You are on page 1of 79

Slicer3TrainingCompendium Slicer3 Training Tutorial

CenterlineExtractionof CoronaryArteriesin3DSlicer usingVMTKbasedTools


DanielHhn StudentofMedicalInformatics UniversityofHeidelberg,Germany
Contact:haehn@bwh.harvard.edu

NationalAllianceforMedicalImageComputing

Learning Objective
Guidingyoustepbystep throughtheprocessof centerlineextractionof CoronaryArteriesinacardiac bloodpoolMRIusingVMTK basedTools.

NationalAllianceforMedicalImageComputing

Background
Coronary heart disease (CHD) is the leading cause of death in highincome countriesandoneofthemaincausesof deathworldwide*. The primary cause for CHD is atherosclerosis of the coronary arteries and is called coronary artery disease (CAD). The extraction of the central lumen line (centerline)ofcoronaryarteriesishelpful for visualization purposes, stenosis quantification or further processing steps.
HumanHeartwithCoronaries,Author:PatrickJ.Lynch (1999),CreativeCommonsLicense

*WHOFactSheet310:http://www.who.int/mediacentre/factsheets/fs310/en/index.html

NationalAllianceforMedicalImageComputing

Materials
ThistutorialrequirestheinstallationoftheSlicer3 softwareandthetutorialdataset.Theyareavailableat thefollowinglocations: Slicer3downloadpage(Slicer3.5NightlyBuild*)
http://slicer.org/pages/Special:SlicerDownloads

UnzippedTutorialMRIdata(3files)
http://www.namic.org/Wiki/index.php/File:TutorialVMTKCoronariesCenterlinesMRI_Data_Winter2010AHM.zip

Disclaimer:ItistheresponsibilityoftheuserofSlicertocomplywithboththeterms ofthelicenseandwiththeapplicablelaws,regulations,andrules.
*oraSnapshotafterDecember2009,theSlicer3extensionsystemhastoworkproperly

NationalAllianceforMedicalImageComputing

Overview
InstallingVMTKin3DSlicer ThePipeline LoadingData ExtractingtheROI VesselnessFiltering LevelSetSegmentation CenterlineComputation Results References Acknowledgements 612 13 1419 2032 3343 4457 5869 7075 76 77

NationalAllianceforMedicalImageComputing

Installing VMTK in 3D Slicer

StarttheExtensionWizard

NationalAllianceforMedicalImageComputing

Installing VMTK in 3D Slicer

ClickNext

NationalAllianceforMedicalImageComputing

Installing VMTK in 3D Slicer

SelectallVMTKExtensions

NationalAllianceforMedicalImageComputing

Installing VMTK in 3D Slicer

ClickDownload&Install

NationalAllianceforMedicalImageComputing

Installing VMTK in 3D Slicer

ClickNext

NationalAllianceforMedicalImageComputing

10

Installing VMTK in 3D Slicer

Restart3DSlicer

NationalAllianceforMedicalImageComputing

11

Installing VMTK in 3D Slicer


1

TheVMTKExtensions appearinthemodules selector(1)insidethe categoryVascular ModelingToolkit

NationalAllianceforMedicalImageComputing

12

The Pipeline

ThispipelineisusedfortheCenterlineExtraction.
NationalAllianceforMedicalImageComputing

13

Loading Data

1 2

Toloadthetutorialdata,choosetheFilemenu (1)andselectAddVolume...(2)
NationalAllianceforMedicalImageComputing

14

Loading Data
1

Selectthefiletutorial_cardiac_mri_data.nrrd(1) oftheunzippedtutorialdataandpressApply(2)

2
NationalAllianceforMedicalImageComputing

15

Loading Data
1 2 Usethelayoutselector(1)to switchtotheConventional layout(2)

NationalAllianceforMedicalImageComputing

16

Loading Data
The2Dsliceviewersshowtheloadedvolume.

