MODELING AND SIMULATION OF FLOW IN CEREBRAL ANEURYSMS JULIA MIKHAL Depromotiecommissie: Voorzitterensecretaris: Prof.dr.ir.A.J.Mouthaan UniversiteitTwente Promotoren: Prof.dr.ir.B.J.Geurts UniversiteitTwente Prof.dr.ir.C.H.Slump UniversiteitTwente Leden: Prof.dr.ing.V.Armenio Universita`degliStudidiTrieste Prof.dr.S.A.vanGils UniversiteitTwente Prof.dr.J.G.M.Kuerten UniversiteitTwente Prof.dr.C.B.L.M.Majoie AcademischMedischCentrumAmsterdam Prof.dr.A.E.P.Veldman RijksuniversiteitGroningen Prof.dr.ir.F.N.vandeVosse TechnischeUniversiteitEindhoven TheresearchpresentedinthisthesiswasdoneinthegroupMultiscaleModelingandSimu- lation(Dept.ofAppliedMathematics),incollaborationwiththegroupSignalsandSystems (Dept.ofElectricalEngineering),FacultyEEMCS,UniversityofTwente,TheNetherlands. ComputingresourcesweregrantedbytheNationalComputingFacilitiesFoundation(NCF), withfinancialsupportfromtheDutchOrganizationforScientificResearch(NWO). Modelingandsimulationofflowincerebralaneurysms.Ph.D.Thesis,UniversityofTwente, P.O.Box217,7500AEEnschede,TheNetherlands. c JuliaMikhal,Enschede,2012. (cid:13) PrintedbyWo¨hrmannPrintService,Zutphen,TheNetherlands. Coverphoto‘LightsandWater’ c JamesAdamson. (cid:13) ISBN:978-90-365-3433-8 DOI:10.3990/1.9789036534338 MODELING AND SIMULATION OF FLOW IN CEREBRAL ANEURYSMS PROEFSCHRIFT terverkrijgingvan degraadvandoctoraandeUniversiteitTwente, opgezagvanderectormagnificus, prof.dr.H.Brinksma, volgensbesluitvanhetCollegevoorPromoties inhetopenbaarteverdedigen opvrijdag19oktober2012om12:45uur door IuliiaOlegivnaMikhal geborenop6juni1984 teKharkiv,Oekra¨ıne Ditproefschriftisgoedgekeurddoordebeidepromotoren: Prof.dr.ir.B.J.GeurtsenProf.dr.ir.C.H.Slump To myparents Contents 1 Introduction........................................................... 1 2 Immersed boundarymethodforthe computationofflow invesselsand cerebralaneurysms .................................................... 7 2.1 Introduction ....................................................... 7 2.2 Computationalmodelforflowinsidecerebralaneurysms ................. 12 2.2.1 Incompressibleflowincomplexdomains ........................ 12 2.2.2 Numericalmethodforsimulatingincompressibleflowwith an immersedboundaryapproach.................................. 14 2.2.3 Maskingfunctionstrategy..................................... 18 2.3 ValidationoftheIBmethod.......................................... 22 2.3.1 Flowinstraightvessels ....................................... 23 2.3.2 Flowincurvedvessels........................................ 27 2.3.3 Flowinamodelaneurysm..................................... 29 2.4 Shearstressincurvedvesselandmodelaneurysm ....................... 30 2.4.1 ValidationoftheIBcomputedshearstress ....................... 30 2.4.2 Analysisoftheshearstressdistributioninsidecurvedvesseland modelaneurysm ............................................. 33 2.5 Concludingremarks ................................................ 36 3 Flowincerebralaneurysmsderivedfrom3Drotationalangiography ........ 39 3.1 Introduction ....................................................... 39 3.2 Computationalmodelforbloodflowincerebralaneurysms ............... 43 3.2.1 Navier-Stokesequationsandimmersedboundarymethod........... 43 3.2.2 Segmentationof3Drotationalangiographydata .................. 45 3.2.3 Elementaryoperationsonthemaskingfunction................... 49 3.3 Flowinarealisticaneurysm.......................................... 52 3.3.1 Motivationanddefinitionofthereferencecase.................... 52 vii viii Contents 3.3.2 Qualitativeimpressionofflowandforcesinsidetheaneurysm....... 55 3.3.3 Flowinpartiallyfilledcerebralaneurysms ....................... 58 3.3.4 ReliabilityofIBpredictions:agridrefinementstudy............... 61 3.4 Sensitivityofflowpredictionstoelementaryoperationsonmaskingfunction. 64 3.5 Sensitivityofflowpredictionsinboundinggeometries ................... 67 3.5.1 Gridcoarseningandboundinggeometries........................ 68 3.5.2 Numericalboundingsolutions ................................. 70 3.6 Concludingremarks ................................................ 73 4 Transitionofpulsatileflowincerebralaneurysms.......................... 75 4.1 Introduction ....................................................... 75 4.2 Computationalmodelofcerebralpulsatilebloodflow .................... 78 4.2.1 ImmersedBoundarymethodandaneurysmgeometry.............. 78 4.2.2 Flowconditionsandpulsatileforcing ........................... 82 4.2.3 Referencepulsatileflow....................................... 84 4.3 Transitionalpulsatileflow ........................................... 86 4.3.1 Shearstressresponseinnormalandpathologicalflow.............. 86 4.3.2 Robustnessofpulsatiletransition............................... 89 4.3.3 Frequenciesofthepulsatilesolution ............................ 91 4.4 Concludingremarks ................................................ 94 5 ConclusionsandOutlook ............................................... 