-
surements.Thisprocedurekeptthemetricproportionsofthephotographsunalteredand
enabledthecomparisonofdifferentfacialstructuresfromthesameperspectiveineachRMG.
AfterestablishingtheRMG,thevariablesineachgroupwereconvertedtopercentageval-
ues(anthropometricratios)ofthereferencemeasurements(Table3).Atthisstage,thesample
stratificationbyagerange(6,10,14,18,and22yearsold)wasusedtoassessanthropometric
variationsduringageing.Despiteknowingofthedifferencesincraniofacialdevelopment
betweensexes[1],thepresentstudydidnotaimtoinvestigatecraniofacialdevelopmentin
functionofsexdimorphism.Forthatreason,thesamplewasanalyzedgroupedbysex.
CalculationofRelativeGrowth(RG)andAverageRelativeGrowth
(ARG)
TheRelativeGrowth(RG)wasassessedforeachagegroup,withtheageof6yearsconsidered
asthe“zeropoint”,fromwhichpercentagevariationswereinvestigatedasafunctionofagein
relationtotheotheragesubgroups.OnceRGwasobtainedforeachmeasurement,Average
RelativeGrowth(ARG)wasassessedbytakingintoconsiderationthevaluescalculatedfor
eachRMG.
Table3. RepresentativeformulasforcalculatingtheproportionalmeasurementsforeachRelativeMeasurementGroupafterimagescaling.
RMGa FixedMeasurementb(n=1000) ProportionalMeasurementsc(AnthropometricRatios)
1:zyzy 2:ecec 3:enen ... 10:sngn
1 zyzy zyzy ¼zyzy ecec ¼ecec enen ¼enen ... sngn ¼sngn
RMG1 zyzy RMG1 zyzy RMG1 zyzy RMG1 zyzy
2 ecec zyzy ¼ zyzy ecec ¼ecec enen ¼enen ... sngn ¼sngn
RMG2 ecec RMG2 ecec RMG2 ecec RMG2 ecec
3 enen zyzy ¼ zyzy ecec ¼ecec enen ¼enen ... sngn ¼sngn
RMG3 enen RMG3 enen RMG3 enen RMG3 enen
... ... ... ... ... ...
10 sngn zyzy ¼ zyzy ecec ¼ecec enen ¼enen ... sngn ¼sngn
RMG10 sngn RMG10 sngn RMG10 sngn RMG10 sngn
aRMG:RelativeMeasurementGroup;
bFixedMeasurement:referencemeasurementusedforcalculatingratiosobtainedthroughthemeanreferencevaluesofeachgroup;
cProportionalMeasurements:measurementcalculatedforeachimagefromtheratiobetweentheimagemeasurementfoundandthemeanvalueofthe
fixed(reference)measurementineachgroup(calculatedforeachimageandeachRMG).
https://doi.org/10.1371/journal.pone.0180330.t003
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 5/19
Irisratiofortheanalysisoffacialgrowth
Table4. Quantificationofthemetricvariationofmeasurement1(zy-zy)asafunctionofageandRelativeMeasurementGrouptodeterminethe
AverageRelativeGrowth.