NationalAllianceforMedicalImageComputing

17

Loading Data
1

Usethelayoutselector(1)to switchtotheRedsliceonly layout(2)

NationalAllianceforMedicalImageComputing

18

Loading Data

Onlytheredsliceviewer(axialview)isdisplayed.

NationalAllianceforMedicalImageComputing

19

Extracting the ROI


1

Usethemodulesselector(1)tostartthe ExtractSubvolumeROI(2)module
NationalAllianceforMedicalImageComputing

20

Extracting the ROI


Thispanelnowappears. BesurethattheInput volumeistheloaded tutorialdata.

NationalAllianceforMedicalImageComputing

21

Extracting the ROI

Createanew MRMLROINode

NationalAllianceforMedicalImageComputing

22

Extracting the ROI

CreateanewVolume asOutputvolume

NationalAllianceforMedicalImageComputing

23

Extracting the ROI


1

Nowclickaroundthehearttoselectthesubvolume. Navigatetothemostinteriorandsuperiorslicesusing thecontrol(1)toselectthecompleteheart.The selectionisshownasablueoverlay.


NationalAllianceforMedicalImageComputing

24

Extracting the ROI


2 1 ClickDoROIresample(1)toextractthe subvolumeandclicktoggletheROIvisibility (2)tohidetheblueoverlay

NationalAllianceforMedicalImageComputing

25

Extracting the ROI

Youcanalsodirectlyloadtheprepared tutorial_cardiac_mri_dataSubvolume resample_scale1.0.nrrdfileoftheunzipped tutorialdatatogettheextractedsubvolume (seetheLoadingDatasection).

NationalAllianceforMedicalImageComputing

26

Extracting the ROI

2 Selecttheextractedsubvolume(2)inthered sliceviewerbyusingthevolumeselector(1)

NationalAllianceforMedicalImageComputing

27

Extracting the ROI


1 2 Fitthevolumetothewindow byusingtheoptionsicon(1) andselectingFittowindow (2)

NationalAllianceforMedicalImageComputing

28

Extracting the ROI


1

2 Usethemodulesselector(1) tonavigatetotheVolumes module(2)

NationalAllianceforMedicalImageComputing

29

Extracting the ROI

Besuretheextractedsubvolume tutorial_cardiac_mri_dataSubvolume resample_scale1.0istheactive Volume(1)andadjusttheWindow/Level settingto1082and1257(2)forbetter visualization

2
NationalAllianceforMedicalImageComputing

30

Extracting the ROI


1 Left Coronary Artery

Aorticarch

Atslicelocation0.80(1)the aorticarchandtheleftcoronary artery(LCA)arevisible


NationalAllianceforMedicalImageComputing

31

Extracting the ROI


1 Right Coronary Artery Aorticarch

Atslicelocation15(1)theaortic archandtherightcoronaryartery (RCA)arevisible


NationalAllianceforMedicalImageComputing

32

Vesselness Filtering
Navigatetothe VMTKVesselEnhancement moduleinsidethecategory VascularModelingToolkitusing themodulesselector

NationalAllianceforMedicalImageComputing

33

Vesselness Filtering
Thispanelappears.Switchto FrangiVesselness.

NationalAllianceforMedicalImageComputing

34

Vesselness Filtering
Selecttheextractedsubvolume (2)astheInputVolume(1)

1 2

NationalAllianceforMedicalImageComputing

35

Vesselness Filtering
Enter0.1(unit:mm)asthe minimaldiameteroftubular structurestodetect

NationalAllianceforMedicalImageComputing

36

Vesselness Filtering
Enter2.0(unit:mm)asthe maximumdiameteroftubular structurestodetect

NationalAllianceforMedicalImageComputing

37

Vesselness Filtering
Choosealowthresholdof0.3to detectlinelikeratherthanplate likestructures

NationalAllianceforMedicalImageComputing

38

Vesselness Filtering
Ahigherthresholdof500limits thedetectionofbloblike structures

NationalAllianceforMedicalImageComputing

39

Vesselness Filtering
Thecontrastofthevesselsin comparisiontothebackgroundin thetutorialdataisveryhigh,so setahigherthresholdof500to detectonlywellvisiblestructures

NationalAllianceforMedicalImageComputing

40

Vesselness Filtering

ClickStart!