97 References.................................................................105 Chapter 1 Introduction There is a growing medical interest in the prediction of flow and forces inside cerebral aneurysms [48, 103], with the ultimate goal of supporting medical procedures and deci- sions by presenting viable scenarios for intervention. The clinical background of cerebral aneurysmsand possible hemorrhagesis well introducedin the literature such as [93, 101]. Thesedays,withthedevelopmentofhigh-precisionmedicalimagingtechniques,thegeome- tryandstructureofbloodvesselsandpossibleaneurysmsthathaveformed,canbedetermined forindividualpatients.Todate,surgeonsandradiologistshavetomakedecisionsaboutpos- sibletreatmentofananeurysmbasedonsize,shapeandlocationcriteriaalone.Inthisthesis wefocusontheroleofComputationalFluidDynamics(CFD)foridentifyingandclassifying therapeuticoptionsinthetreatmentofaneurysms. CFD allows to add qualitative and quantitative characteristics of blood flow inside the aneurysmtothecomplexprocessofmedicaldecisionmaking.Wecontributetothisbycom- puting the precise patient-specific pulsatile flow in all spatial and temporal details, using a so-called‘ImmersedBoundary’(IB)method.Thisrequiresanumberofsteps,frompreparing therawmedicalimagerytodefinethecomplexpatient-specificflowdomain,totheexecution ofhigh-fidelitysimulations.Subsequently,detailedinterpretationoftheconsequencesofthe flow shouldbe givenjointly by medicalexpertsandCFD specialists in termsof flow visu- alization and quantitative measures of relevance to medical practice. We compute the flow inside theaneurysmto predicthighandlowstress regions,indicativeofpossiblegrowthof ananeurysm.Wealsovisualizevorticalstructuresintheflowindicatingthequalityoflocal blood circulation. We show that, as the size of the aneurysmincreases, qualitative changes intheflowbehaviorcanarise,whichexpressthemselvesashigh-frequencyvariationsinthe flow and shear stresses. These rapid variations could be used to quantify the level of risk associated with the growing aneurysm. Such computationalmodeling may lead to a better understandingoftheprogressiveweakeningofthevesselwallanditspossiblerupture. 1 2 1 Introduction Beforepresentingthecontentofthisthesis,wewillbrieflydiscussthemedicalmotivation oftheproblem,thevariousrolesCFDcanplayandgiveanoverviewofnumericalmethods beingusedforbloodflowsimulations.Subsequently,thefocuspointsofeachchapterwillbe presentedandfinallythegeneraloutlineofthisthesiswillbegiven. Cerebral Aneurysm. Medical motivation. The medical condition known as cerebral aneurysmisoneofthecardiovasculardiseasesthatinvolvesvesselwallsofcerebralvessels. Bulgeformationsmightdeveloponthevesselwalls,andovertime,underpermanentpulsat- ingforces,theaneurysmmaygrowfurtherandevenrupture.Thedegradationofendothelial cells in the vessel walls is often associated with regions of relatively low shear stress [9] while locations of relatively high shear stress could be important for initial formation of a bulge[84].Commonly,cerebralaneurysmsarelocatedinorneartheCircleofWillis[101]– the centralvesselnetworkforthe supplyofbloodto thehumanbrain.Therisk-areasareat ‘T’and‘Y’-shapedjunctionsinthevessels[34].Treatmentofcerebralaneurysmsoftenin- volvesinsertionofaslendercoil.Thiscoilingprocedurerepresentsconsiderableriskduringa surgicalintervention,aswellasuncertaintyaboutthelong-termstabilityofcoiledaneurysms [86,94].Bloodvesselsandaneurysmsarerathercomplexbytheirstructureandgeometrical shapes.Thewallsofbloodvesselscontainseverallayersofdifferenttypesofbiologicalcells, which provide elasticity to the vessels and determine their compliance [76]. The shape of cerebralaneurysmsdevelopingin patientscanbe inferredbyusingseveraltechniquessuch asCTA(ComputedTomographicAngiography),DSA(DigitalSubtractionAngiography)and 3DRA (Three-DimensionalRotationalAngiography)[64]. In these proceduresa small part ofthebrainisscanned,andaneurysmsevenofasizelessthan3mmcanbeobserved[8,95]. These techniques allow a reconstruction of three-dimensional arteries and aneurysms and hence an approximate identification of the blood vessels and parts of the soft tissue in the scannedvolume. ComputationalFluidDynamics.ThetremendouspotentialofCFDinsupportingmedi- caldecisionsandproposingtherapeuticoptionsinthetreatmentofcerebralaneurysms,was already anticipatedin [57]. The value of numericalsimulationsfor treating aneurysmswill likely increase further with better quantitative understanding of hemodynamicsin cerebral bloodflow.AprimarychallengeforanyCFDmethod,whetheritisabody-fittedmethod[35] oranIBmethod[37,62,73],istocapturetheflownearsolid-fluidinterfaces.Inthisregion thehighestvelocitygradientsmayoccur,leadingtocorrespondinglyhighestlevelsofshear stress, but also potentially highest levels of numerical error. In methods employing body- fittedgrids,thequalityofpredictionsisdirectlylinkedtothedegreetowhichgrid-linescan beorthogonaltothesolid-fluidinterfaceandtoeachother.Also,variationinlocalmeshsizes andshapesofadjacentgridcellsisafactordeterminingnumericalerror.Thegenerationofa suitablegridisfurthercomplicatedastherawdatathatdefinetheactualaneurysmgeometry
Description: