You are on page 1of 10

9/5/2016 DiameterJImageJ

DJ
DiameterJ
DiameterJ(ImageJ1.48ornewer(includingImageJ
DiameterJ[1]isafree,opensourceplugincreatedforImageJ,ImageJ 2.XX)andFiji)
2,andFijidevelopedattheNationalInstituteofStandardsand Author NathanHotaling
Technology.DiameterJisavalidatednanofiberdiameter Maintainer NathanHotaling
characterizationtool.DiameterJisabletoanalyzeanimageandfind
File ImageJ1.48ato2.XXXDiameterJv1.018
thediameterofnanofibersormicrofibersateverypixelalongafibers
axisandproducesahistogramofthesediameters.Includedwiththis FijianyversionDiameterJv1.018
histogramaresummarystatisticssuchasmeanfiberdiameterand
mostoccurringfiberdiameter(mode).DiameterJalsobundles Source SourceCode
OrientationJ[2]foracompleteanalysisoffiberorientationwithinan Initialrelease February2015
imageaswellasthe"AnalyzeParticles"functionbuiltinto
ImageJ/Fijitoanalyzeporespacewithinscaffoldsandproduce Latestversion August5th,2016
summarystatisticsforpores. Developmentstatus vX.003(firstversionreleasedpublicly)
Category PluginsAnalysis

Contents
1 DiameterJ
1.1 Overview
1.1.1 Citation/ReferenceInformation
1.2 DownloadLink
1.3 HowDiameterJWorks
1.3.1 Segmentation
1.3.2 SuperPixelDiameter
1.3.3 FiberDiameterHistogram
1.3.4 MeshHoleAnalysis
1.3.5 FiberOrientation
1.4 HowtoUseDiameterJ
1.4.1 LearnDiameterJTrainingModule
1.5 DiameterJOutput
1.5.1 SummariesFolder
1.5.2 HistogramsFolder
1.5.3 DiameterAnalysisImagesFolder
1.6 Limitations
1.7 Installation
2 FAQs
3 ComplementaryTools
4 FutureDevelopment
5 References
6 ExternalLinks

Overview

DiameterJ[1]isatwostepprocessofimageanalysis:

1.ImageSegmentationintoabinaryimage(blackandwhitepixelsonly)
SixteendefaultsegmentationalgorithmshavebeenincludedwithDiameterJinthe"SegmentSRM"and"SegmentMixed"
plugins.However,thesealgorithmsmaynotworkforallSEMimages.
Iftheuserisnothappywiththeresultsofthesegmentationalgorithms(i.e.theblackandwhiteimagesdonotproducean
accuraterepresentationoftheoriginalimage)thenDiameterJwillstillworkwithanybinaryimagethathasbeensegmented
throughsomeothermeans.
2.AnalysisofSegmentedimage
AllmeasuresgivenbyDiameterJareinpixelsbydefault
DiameterJhasbeenvalidatedwithover130digitalimagescreatedinsilicoandwithscanningelectronmicroscopeimagesof
referencewireswithknowndiameters.

http://imagej.net/DiameterJ 1/10
9/5/2016 DiameterJImageJ
Fibersthataresmallerthan10pxor
greaterthan10%ofthesmallest
dimensionoftheimageproduce10%
orgreatererror
FornowDiameterJonlyanalyzes.tif,
jpeg,png,.bmp,and.giffiles
Fibersmustbelessthan512pxin
diametertobeanalyzed

IfyouwouldliketociteDiameterJinyourwork,citation
informationcanbefoundhereorusethebelow:

Citation/ReferenceInformation OverviewofDiameterJanalysisflow(Top)SEMimage>segmentedimage>
stylizedEuclideandistancetransform.(Bottom)Afewofthegraphscapableofbeing
HotalingNA,BhartiK,KrielH,SimonJr.CG. producedfromdatagivenbyDiameterJ
DiameterJ:Avalidatedopensourcenanofiber
diametermeasurementtool.Biomaterials201561:32738.doi:10.1016/j.biomaterials.2015.05.015.

http://www.sciencedirect.com/science/article/pii/S0142961215004652

DownloadLink

ForImageJ1.48ornewer:DiameterJv.1.018forImageJ
ForFijilatestrelease:DiameterJv.1.018forFiji

HowDiameterJWorks

TheoverallgoaloftheDiameterJ[1]algorithmwastobeabletoanalyze
an8bitSEMimageofanyresolutionusingadesktopcomputerinless
than60seconds.ForablockdiagramandoverviewofhowtheDiameterJ
algorithmanalyzesfiberdiameterandotherscaffoldpropertiesseebelow.