NationalAllianceforMedicalImageComputing

41

Vesselness Filtering
Thefilteringproceduretakes approx.610minutes.

Youcanalsodirectlyloadtheprepared EnhancedVesselnessOut.nrrdfileofthe unzippedtutorialdatatogetthevesselness filteredvolume(seetheLoadingData section).

NationalAllianceforMedicalImageComputing

42

Vesselness Filtering

Right Coronary Artery

Theredsliceviewershowsthevesselness filteredvolume.Theenhancedtubesare visible(f.e.atslicelocation15).


NationalAllianceforMedicalImageComputing

43

Level Set Segmentation


Navigatetothe VMTKEasyLevelSetSegmentation moduleinsidethecategory VascularModelingToolkitusing themodulesselector

NationalAllianceforMedicalImageComputing

44

Level Set Segmentation


Thispanelnowappears. TheLevelSetSegmentation processconsistsoftwo steps:Initializationand Evolution

NationalAllianceforMedicalImageComputing

45

Level Set Segmentation


1 2

SelecttheEnhancedVesselnessOut volume(2)astheInputVolume(1)

NationalAllianceforMedicalImageComputing

46

Level Set Segmentation


1 2

UsetheSourceSeedsselector(1)to createanewFiducialList(2)which automaticallybecomesactive 3 SwitchtoPlacemodebyusingthe icon(3)onthetoolbar

NationalAllianceforMedicalImageComputing

47

Level Set Segmentation

ClickinsidetheRCAontheredslice viewertoplaceoneseedpoint(f.e.at slicelocation15)


NationalAllianceforMedicalImageComputing

48

Level Set Segmentation

ClickinsidetheLCAontheredslice viewertoplaceoneseedpoint(f.e.at slicelocation0.80)


NationalAllianceforMedicalImageComputing

49

Level Set Segmentation

DeactivateUseVolumeRendering becausePolydataisneededlater

NationalAllianceforMedicalImageComputing

50

Level Set Segmentation


Setalowerthresholdof 0.143 Thisresultsinimmediate visualizationfeedbackat thesliceviewers

NationalAllianceforMedicalImageComputing

51

Level Set Segmentation


1 2 Usethelayoutselector(1)to switchtotheConventional layout(2)

NationalAllianceforMedicalImageComputing

52

Level Set Segmentation

ClickStart!

NationalAllianceforMedicalImageComputing

53

Level Set Segmentation


Thesegmentationisshownasa modelinthe3Drenderingwindow (1)andasanoverlayedlabelmap inthe2Dsliceviewers(2)

2
NationalAllianceforMedicalImageComputing

54

Level Set Segmentation

1 2

UsetheInputVolumeselector(1)to settheextractedsubvolume(2)asthe inputfortheevolutionstage


NationalAllianceforMedicalImageComputing

55

Level Set Segmentation


1 2 3
Specifythebehavioroftheevolutionbyusingthesliders. Theinitializationisalreadyclosetotheedgesofthevesselssono inflationisneeded(1). Togetasmoothsurfaceahighercurvatureweightof70isimportant (2). Toattractthesegmentationtothegradientridgesahighattractionweight of100isnecessary(3).
NationalAllianceforMedicalImageComputing

56

Level Set Segmentation


Theresultisshownasablue modelinthe3Drenderingwindow (1)andasanoverlayedlabelmap inthe2Dsliceviewers(2)

2
NationalAllianceforMedicalImageComputing

57

Centerline Computation
NavigatetotheVMTKCenterlines moduleinsidethecategory VascularModelingToolkitusing themodulesselector

NationalAllianceforMedicalImageComputing

58

Centerline Computation
Thispanelnowappears. TheCenterlinesextraction consistsoftwosteps:Model preparationandCenterline Computation