Measurement Age(years) RMG1 RMG2 RMG3 ... RMG10: ARG
zy-zy ec-ec en-en sn-gn
1.zy-zy 6to10 0 D%zyzyRMG2 D%zyzyRMG3 ... D%zyzyRMG10 ARGzyzy
6(cid:0)10 6(cid:0)10 6(cid:0)10 6(cid:0)10
10to14 0 D%zyzyRMG2 D%zyzyRMG3 ... D%zyzyRMG10 ARGzyzy
10(cid:0)14 10(cid:0)14 10(cid:0)14 10(cid:0)14
14to18 0 D%zyzyRMG2 D%zyzyRMG2 ... D%zyzyRMG10 ARGzyzy
14(cid:0)18 14(cid:0)18 14(cid:0)18 14(cid:0)18
18to22 0 D%zyzyRMG2 D%zyzyRMG2 ... D%zyzyRMG10 ARGzyzy
18(cid:0)22 18(cid:0)22 18(cid:0)22 18(cid:0)22
6to22(cumulative) 0 Δ%zyzyRMG2 Δ%zyzyRMG3 Δ%zyzyRMG10 ARGzyzy
6(cid:0)22 6(cid:0)22 6(cid:0)22 18(cid:0)22
https://doi.org/10.1371/journal.pone.0180330.t004
Table4detailstheprocessfollowedtoquantifytheARGwhenusingthecraniofacialmea-
surement1(zyzy)asreference.Analysisofthecolumnsallowsforthevisualizationofthe
proportionalgrowthofthezyzymeasurementwithineachofthe10RMGs,ineachofthedif-
ferentageranges.Zerovaluesinthefirstcolumnareexplainedbythescalingprocess,which
madethereferencemeasurementfixedforallagerangesinRMG1(zy-zy).Thecodification
Δ%zyzyRMG2canbeinterpretedasthepercentagevariationofmeasurementzy-zy,within
6(cid:0) 10
RMG2,whenconsideringthemeanvaluesobtainedfromthecohortof6to10yearolds
(6–10);thecolumnonthefarrightshowstheARGforthezy-zymeasurementineachage
rangeobtainedfromthemeanvaluesoftherelativezy-zygrowth,byconsideringallRMGs
(i.e.ARGzyzy,referstotheagerangeof6–10years).ThebottomlineofTable4showsthe
6(cid:0) 10
cumulativerelativegrowthofzy-zyfrom6to22yearsofage.
AfterARGquantificationforallmeasurements,arankingoftherelativevariationswasper-
formedconsideringtheagerangesseparatelyandcombined(cumulativegrowth).Inthispro-
cedure,positivevaluesindicatedpositivevariationsofaspecificmeasurementcomparedtothe
meangrowthoftheothers,whiletheoppositewasinterpretedfornegativevalues.
Themeasurementsthatusedirisdiameterasreference(RMG5)–proposedasanew
approachinthepresentstudy—wereusedastheparametertoanalyzevariationintheremain-
ingmeasurements.Basedonthisanalysis,anewrankingsystemwasdesignedtoprovideout-
comesclosertotherealgrowthofthemeasurementsconsidered.
Statisticalanalysis
TheKolmogorov-SmirnovtestwasusedtoassessdatanormalityandtheIntraclassCorrelation
Coefficient(ICC)appliedtotheevaluationoftheintra-andinter-examinerreliabilityforposi-
tioninglandmarksandperformingmeasurements.Descriptivestatisticswereusedforinitial
datascreening.Pearson’sCorrelationCoefficientwasusedtoassociatemeasurementswithin
eachgroupandage.ThestatisticaldifferencesineachRMGwereassessedbyapplyingthe
ANOVAtestasafunctionofage(consideringthedifferentageranges).APost-hocTukeytest
wasusedtosearchforstatisticaldifferencesbetweenspecificagerangesandtoverifytheeffi-
ciencyofeachfixedmeasurementindetectingtherelativegrowth.
AllstatisticaltestswereperformedwiththeSPSS121.0softwarepackage(IBM1,New
York,USA),withconsiderationforasignificancelevelof5%andconfidenceintervalof95%.
Researchdatabasesusedforstasticalanalysismaybefoundwithinthesupportinginformation
filesS1–S6Files.
Ethicsstatement
ThepresentresearchwasconductedwiththeapprovaloftheCommitteeofEthicsinResearch
oftheUniversityofSãoPaulo,underprotocolnumber17017213.0.0000.5440.Theimages
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 6/19
Irisratiofortheanalysisoffacialgrowth
Table5. Cephalometriclandmarkmeasurementsexpressedinpixels.