Segmentation

SEMmicrographswerefirstsegmentedusingavarietyofthresholding
techniquesavailableinImageJ/Fiji.BoththeSegmentSRMandSegment
Mixedpluginsareautomatedinclusionsofotherssegmentationalgorithm
work.Specifically,wehavefoundthatforSEMimagesofnanofibersthe
StatisticalRegionMergingalgorithm[3]developedbyJohannes
Schindelindoesagreatjobofblendingfibersacrosstheirdiameterandby
depthtocreategreatrepresentationsofthefiberswhentheyare
segmented.Additionally,weusemoreconventionalsegmentation
algorithmsdevelopedbyOtsu[4],Huang[5],Kittler(minerror)[6],Doyle
(percentile)[7],orZack(triangle)[8]

Aftersegmentationallimageshadremainingnoiseandmorphological
featuresthatweresmoothedaccordingtotheprotocolsoutlinedby
D'Amore[9]andbyGonzalez[10].Briefly,successiveroundsofnoise
removal(viaImageJsdespecklecommand)wereperformeduntilno
changeintheimagewasfound.Erosion(throughImageJserode
command,anddilation(throughImageJsdilatecommand),andafinal
erosion(throughImageJserodecommand),operationsservedtorefinethe
image,highlightingfiberedgesandeliminatingisolatedpixelareas.The DiagramofDiameterJcode
describedmorphologicalprocedureswereperformedtoimprovethe
precisionofthecenterlinedeterminationsasperthemethoddevelopedby
Lamet.al.[11].

http://imagej.net/DiameterJ 2/10
9/5/2016 DiameterJImageJ

SuperPixelDiameter

Afterimagesegmentationwhitepixelsweresummedfortotalfiberareaineachimage.Twodifferentcenterlineswerethencalculatedforthe
image,oneusinganaxialthinningalgorithmdevelopedbyZhangandSuen[12](SkeletonizecommandinImageJ)andtheotherusinga
Voronoitessellation[13](VoronoicommandinImageJ).Theaxialthinningalgorithmisverysensitivetochangesinthefibersurfaceresulting
inbranchestoareasthatwerenotnecessarilynewfibers.TheVoronoialgorithmessentiallymaximizesthedistancebetweendiscreteblack
pixelclustersandthusiscompletelyinsensitivetofibermorphology.Thelengthofeachcenterlinewasthenaveragedandthetotalareaof
fiberswasdividedbytheaverageoftheaxiallythinnedandVoronoicenterlinelengths.Thetwocenterlinelengthswereaveragedbasedon
resultsfromanalyzingdigitalsyntheticimagesandthefindingthatonemethodconsistentlyoverestimatedfiberlengthwhiletheother
consistentlyunderestimatedfiberlength.

TheSuperPixelnamewaschosenbecausethefiberarea,inpixels,wasdividedbythecenterlinelengths,inpixelsthusproducingaunitless
valuethatisequivalenttomeanfiberdiameter(fiberlengthxdiameter=fiberarea).Thisvalueisthereforeatransformedpixelunitandis
equivalenttothemeanfiberdiameterundertheassumptionthatthefibersarejustlongrectangleswhensegmentedinto2Dshapes.

Afteritsinitialcalculationthediametercalculationwasthenfurtherrefinedviaintersectioncorrection.Intersectioncorrectionwasdoneby
takingtheaveragelengthofthecenterlineandsubtractingaradiusvalue(obtainedfromfirstapproximationofthediameterasdetermined
withoutintersectioncorrection)foreachthreepointintersectionandadiametervalue(obtainedfromfirstapproximation)foreachfourpoint
intersectionofthefibers.IntersectionsofeachcenterlinewerefoundusingthealgorithmdevelopedbyArganaCarrerasetal.[14].

Anewdiameterwasthencalculatedusingthenewcorrectedlengthandthetotalfiberareaandthisprocesseswasloopeduntilthediameter
convergedto1/1000thofapixel.Additionally,thenumberofintersectionswasalsosavedandtheintersectiondensity(ID)wascalculatedfor
a100pxx100px(104)pixelareabydividingthetotalnumberofintersectionsbythetotalarea(inpixels)oftheimageandmultiplyingby
104: .Thecharacteristiclength(CL)offiberswasdefinedasmeanlengthoffiberbetween

intersectionsandwascalculatedbydividingthetotalcenterlinelengthbythenumberofintersections: .The

centerlinelengthwascalculatedastheaverageoftheaxialthinningandVoronoitessellationcenterlines.

FiberDiameterHistogram

Toobtainthedistributionoffiberdiametersthesegmentedimagewastransformedwitha
Euclidiandistancetransformationalgorithm[15](DistanceMapcommandinImageJ).
Thisalgorithmtakesafiberpixelandfindsthedistancetothenearestorthogonalmesh
holeusingthesquarerootofthesumofthesquareoftheverticalandhorizontaldistances
totheholeandthentransformsthefiberpixeltoagreyscalevalueequaltothatdistance.
Theresultingimageisagreyscaleimageratherthanblackandwhite.Thecenterline
calculatedbytheaxialthinningalgorithmaboveisthenoverlaidontopofthedistance
transformedimage.Ateachintersectionofthecenterlinethegreyscalevalueisfoundand
radiivalueswithinthatrangearesubtractedoutfromthecenterline.Thegreyscalevalues
undertheremainingcenterlinearethenobtainedandmultipliedby2togetthevalueof
alldiametersnotinanintersectionarea.Thesubsequenthistogramofgreyscalevaluesis
thenfoundandplacedina.csvfilealongwiththeoverallaverage,standarddeviation,
medianandmodeofalldiametervalues.Thesensitivecenterlinewaschosenbecauseit
wasthemorediscriminatingoptionandthus,hadahigherlikelihoodofeliminatingpixels
thathadahighervaluethantherealradiusvalue. RadiushistogramproducedbyDiameterJ

MeshHoleAnalysis

Segmentedpicturescontainonlyblackandwhitepixelswithblackpixelsrepresentingbackgroundandwhitepixelsrepresentingfibers.
BlackpixelswereanalyzedusingtheAnalyzeParticlescommandinImageJ.Thisalgorithmessentiallyfindsdiscreteclustersofblackpixels,
countsthenumberofpixelsineachclusterandthenreportstheirarea.Pixelunitswereselectedforparticleanalysisaswellasacircularity
from0.001.00,theoptiontoexcludeclustersthattouchtheedgewasalsochosen.Thesubsequentparticleanalysiswasthensavedtolater

produceameshholehistogram,meanmeshholearea(producedbyaveragingallcluster
http://imagej.net/DiameterJ 3/10
9/5/2016 DiameterJImageJ
produceameshholehistogram,meanmeshholearea(producedbyaveragingallcluster
areas),andpercentmeshhole(producedbytakingthetotalnumberofblackpixelsand
dividingitbythetotalimageresolution).

FiberOrientation

FiberorientationwasdeterminedusingawellestablishedpluginforImageJcalled
OrientationJ[2].Todeterminefiberorientationanaxialthinningalgorithmwasusedand
thenthecenterlinewasenlargedby2pixels(usingtheEnlargecommandinImageJ)to
ensureaccuratemeasureoftheline.WithinOrientationJaFouriergradientwasusedwith
agaussianwindowof7pixels.Thesubsequentfrequencyhistogramoffiberorientation
wasthensavedasanimage.OrientationJlimitsaccesstotherawdataforthishistogram
thus,iftheuserdesirestherawdatatheymustusethe"OrientationJDistribution"plugin.
IntheFijiversionofDiameterJtheDirectionality[16]plugin,writtenbyJeanYves
Tinevez,canalsobeusedtoobtaintherawdistributiondata.Thisapproachis
considerablyslowerandlesselegant(popupwindowsremainopen)thanOrientationJ
butfornowitistheonlywaytoobtaintheraworientationdatainbatch.

HowtoUseDiameterJ

LearnDiameterJTrainingModule

Anindepthtraining,calledLearnDiameterJ,hasbeendevelopedforusersofDiameterJ.
Intotalthetrainingasbeenbrokeninto7components:
MeshholesmeasuredbyDiameterJ
1.InitialInstallation/Training/UseofDiameterJ
OverviewofhowtoinstallDiameterJ
TrainingdocumentsexplaininggeneralfeaturesofDiameterJsoftware/operation
Detaileddescriptionofhowtoconvertpixelstounitsforanalysis
Quiztoensureunderstandingofconceptscoveredinthetrainingdocuments
2.PreTest
Forresearchpurposes,thistrainingisnotrequiredtolearnhowtouseDiameterJ
However,we'dappreciateitifyoutookthePreTestsothatwecanpublishtrainingresultslater
3.Training1CroppingandSegmenting
HowtocropimagesforusewithDiameterJ
HowtosegmentimagesusingDiameterJ
Segmentationselectioncriteriaforselectingthebestsegmentationfromtheoptionsavailable
4.Training2ManualSegmentation
Thistrainingcovershowtoefficientlymanuallycorrectimagesegmentations
5.Training3AssessmentofFiberDiameter
Determiningaveragefiberdiameter,
Identificananduseofhistogramsofthefiberdiameter
Multiplefiberdiametersidentification
Selectioncriteriaforwhentoacceptafiberanalysis
6.Training4OtherMetricAssessment
HowtouseDiameterJtoassessothermetricsofscaffolds
7.PostTest
Forresearchpurposes,thistrainingisnotrequiredtolearnhowtouseDiameterJ
However,we'dappreciateitifyoutookthePostTestsothatwecanpublishtrainingresultslater

PleasegotoLearnDiameterJtotakethetraining!