NationalAllianceforMedicalImageComputing

59

Centerline Computation

1 2 UsetheInputModel selector(1)tosetthe VMTKEvolutionModel(2) astheinput

NationalAllianceforMedicalImageComputing

60

Centerline Computation

ClickPrepareModel! Thebluemodelinthe3D RenderingWindowturns green

NationalAllianceforMedicalImageComputing

61

Centerline Computation

NowusetheCenterlines Computationpanelforstep2

NationalAllianceforMedicalImageComputing

62

Centerline Computation
2 1

UsetheSourceSeeds selector(1)tocreateanew Fiduciallist(2)


Note:ItisrecommendedtousetheFiducialsmoduletohidetheFiducial listsoftheLevelSetSegmentationprocess
NationalAllianceforMedicalImageComputing

63

PlacetwoSeedsinthe3DRendering Windowdirectlyonthegreenmodelwhere theCoronariesstart

Centerline Computation

NationalAllianceforMedicalImageComputing

64

Centerline Computation
1

UsetheTargetSeeds selector(1)tocreateanew Fiduciallist(2)


NationalAllianceforMedicalImageComputing

65

Centerline Computation
1

ToplacetheTargetSeedscorrectly,itis recommendedtofirstusetheTransformmode(1)to rotatethemodelandthenthePlacemode(2)toset thefiducials

NationalAllianceforMedicalImageComputing

66

PlacefourSeedsinthe3DRendering Windowdirectlyonthegreenmodelwhere theCoronariesend

Centerline Computation

NationalAllianceforMedicalImageComputing

67

Centerline Computation

ClickGetCenterlines!

NationalAllianceforMedicalImageComputing

68

Centerline Computation

TheVoronoidiagramandthecorrespondingCenterlines appearinthe3DRenderingWindow.Usetheright mousebuttontoZoomintothe3Dview


NationalAllianceforMedicalImageComputing

69

Centerline Computation
Thisisagoodtimetosavethelumensegmentation,the generatedVoronoidiagram,theCenterlinesasPolydata andallotherMRMLdatabyusingtheFilemenuand Save.

NationalAllianceforMedicalImageComputing

70

Results

Allsegmentationpartsare availableasMRMLnodes inthecurrentscene.The Datamoduleshowsthe MRMLtree.

NationalAllianceforMedicalImageComputing

71

Results
DeactivatetheVMTKCenterlinesPrepOutmodelto hidethesegmentedlumen(1). The3DRenderingWindowthen showstheVoronoidiagramandthe correspondingCenterlinesonly.

NationalAllianceforMedicalImageComputing

72

Results
HidetheVMTKVoronoiOutmodel (1)toshowtheCenterlinesinthe 3DRenderingWindowonly.

NationalAllianceforMedicalImageComputing

73

Results
1 2

TheVMTKCenterlinesmodulesupportstheexportof extractedCenterlinesascloudsofpointstothe filesystem. Toexportdetailslikethemaximuminscribedsphere radiusactivatethecheckbox(1),choosea destination(2)andclickExport!(3).


NationalAllianceforMedicalImageComputing

74

Results
1 2

Theexportedfileincludestheworld coordinates(1)oftheCenterlinesand alsotheMaximumInscribedSphere Radius(2)foreachpoint.


NationalAllianceforMedicalImageComputing

75

Conclusion
VMTKextensionsinstallableusingtheextensionwizard VesselnessFilteringusingVMTKVesselEnhancement LumenSegmentationusingVMTKEasyLevelSetSegmentation CenterlineComputationusingVMTKCenterlines 3DSlicerIntegrationforfurtherprocessingofthedata(MRML nodes) OpenSourceEnvironment

NationalAllianceforMedicalImageComputing

76

References

NationalAllianceforMedicalImageComputing

77

Acknowledgements
NationalAllianceforMedicalImageComputing SurgicalPlanningLaboratory(BWH) StevePieper,RonKikinis MarioNegriInstitute LucaAntiga

NationalAllianceforMedicalImageComputing

78

Thankyouforusingthis tutorial!

NationalAllianceforMedicalImageComputing

79

You might also like