Simulated vascular reconstruction in a virtual operating theatre
description
Transcript of Simulated vascular reconstruction in a virtual operating theatre
Simulated vascular Simulated vascular reconstructionreconstruction
in a virtual operating theatrein a virtual operating theatre
Robert G. Belleman, Peter M.A. SlootRobert G. Belleman, Peter M.A. Sloot
Section Computational Science, University of Section Computational Science, University of Amsterdam,Amsterdam,
Kruislaan 403, 1098 SJ Amsterdam, the Netherlands.Kruislaan 403, 1098 SJ Amsterdam, the Netherlands.
Email:Email: (robbel|sloot)@science.uva.nl(robbel|sloot)@science.uva.nl
OverviewOverview
Interactive distributed simulationInteractive distributed simulation– Simulated vascular reconstruction in Simulated vascular reconstruction in
a virtual environmenta virtual environment Performance issuesPerformance issues
– High speed simulationHigh speed simulation– High speed communicationHigh speed communication– Responsive interactive explorationResponsive interactive exploration
Time/location independenceTime/location independence
Interactive distributed Interactive distributed simulationsimulation
Human in the loop Human in the loop experimentationexperimentation– interactive exploration of interactive exploration of
data/parameter spacesdata/parameter spaces
Performance issuesPerformance issues
Asynchronous, pipelined Asynchronous, pipelined configuration configuration – Distributed to exploit specialised Distributed to exploit specialised
resourcesresources– To increase performance:To increase performance:
decrease component execution timesdecrease component execution times shorten delaysshorten delays
Vascular reconstructionVascular reconstruction
““Traditional” treatmentTraditional” treatment
Vascular reconstructionVascular reconstruction
Angio w/ Fem-Fem &Fem-Pop
AFB w/ E-S Prox.
Anast.
Angio w/Fem-Fem
AFB w/ E-E Prox.
Anast.
Preop
(Figure courtesy C.A. Taylor, Stanford University)
Simulated vascular Simulated vascular reconstructionreconstruction
Simulated treatment planningSimulated treatment planning
Simulated vascular Simulated vascular reconstructionreconstruction
Components:Components:– blood flow simulationblood flow simulation– visualization in a virtual environmentvisualization in a virtual environment– interaction with simulation and interaction with simulation and
visualization (treatment planning)visualization (treatment planning) Requirements:Requirements:
– Interactive system; Interactive system; fast response, fast fast response, fast update rateupdate rate
Fluid flow simulationFluid flow simulation Lattice Boltzmann Method (LBM)Lattice Boltzmann Method (LBM)
– Lattice based particle methodLattice based particle method Regular lattice, similar to CT or MRI datasetsRegular lattice, similar to CT or MRI datasets Allows irregular 3D geometryAllows irregular 3D geometry Allows changes at run-timeAllows changes at run-time
Velocity, pressure and shearVelocity, pressure and shearstress calculated fromstress calculated fromparticle densitiesparticle densities
Non-compressibleNon-compressiblehomogeneous fluid,homogeneous fluid,laminar flowlaminar flow
Spatial and temporal localitySpatial and temporal locality– Ideal for parallelIdeal for parallel
implementationimplementation
PreprocessingPreprocessing Segmentation of patient specific MRA/CTA scanSegmentation of patient specific MRA/CTA scan
– Isolates region of interestIsolates region of interest Lattice Boltzmann grid generationLattice Boltzmann grid generation
– Defines solid and fluid nodes, inlet and outlet conditionsDefines solid and fluid nodes, inlet and outlet conditions
Interactive exploration in Interactive exploration in VRVR
Visualize simulation resultsVisualize simulation results– Flow field, pressure, shear stressFlow field, pressure, shear stress– Real timeReal time
Interactive explorationInteractive exploration– VR interaction to locateVR interaction to locate
regions of interestregions of interest Interactive grid editingInteractive grid editing
– Simulate vascularSimulate vascularreconstruction procedurereconstruction procedure
Interactive exploration in Interactive exploration in VRVR
Visualize simulation resultsVisualize simulation results– Flow field, pressure, shear stressFlow field, pressure, shear stress– Real timeReal time
Interactive explorationInteractive exploration– VR interaction to locateVR interaction to locate
regions of interestregions of interest Interactive grid editingInteractive grid editing
– Simulate vascularSimulate vascularreconstruction procedurereconstruction procedure
Parallel fluid flow simulationParallel fluid flow simulation
Performance indication:Performance indication:
Communication delayCommunication delay
10-100Mb of data per iteration10-100Mb of data per iteration– Velocity, pressure, shear stressVelocity, pressure, shear stress– May take seconds to transfer May take seconds to transfer
Increasing communication Increasing communication throughputthroughput
1.1. Latency hidingLatency hiding DecreaseDecrease latency latency: : faster response timefaster response time, ,
increased updateincreased update rate rate
2.2. Payload reductionPayload reduction Less dataLess data:: shorter transfer times shorter transfer times
Latency hidingLatency hiding
Multiple connectionsMultiple connections– Waiting for acknowledgements is Waiting for acknowledgements is
hiddenhidden– Packets can travel through different Packets can travel through different
routesroutes– Uses CAVERNUses CAVERN
Payload reductionPayload reduction
Data encodingData encoding– Decreases level of detailDecreases level of detail
Accuracy determined by type of representationAccuracy determined by type of representation
– Lossy compressionLossy compression– InducesInduces latency latency
““Standard” compression librariesStandard” compression libraries (zlib) (zlib)– Reduction to 10% not uncommonReduction to 10% not uncommon– Lossless compressionLossless compression– InducesInduces latency latency
Communication pipelineCommunication pipeline
Cascade of encoding, compression Cascade of encoding, compression and multiple socket connectionsand multiple socket connections
Pipeline performancePipeline performance
Latency hidingLatency hiding:: Payload reductionPayload reduction::
The Virtual LaboratoryThe Virtual Laboratory Shared use of distributed computing resources:Shared use of distributed computing resources:
high performance computers, scanners, algorithms, etc.high performance computers, scanners, algorithms, etc.– Connected via high performance networksConnected via high performance networks– Common infrastructure: the Common infrastructure: the Virtual LaboratoryVirtual Laboratory
Multi-disciplinary scientific experimentationMulti-disciplinary scientific experimentation– Problem solving environments (PSE)Problem solving environments (PSE)– Time/location independent scientific experimentationTime/location independent scientific experimentation– Collaborative scientific researchCollaborative scientific research
For additional information...