Measurement 6years(n=200) 10years(n=200) 14years(n=200) 18years(n=200) 22years(n=200) Totalsample
(n=1000)
Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD
1.zy-zy 254.84 23.70 249.77 16.78 245.85 16.82 247.15 24.77 250.01 26.23 249.52 22.21
2.ec-ec 169.15 9.20 167.65 8.41 166.74 8.91 166.26 8.55 166.70 7.74 167.30 8.62
3.en-en 62.29 4.92 61.07 4.57 60.27 4.54 59.84 4.43 60.24 4.38 60.74 4.64
4.pu-pu 114.68 6.15 114.58 5.44 114.78 5.76 115.50 5.56 116.17 5.07 115.14 5.63
5.iris 24.96 1.63 23.58 1.64 22.20 1.67 21.67 1.64 21.56 1.45 22.79 2.07
6.al-al 64.86 4.74 65.04 5.69 67.56 6.00 68.04 5.79 67.81 5.85 66.66 5.80
7.ch-ch 80.10 8.29 83.98 7.77 85.27 8.22 85.94 8.35 87.32 6.64 84.52 8.24
8.n-sn 93.67 7.79 96.17 9.02 98.22 8.52 98.32 9.49 99.44 9.50 97.16 9.10
9.n-gn 205.58 13.77 211.15 13.93 216.44 15.92 219.91 15.76 223.73 15.40 215.36 16.27
10.sn-gn 111.97 10.35 115.03 10.07 118.27 11.19 121.64 10.90 124.36 13.76 118.25 12.15
https://doi.org/10.1371/journal.pone.0180330.t005
usedtoillustratethepresentstudywereacquiredandusedwiththesignedconsentofthepho-
tographedsubjectandhisrelatives.
Results
Thedatahereinstudiedpresentedwithanormaldistribution.TheIntraclassCorrelationCoef-
ficientrevealedexcellentoutcomesforthelandmarkingperformedintheintra-andinter-
examinerreliabilitytests,whichwereperformedpriortoandduringtheresearchproject
(ICC>0.90;p<0.001).Themeanandstandarddeviationofeachofthecephalometric
measurementsobtainedbeforesizenormalizationarepresentedinTable5.Thevaluesare
expressedinpixels,withmalesandfemalesgrouped,consideringthefiveagegroupsandthe
completesample(6–22yearsold).
TheARGofeachmeasurementasafunctionoftheagerangesarepresentedinTable6.
Usingirisdiameter,ARGreachedthelowestvaluesinallagegroupsincomparisontothe
othermeasurements.Positivevaluesindicateameasurementwitharelativegrowthhigher
thanthemeanvariationofmeasurements,whilenegativevaluesindicatetheopposite.The
irisvalueswereconsistentlynegativeandwasthemeasurementleastalteredwithage.Inthe
cumulativeanalysis(6–22years),irismeasurementsreachedavalue4.2timeslower(-14.83%)
thanthesecondhighestrankedmeasurement(en-en,-4.62%).
Table6. AverageRelativeGrowthforthe10measurementsstudiedasafunctionofage.
Measurement AverageRelativeGrowth
6to10years 10to14years 14to18years 18to22years 6to22years
1.zy-zy -2.18% -1.90% 0.23% 0.39% -3.29%
2.ec-ec -1.08% -0.85% -0.57% -0.49% -2.80%
3.en-en -2.14% -1.57% -1.05% -0.10% -4.62%
4.pu-pu -0.27% -0.12% 0.33% -0.17% -0.07%
5.Iris -5.74% -6.20% -2.62% -1.23% -14.83%
6.al-al 0.01% 3.52% 0.46% -1.14% 3.00%
7.ch-ch 4.60% 1.23% 0.52% 0.87% 7.54%
8.n-sn 2.39% 1.84% -0.22% 0.51% 4.74%
9.n-gn 2.49% 2.17% 1.31% 0.96% 7.28%
10.sn-gn 2.57% 2.45% 2.58% 1.33% 9.41%
https://doi.org/10.1371/journal.pone.0180330.t006
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 7/19
Irisratiofortheanalysisoffacialgrowth
Fig2.AverageRelativeGrowthofthe10measurementsstudiedaccordingtoanage-segmentalanalysis.
https://doi.org/10.1371/journal.pone.0180330.g002
GraphicalrepresentationofARGprogressbetweenagegroups(Fig2)enablesaclearinter-
pretationofthedatareportedinTable6,showingthatfacialmeasurementshaddifferentrela-
tivegrowthbetweentheagegroups.SomeofthespecificmeasurementspresentedARGvalues
thatrangedbetweenthepositiveand/ornegativescaleforthedifferentageranges(e.g.width
oftheface:zy-zy),whileothermeasurementsconsistentlyremainedpositiveornegative,such
astheheightofthelowerthirdoftheface(sn-gn)andthediameteroftheiris.Thecumulative
approachforARG(6–22yearsold)enabledtheconstructionofaproportionalgrowthrankfor
eachofthemeasurements(Fig3).Theheightofthelowerthirdofthefacepresentedthehigh-
estpositivevalue(+9.41%),indicatingagreaterrelativegrowth,whilethediameteroftheiris
hadthemostnegativevalue(-14.83%),suggestingitisthemoststablemeasurementofthe
study.