DiameterJOutput

http://imagej.net/DiameterJ 4/10
9/5/2016 DiameterJImageJ

SummariesFolder

"FileSourceName"_TotalSummary.csv "FileSourceName"_TotalSummary.csvContinued

1.SuperPixel:Themeanfiberdiameterascalculated 1.FiberLength:ThetotallengthoftheFibercenterlinesin
usingasuperpixeldetermination.Adetailed thesegmentedimage.
explanationofhowthisalgorithmworkscanbefound 2.MeanPoreArea:(Totalnumberofblackpixelscounted
above. inpores)/(Totalnumberofporesinimage)
2.HistogramMean:Themeanfiberdiameterascalculated 3.PoreAreaSD:Thestandarddeviationofallporeareas
byfittingaGaussianCurvetotheradiusdataand measured
findingthecurvesmeanvalue. 4.Min.PoreArea:Theminimumporeareameasured
3.HistogramSD:Thestandarddev.oftheGaussianfitof 5.MaxPoreArea:Themaximumporeareameasured
theradiushistogram 6.PercentPorosity:(Totalnumberofblackpixels)/(Total
4.HistogramMode:Mostoccurringfiberdiameterinthe pixelsinanimage)
histogram 7.IntersectionDensity(100x100px):(Numberoffiber
5.HistogramMedian:Middlefiberdiameterinthe overlaps)*10000/(Totalpixelsinimage)
histogram 8.CharacteristicLength:(Totallengthofallfiber
6.HistogramMin.Diam.:Thesmallestdiametermeasured centerlines)/(#offiberoverlaps)
7.HistogramMax.Diam.:Thelargestdiametermeasured
8.HistogramIntegratedDensity:Theproductoflengthof
thefibersandtheaverageradius
9.HistogramRawIntegratedDensity:Thesumoftheradii
atallpixelsintheimageorselection
10.DiameterSkewness:Thethirdordermomentaboutthe
mean
11.DiameterKurtosis:Thefourthordermomentaboutthe
mean.

HistogramsFolder

"FileSourceName"_PoreData.csv "FileSourceName"_Histogram.csv

1.Slice:Imagename 1.RadiusValue:Radiuslength(inpixels)
2.Count:Totalnumberofporesfoundinimagethatare 2.RadiusCount:Numberoftimestheradiusvalue
nottouchingtheside occurredintheimage.Alsoknownasthefrequencyof
3.Area:Totalnumberofblackpixelsinanimagenotin occurrenceandcanalsobeinterpretedasthelengthof
groupstouchingthesides fiberinanimagethathasagivenradius
4.StdDev:0(individualporemeasurementssonoSD Theradiusordiameterhistogramisconstructed
5.%Area:(Totalnumberofblackpixels)/(Totalpixelsin fromtheradiusvalue(xaxis)andradiuscount
animage) (frequencyofoccurrenceonyaxis)
6.Major:Thelengthoftheprimaryaxisofthebestfitting
ellipseforeachpore.
7.Minor:Thelengthofthesecondaryaxisofthebest
fittingellipseforeachpore
8.Angle:Theanglebetweentheprimaryaxisandaline
paralleltotheXaxisoftheimage
9.Circ.:4[Area]/[Perimeter^2]withavalueof1.0
indicatingaperfectcircle.Asthevalueapproaches0.0,
itindicatesanincreasinglyelongatedshape:::#Values
maynotbevalidforverysmallparticles.Usesthe
headingCirc.
10.Skew:ThethirdordermomentaboutthemeanNaN
becauseonlyonepore
11.Kurt:ThefourthordermomentaboutthemeanNaN
becauseonlyonepore
12.AR:Theaspectratiooftheparticlesfittedellipse,i.e.,
[MajorAxis]/[MinorAxis]
13.Round:4[Area]/([Majoraxis]2)ortheinverse
ofAspectRatio.
14.Solidity:[Area]/[Convexarea]

http://imagej.net/DiameterJ 5/10
9/5/2016 DiameterJImageJ
"FileSourceName"_IntersectionCoordinates.txt "FileSourceName"_RadiusHistogram.tif

1.Column1Greyscalevalueatintersection 1.Imageofthehistogramofallfiberradiiintheimage
2.Column2xcoordinateofintersection
Upperlefthandcorneris0,0
3.Column3ycoordinateofintersection
Upperlefthandcorneris0,0