DutchGrid initiativeDutchGrid initiative : : http://www.dutchgrid.nl/http://www.dutchgrid.nl/
VLAM-G VLAM-G : : http://www.http://www.dutchgriddutchgrid..nlnl//VLAM-G/VLAM-G/
Experiment definitionExperiment definition Simulated vascular reconstructionSimulated vascular reconstruction
– Patient specific angiographyPatient specific angiographydatadata
– Fluid flow simulationFluid flow simulationsoftwaresoftware
– Simulation of reconstructiveSimulation of reconstructivesurgical procedure in VRsurgical procedure in VR
– Interactive visualization ofInteractive visualization ofsimulation results in VRsimulation results in VR
Pre-operative planningPre-operative planning– Explore multiple reconstructionExplore multiple reconstruction
proceduresprocedures
Where are we going?Where are we going?
Improve blood flow simulationImprove blood flow simulation High performance visualization and High performance visualization and
renderingrendering Time/location(Time/location(/device?)/device?) independent independent
interactive simulationinteractive simulation– Access to high performanceAccess to high performance
computing resources computing resources overoverlowlow bandwidth connections bandwidth connections
– Use aUse anywhere, at any timenywhere, at any time……onon anythinganything??
Where are we going?Where are we going?
Improve blood flow simulationImprove blood flow simulation High performance visualization and High performance visualization and
renderingrendering Time/location(Time/location(/device?)/device?) independent independent
interactive simulationinteractive simulation– Access to high performanceAccess to high performance
computing resources computing resources overoverlowlow bandwidth connections bandwidth connections
– Use aUse anywhere, at any timenywhere, at any time……onon anythinganything??
Where are we going?Where are we going?
Improve blood flow simulationImprove blood flow simulation High performance visualization and High performance visualization and
renderingrendering Time/location(Time/location(/device?)/device?) independent independent
interactive simulationinteractive simulation– Access to high performanceAccess to high performance
computing resources computing resources overoverlowlow bandwidth connections bandwidth connections
– Use aUse anywhere, at any timenywhere, at any time……onon anythinganything??
Where are we going?Where are we going?
Improve blood flow simulationImprove blood flow simulation High performance visualization and High performance visualization and
renderingrendering Time/location(Time/location(/device?)/device?) independent independent
interactive simulationinteractive simulation– Access to high performanceAccess to high performance
computing resources computing resources overoverlowlow bandwidth connections bandwidth connections
– Use aUse anywhere, at any timenywhere, at any time……onon anythinganything??
Thanks!Thanks! UvA, Section Computational ScienceUvA, Section Computational Science
– Sloot, Hoekstra, Zhao, Artoli, Merks, Shulakov, ShamoninSloot, Hoekstra, Zhao, Artoli, Merks, Shulakov, Shamonin Leiden University Medical Center (LUMC)Leiden University Medical Center (LUMC)
– LKEB (Reiber, vd Geest, Koning, Schaap)LKEB (Reiber, vd Geest, Koning, Schaap) Stanford UniversityStanford University
– Biomedical department (Zarins, Taylor)Biomedical department (Zarins, Taylor) SARA Computing and Networking ServicesSARA Computing and Networking Services
Related work:Related work: Lattice Boltzmann flow simulation: Artoli, Hoekstra (UvA/SCS)Lattice Boltzmann flow simulation: Artoli, Hoekstra (UvA/SCS) Segmentation of MRA images: Koning, Schaap (LUMC)Segmentation of MRA images: Koning, Schaap (LUMC) Agent based solutions to interactive systems: Zhao (UvA/SCS)Agent based solutions to interactive systems: Zhao (UvA/SCS)