Table7expressesPearson’sCorrelationCoefficient(r)betweenageandtherespective
RMGmeasurements.Amongthe81possibilitiesexpressedintheTable(9×9),onlytwocombi-
nationsresultedinverystrongcorrelation(0.8(cid:20)|r|<1)withage:thewidthofthelips(ch-ch)
inRMG8(n-sn),andthediameteroftheirisinRMG9(n-gn).Themeasurementsreachedout-
comesthatweremoreconsistentlyandstronglycorrelatedtoagewithinRMG5(Irisratios),
reportingsevenvalueswithstrongcorrelation(0.6(cid:20)|r|<0.8)andtwowithmoderatecorrela-
tion(0.4(cid:20)|r|<0.6),amongninepossibilities.Apartfromthesemeasurements,mostcombi-
nationsresultedinweak(0.2(cid:20)|r|<0.4)orveryweak(0<|r|<0.2)correlationswithage.
Consideringthemeanvaluesofthecorrelations(bottomline),itispossibletoinferthatall
groupshadaweakcorrelationwithage,exceptforthegroupthatusedthediameteroftheiris
asreference(RMG5).
TheANOVAtestindicatedastatisticallysignificantgrowth(p<0.05)inallRMGs,for
almostalltherelativemeasurements,andshowedmetricdifferencesfrom6to22years.On
theotherhand,thepost-hoctest(Table8)indicatedthatthedifferencesingrowthmainlytook
placeinthetwolowerageranges(6–10,10–14yearsold),whileRGdetectionfelldramatically
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 8/19
Irisratiofortheanalysisoffacialgrowth
Fig3.RankingofthecumulativeAverageRelativeGrowthexpressedaspercentagevaluesforthecumulativeagerangeof
6to22years.
https://doi.org/10.1371/journal.pone.0180330.g003
Table7. Pearson’sCorrelationCoefficient(r)assessmentofthecorrelationbetweenthe10measurementsandage(6–22years)inrelationtoeach
RMG.
Measurement RelativeMeasurementGroup—Pearson’sCorrelationCoefficient(r)
RMG1: RMG2: RMG3: RMG4: RMG5: RMG6: RMG7: RMG8: RMG9: RMG10:
zy-zy ec-ec en-en pu-pu Iris al-al ch-ch n-sn n-gn sn-gn
1.zy-zy .a -.024 .055 -.155** .489** -.265** -.287** -.261** -.400** -.298**
2.ec-ec .047 .a .109** -.445** .677** -.320** -.333** -.327** -.567** -.318**
3.en-en -.058 -.113** .a -.284** .417** -.323** -.331** -.290** -.431** -.320**
4.pu-pu .219** .445** .281** .a .709** -.186** -.237** -.185** -.404** -.234**
5.Iris -.596** -.682** -.421** -.713** .a -.645** -.628** -.696** -.804** -.567**
6.al-al .310** .317** .323** .189** .629** .a -.085** -.008 -.150** -.142**
7.ch-ch .356** .376** .343** .262** .672** .104** .a .081* -.019 -.053
8.n-sn .303** .324** .286** .189** .685** .017 -.067* .a -.132** -.062
9.n-gn .512** .566** .426** .407** .799** .152** .014 .149** .a -.062
10.sn-gn .445** .444** .412** .350** .684** .196** .075* .150** .133** .a
MEANb 0.32 0.37 0.30 0.33 0.64 0.25 0.23 0.24 0.34 0.23
a.Notcalculatedsinceoneormorevariablesareconstant.
b.Meancalculatedusingabsolutevalues.
**p<0.01,levelofsignificance;
*p<0.05,levelofsignificance
https://doi.org/10.1371/journal.pone.0180330.t007
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 9/19
Irisratiofortheanalysisoffacialgrowth
intheolderageranges(14–18,18–22yearsold).TheRMGthatemployedthediameterofthe
irisasthereferencemeasurement(RMG5)wasthegroupwithmoremeasurementscapableof
detectinggrowthintheageranges6–10,10–14,and14–18years.Intheagerangeof18–22
years,growthdifferenceswereonlyobservedforsevenmeasurementassociationsoutofthe81
possibilitieswhenconsideringallRMGs(Table8).
Oncetheirismeasurementwasobservedtoofferthebestoutcome,itwasusedtoevaluate
facialgrowthwithamoreaccurateperceptionofthemetricchangesthatoccurinthefacial
structuresaccordingtoage(Fig4).Insubjectsagedbetween6–10years,mouthwidthwasthe
facialdimensionwiththehighestgrowthpercentage(10.9%).Yetinsubjectsaged10–14years,
thenasalwidthpresentedhighestgrowthpercentage(10.3%),whileinsubjectsaged14–18
and18–22years,theheightofthelowerthirdofthefacewasthefacialdimensionwithhighest
growthpercentage—presenting5.5%and2.8%growth,respectively.Theheightofthelower
thirdofthefacealsodisplayedthehighestgrowthpercentage(28.8%)inthecumulativeanaly-
sis(6–22yearsold).Thisgrowthpercentagewasmorethantwicethatoffacewidth(zy-zy:
13.63%).
Discussion
Craniofacialgrowthisacontinuousmorphogeneticprocessthatmodifiesasetofrelatedana-
tomicstructuresinamorphological,functional,andbalancedmanner.Thisprocessisprimar-
ilyexploredinthefieldsofanthropology,genetics,dentistry,medicine,andforensics[1,8].
Allometry,whichisfoundedonthecomparisonsofanatomicproportions,isoneofthemech-
anismsforexaminingthecraniofacialgrowthprocess.Despitebeingmorecommonlyusedin
comparativeanimalbiologythaninhumananthropology[2],allometryrecentlyemergedasa
potentialtoolforforensicageestimation.
AccordingtotheStudyGrouponForensicAgeDiagnostics,standardparametersforage
estimationarenecessarytoguidebestpracticesinternationally[23–24].Specialattentionis
giventoagesofmajorlegalimportance,suchas14,18,and21years[25–28].Asrecom-
mendedbytheGroup,estimatedageshouldbebasedonthreeindependentevaluations,
namely:1)clinicalexaminationusinganthropologicalaspectsandsexualmaturity;2)hand
andwristexaminationusingcarpalradiographs;and3)dentalexaminationusingpanoramic
radiographs[29,30].However,underspecialcircumstances,theonlyevidenceavailablefor
analysisisaphotograph,inwhichtheabovementionedguidelinesarenotapplicable.Analy-
sisofthematurationofsexualtraitsmaybeconsideredwhenfullbodyimagesareobtained;
however,ageestimationthroughananalysisofsexualmaturationusingphotographsshould
notbeconsideredforforensicpurposes[17,30].Thisisprimarilyjustifiedbythefactthat
outcomesmaybebiasedbyintentionalmodifications,suchastheremovalofpubichair[16,
17].Thistypeofevidenceismostcommonincybercrimesinvolvingchildabuse.Inthese
crimes,victimsandperpetratorsmaybephotographed[30]andtheirfacialtraitsmaycon-
tributetocrimecharacterization.Nevertheless,ageestimationthroughfacialphotographsis
achallengingprocedure.
ThesearchforageindicatorsthroughfacialtraitsisjustifiedbyCattaneoetal.[16].The
authorsprovidedimagesofmodels(aged>18yearsold)fromapornographicwebsiteto23
examiners,consistingofforensicexperts,pediatricians,gynecologists,andlaypersons.The
examinerswereaskedtoestimatetheageofthemodelsmentioningwhichphotographicindi-
catorofagecontributedmostsignificantlyineachcase.Facialtraitsweredescribedasthe
mostcitedindicator,reaching64%oftheforensicexperts’opinions.Thisoutcometriggered
furtherresearchinthisfield,includingthedevelopmentofmethodologiesbasedonfacialindi-
cesforforensicageestimation[17].
PLOSONE|https://doi.org/10.1371/journal.pone.0180330 July7,2017 10/19
Description:and forensics. In the field of forensics, it acts as a valuable tool for combating child pornogra- phy. The present research proposes a new method, based on relative measurements and photo-anthropometry remains a popular approach for epidemiological and forensic studies. It consists of landmarking