DiameterAnalysisImagesFolder

"FileSourceName"_Compare.png "FileSourceName"_Orientation.tif

Amontageimagewithfourimagesinit 1.Animagewiththefrequencyoforientationofthe
centerlineofallfibers.ThisisanoutputofOrientationJ
1.TopLeftImageOriginalsegmentedimage andwasnotcodedbyanydevelopersfromDiameterJ
2.ToprightImageImageofthecenterlineasdetermined
bytheVoronoitessellationalgorithm
3.BottomLeftImageImageofallcenterlinescountedin
thehistogramoverlaidontheEuclideandistance
transformedofthefibersfibers.
Yellowlinesarethelocationswhereradiiwere
counted
Fibersareingreyscaleastransformedbythe
Euclideandistancetransform
4.BottomRightImagePoresfoundandanalyzedby
Diameterj

Limitations

1.Duetomathematicallimitationsfibersthataresmallerthan10pxorgreaterthan10%ofthesmallestdimensionoftheimage
produceerrorsthatareabove10%andthusdeemedastoohighforaccuratemeasurement.Thisisbecauseatsmalldiameters
(lessthan10px)asinglepixelerrorcancause10%errorinmeasurement.Whilewithlargefibers(greaterthan10%ofthe
smallestdimensionoftheimage)lowsamplingandedgeeffectsoffibersbecomeadominantfactorindiameterdistribution.This
limitationcanbeovercomebyalteringyourmagnificationwhentakingtheimagetomakefiberssmaller/largerasappropriate.
2.Duetoalgorithmicreasonsfibersthathaveadiameterlargerthan512pxcannotbeanalyzedbyDiameterJ.Ingeneral,thisisnot
aconcerniftheresearcheristakingimagesthatdonotviolatethefirstlimitationbecauseanimagewouldhavetobeover5120
pxonitssmallestdimensionforthislimitationtoaffectthem.However,withstitchedimagesandwithultrahighimage
resolutions(36MPorgreater)thiscouldprovetobealimitingfactor.WorkisunderwaytoconverttheEuclideanDistance
Transformtoa16bitalgorithmthatcananalyzeimageswithfibersthatare131,072pxorlessindiameter.
3.Imagesthathavemanyfibersthatentwineorthattouchandrunparalleltoeachotherforlongdistancesinanimagecanproduce
errantpeaksinDiameterJ.Thisisusuallyduetosegmentationoftheseimagesshowingonelargerfiberinsteadofmanysmaller
fibersbundled.Additionally,imageswherepartialfibersrunalongtheedgeoftheimagecancauseDiameterJtoshowfiber
diametersthataresmallerthantheactualfiber.Thus,a"commonsense"checkonallimagesmustbeperformedandwhiletaking
imagestheresearchermustbecarefultoselectareaswherefibersdonotoverlapandrunparallelconsiderablyorrunalongthe
edgeoftheimage.
4.AnintegralpartofanalysisforDiameterJisimagesegmentation.However,howanimageissegmentedintoforegroundor
backgroundcandrasticallyaltertheoutputsproducedbyDiameterJlistedbelow:

NormalizedOrientationIndex
MeanMeshHoleSize
PercentPorocity
IntersectionDensity
CharacteristicFiberLength
FiberOrientationHistogram

Thisisbecausethemorefibersthataresegmentedoutofanimagethelargertheporesize,thelowertheintersection
densityandthehigherthecharacteristicfiberlengthbetweenintersections.Additionally,fiberorientationcanbeshifted
becausetheorientationofunderlyingfiberscanbedifferentthanthefibersthatareclosertothetoplayer.Thus,whilethe
algorithmsanalyzingthesemetricsareabletoproducehighlyaccurateresultsoncalibrationimagesonrealsegmented
imagestheanalysisisentirelydependentonsegmentationalgorithmused.While,thesealgorithmswereincorporatedinto

http://imagej.net/DiameterJ 6/10
9/5/2016 DiameterJImageJ
DiameterJtoprovideareferencewaytocalculatethesevalueswecautionthatunlessidenticalsegmentationalgorithms
andimagecapturesettingsareused,resultsfromthesealgorithmsarenotcomparablebetweenimages.

Fiberdiameterisrelativelyinvarianttosegmentationbecausegoodimagesegmentationleadstohavingnopartialfibers,onlyless
fibersbeingincludedinthesegmentation.Thus,whiledifferentsegmentationalgorithmsmayproduceincreasedordecreasedfiber
samplingthemeasureoffiberdiameterswithineachsegmentationisaccurate.Developingasegmentationalgorithmthatdoesnot
havethisproblemacrossanyimageiscurrentlyatopicofheavyresearchintheimageanalysisfield.

5.WhenanalyzingtheimageswithmultipledistinctfiberdiametersithasbeenfoundthatDiameterJunderrepresentedthe
prevalenceoflargerdiameterfibersascomparedtosmallerfibers.Thisbiasleadstoweightedaveragesbeinginaccurateandfor
globalaveragesoffiberdiameterstobebiasedtowardsmallerfiberdiameters.Thebiasisduetointersection
correction/subtractionandtheincreasedprobabilityoflargerfibershavingmoreintersectionsduetotheirincreasedsize.This
limitationcanbepartiallyovercomebydirectionalintersectioncorrectionoffiberintersection,whichiscurrentlyunder
development.However,thislimitationisinherenttoalltwodimensionalimageanalysistechniquesandcannotbecompletely
eliminated,makingweightedaveragesoffiberdiameterslightlyinaccuratenomatterwhat.
6.Distinguishingbetweenfiberdiametersthatarecloserindiameterthan3pixelsoftencreatesresolutiondifficultiesforpeak
fittingandtheprevalenceofeachfiberdiametercanbeskewed.Thishastodowithinherentlimitationsduetopixelbinningof
diagonallinesinEuclideanDistanceTransforms.Thislimitationcanbeovercomebyincreasingmagnificationofyourimageso
thatfiberswithdifferentdiametershavemorethan3pixelsofdifferencebetweenthem.

Installation

IfyouinstalledimageJbeforetheendof2013youshoulduninstallyourcurrentversionofImageJ(DONOTUPDATE)andreinstallImageJ
1.48ornewer.

Beforeuninstallbesuretocopyallofyouroldpluginsintoaseparatefolderasthesewillberemovedwhenyouuninstallyour
oldversionofImageJ.
WerecommendImageJoverFijiifyouhavenoexperiencewitheithersoftwarebecauseitissimplertouse.

DownloadandinstallImageJ1.48ornewerorFiji(anyversion)
Windows OSX Linux

1.Downloadandunzipthe 1.FollowinstructionsHereorHere 1.Downloadandunzipthe


DiameterJfiles(Findin"Source" forinstallationofImageJ/Fijion DiameterJfiles(Findin"File"in
above)andmoveorcopythethree OSX theInfoboxabove)
foldersintothepluginsfolderof 2.Moveorcopythethreefolders
ImageJ. intothepluginsfolderofthe
2.Thatshouldbeindirectory: directorywhereyouhaveplaced
ImageJ.
"C:\Program
Files\ImageJ\plugins" DiameterJwillwork
withx86(32bit)or
Or x64(64bit)versions
ofJava/ImageJ
"C:\ProgramFiles
(x86)\ImageJ\plugins" 3.RestartImageJ
DiameterJwillwork
withx86(32bit)or
x64(64bit)versions
ofJava/ImageJ

3.RestartImageJ

FAQs
1.Q:Whenrunningeithersegmentationalgorithmanerroroccursthatsayseither"Unrecognizedcommand:"AutoThreshold""or
"Unrecognizedcommand:"AutoThreshold..""

A:YouhaveinstalledthewrongversionofDiameterJintoyourImageJ/Fijirelease.Fijigivesthe"AutoThreshold.."
erroranditmeansyouinstalledImageJ'spluginintoFiji.ImageJgivesthe"AutoThreshold"erroranditmeans
you'veinstalledFiji'ssoftwareversionintoImageJ0.PleasedownloadandinstallthecorrectversionofDiameterJ
forthepieceofsoftwareyouareusing.

http://imagej.net/DiameterJ 7/10
9/5/2016 DiameterJImageJ
2.Q:WhenIstartImageJorFijiforthefirsttimeafterinstallingDiameterJIget"Pluginconfigurationerror:C:\...Duplicate
command:"XXXX"(alreadyin"YYYY")

Where"..."isthedirectorywhereyourpluginislocated,"XXXX"isthenameofthepluginand"YYYY"isthenameofthedirectorywherethat
pluginisalready.
A:Youhaveduplicateplugins!GotothefilewhereyouunzippedDiameterJanditsotherpluginsopenthe
"DiameterJ","OrientationJ",or"AnalyzeSkeleton2D3D"folder"anddeletethefilenamedXXXX

3.Q:WhenIrunDiameterJanerroroccursthatsays"UnrecognizedCommand:"SkeletonIntersections""or"Unrecognized
Command:"OrientationJ""or"UnrecognizedCommand:"AnalyzeSkeleton(2D/3D)""or"UnrecognizedCommand:"Statistical
RegionMerging""

A:DuringinstallationoneormoreofthepluginsthatisneededforDiameterJwasmissed.Pleasegobacktothezip
filethatyoudownloadedforDiameterJandcopyallfilesintothepluginsfolderofImageJ/Fiji

4.Q:WhenIrunDiameterJanerroroccursthatsays"UnrecognizedCommand:AAAA"whereAAAAisanycommandnotlisted
above.

A:YouareprobablyusingaversionofImageJthatwasupdatedtov.1.48orneweranddidnotcreateafreshinstall
ofImageJ.DiameterJusesseveralplugins/scriptsthatImageJdoesnotincludeinitsupdates,theyONLYinclude
theminfreshinstalls.ThusyoumustuninstallImageJandreinstallanewversion1.48ornewer.Pleaseremember
saveacopyofanypluginsyouaddedtoImageJbeforeuninstallingit,thesewillbelostduringtheuninstallprocess
unlessyousaveacopyinanotherlocationonyourcomputer.

5.Q:WhenIrunDiameterJanerrorappearsinthelogthatsays"Errortherearenofibersin"BBBB".tiftoanalyze".

A:Theimagesthatareinthefolderthatyouhaveselectedarenotbinary(blackandwhite)imagesortheimageis
completelyblack.PleasesegmentyourimagebeforetryingtoanalyzeitwithDiameterJandthenmovethe
segmentedimageintoaseparatefolderwithonlyblackandwhiteimagesinit.

6.Q:NoerroroccursbutwhenIasktheSegmentXXXorDiameterJtoanalyzeafolderthatIhaveimagesin,nooutputis
producedbySegmentXXX/DiameterJ.

A:Theimageisprobablynota.tiffile.FornowDiameterJonlyanalyzes.tifimages.Pleasesaveyourimagesas.tif
filesandthenanalyzewithDiameterJ

7.Q:DiameterJkeepsgivingmeanerroronafileandwon'tcontinueontothenextfile

A:UnfortunatelyDiameterJgoesseriallythroughfilesandisn'tcapableofskippingafilewithanerror.Simply
removethefilethatisgivingtheerrorfromthefolderyouwishtoanalyzeandrerunDiameterJ.

8.Q:NoneoftheimagesIamanalyzingaresegmentingwellwithyouralgorithms,whynot?

A:ThealgorithmsincludedbydefaultwithDiameterJrelyheavilyonuniformityoffibercolorand/oradark
background.Belowarefourgoodexamplesandfourexamplesthatworkpoorlyforimagesegmentationwiththe
defaultalgorithms.KeepinmindtherearemanymoresegmentationalgorithmsthanIhaveincludedwithDiameterJ
inbothFijiandImageJ.SeetheComplementaryToolsorImageSegmentationsectionsofthisworkforafewofthe
optionsavailable.

ComplementaryTools
DiameterJ[1]workswithseveralpluginsofImageJ/Fiji.FirstandforemostOrientationJ.Also,othergreattoolsthatwehaveincorporatedare
http://imagej.net/DiameterJ 8/10
9/5/2016 DiameterJImageJ

DiameterJ[1]workswithseveralpluginsofImageJ/Fiji.FirstandforemostOrientationJ.Also,othergreattoolsthatwehaveincorporatedare
AnalyzeSkeletonfunctionfromIgnacioArgandaCarreras,aswellasSkeletonIntersectionsfromGabrielLandini'smorphologyplugin.We'd
alsoliketogiveabigthankyoutotheStatisticalRegionMergingalgorithmwhichmakesoursegmentationsworkalotbetter.Speakingof
which,weincludealotofsegmentationalgorithmsthatwearen'tnearlytalentedenoughtoofdevelopedourselves.Inparticularthose
techniquesoutlinedaboveintheSegmentationsectionofthisarticleandcanbefoundinthelinksbelow.

We'dalsoliketosaythatDiameterJplaysnicelywiththeoutputfromanyothersegmentationalgorithmthatproducesabinaryimage.Fijihas
atonofSegmentationalgorithmsthataregreatfordifferenttypesofimages.Additionally,manypluginshavebeencreatedforuseinImageJ
orFiji:IJPluginsAutoThresholdandAutoLocalThreshold.Ifourdefaultsdon'tworkthentryany/allofthese.

Finally,we'dliketoencourageeveryonetodopeakfittingofthediameterhistogramsthatDiameterJproduces.Todothisanypeakfitting
toolcanbeused.AfreeresourceforWindowsisFitykhowever,wedon'trecommendanysoftwareinparticular.

FutureDevelopment
1.ReducingbiasoftheHistogramintersectioncorrectionbydirectionallysubtractingfiberintersectionsratherthanblanket
subtractionwithinagivenintersectionradius
2.incorporationofGaussianpeakfittingalgorithmswithinDiameterJitself.
3.CurrentlyweareworkingonanativeJAVAapplicationonDiameterJ.Thiswillnotfundamentallychangethefunctionof
DiameterJhowever,itwillmakeitfaster,lookcleaner,andshouldsolvecontinuityissues
4.16bitEuclideandistancetransformcalculator
5.EnableDiameterJtoskiptothenextfileifanerroroccursduringanalysis

CompletedGoals
1.CombinethethreedifferentreleasesofDiameterJintoasinglereleaseforallversionsofImageJ/Fiji
2.EasiertouseGUI(anyGUIatall)withmoreoptionsforwhattheoutputsofDiameterJareandwhattypesofanalysisit
performs.
3.MakeDiameterJcompatiblewithimagesotherthan.tiffiles

Helpiswelcomeinany/alloftheseimprovements!

References
1.HotalingNA,BhartiK,KrielH,SimonJr.CG.DiameterJ:Avalidatedopensourcenanofiberdiametermeasurementtool.Biomaterials201561:327
38.doi:10.1016/j.biomaterials.2015.05.015http://www.sciencedirect.com/science/article/pii/S0142961215004652
2.R.Rezakhaniha,A.Agianniotis,J.T.C.Schrauwen,A.Griffa,D.Sage,C.V.C.Bouten,F.N.vandeVosse,M.UnserandN.Stergiopulos,
Experimentalinvestigationofcollagenwavinessandorientationinthearterialadventitiausingconfocallaserscanningmicroscopy,Biomechanicsand
modelinginmechanobiology,SpringerLink(DOI:10.1007/s102370110325z)
3.R.Nock,F.Nielsen(2004),"StatisticalRegionMerging",IEEETrans.PatternAnal.Mach.Intell.26(11):14521458
4.OtsuN.Athresholdselectionmethodfromgraylevelhistograms.Automatica197511:237.
5.HuangLK,WangMJJ.Imagethresholdingbyminimizingthemeasuresoffuzziness.PatternRecognit199528:4151.doi:10.1016/0031
3203(94)E0043K.
6.KittlerJ,IllingworthJ.Minimumerrorthresholding.PatternRecognit198619:417.doi:10.1016/00313203(86)900300.
7.Doyle,W(1962),"Operationusefulforsimilarityinvariantpatternrecognition",JournaloftheAssociationforComputingMachinery9:259267,
doi:10.1145/321119.321123
8.ZackGW,RogersWE,LattSA(1977),"Automaticmeasurementofsisterchromatidexchangefrequency",J.Histochem.Cytochem.25(7):74153,
PMID70454
9.DAmoreA,StellaJA,WagnerWR,SacksMS.Characterizationofthecompletefibernetworktopologyofplanarfibroustissuesandscaffolds.
Biomaterials201031:534554.doi:10.1016/j.biomaterials.2010.03.052.
10.GonzalezRC,EddinsSL.DigitalImageProcessingUsingMATLAB,2nded.2ndedition.S.I.:GatesmarkPublishing2001.
11.LamL,LeeSW,SuenCY.ThinningMethodologiesAComprehensiveSurvey.IEEETransPatternAnalMachIntell199214:86985.
doi:10.1109/34.161346.
12.ZhangTY,SuenCY.AFastParallelAlgorithmforThinningDigitalPatterns.CommunACM198427:2369.doi:10.1145/357994.358023.
13.OkabeA.SpatialTessellations:ConceptsandApplicationsofVoronoiDiagrams.2edition.ChichesterNewYork:Wiley2000.
14.ArgandaCarrerasI,FernndezGonzlezR,MuozBarrutiaA,OrtizDeSolorzanoC.3Dreconstructionofhistologicalsections:Applicationto
mammaryglandtissue.MicroscResTech201073:101929.doi:10.1002/jemt.20829
15.F.Leymarie,M.D.Levine,in:CVGIPImageUnderstanding,vol.55(1992),pp8494http://dx.doi.org/10.1016/10499660(92)90008Q
16.Liu.Scalespaceapproachtodirectionalanalysisofimages.Appl.Opt.(1991)vol.30(11)pp.13691373

ExternalLinks
OrientationJ
NIST

http://imagej.net/DiameterJ 9/10
9/5/2016 DiameterJImageJ

Retrievedfrom"http://imagej.net/index.php?title=DiameterJ&oldid=21558"

Thispagewaslastmodifiedon8August2016,at13:18.

Privacypolicy AboutImageJ Imprint

Categories: Plugins Analysis

http://imagej.net/DiameterJ 10/10

You might also like