Department Of Physics, Northern Illinois University, And .

1m ago
2.66 MB
29 Pages

A Low-dose, Accurate MedicalImaging Method for Proton Therapy:Proton Computed TomographyBela ErdelyiDepartment of Physics, Northern Illinois University,and Physics Division, Argonne National LaboratoryFFAG’09, Fermilab, Batavia, September 21-25, 2009

Acknowledgments Northern Illinois University– Physics: G. Coutrakon, V. Rykalin, K. Wong– Computer Science: N. Karonis, K. Duffin, K.Naglich, J. Panici Loma Linda University Medical Center– R. Schulte, V. Bashkirov, F. Hurley, S. Penfold Santa Cruz Institute for Particle Physics– H. Sadrozinski, M. Petterson, N. Blumenkrantz,B. Colby, J. Feldt, J. Heimann, R. Johnson, D.Lucia, D. C. WilliamsSeptember 21-25, 2009Proton Computed Tomography2

Outline The main ideaA little historyThe fundamentalsCurrent statusSummarySeptember 21-25, 2009Proton Computed Tomography3

The Main Idea For proton therapy, one positions theBragg peak onto the tumor For pCT, raise the initial energy so protonstraverse the object to be imaged Measure the phase space data of eachproton individually Use this data to construct the electrondensity map of the object traversed byprotons Use the resulting electron density map fordiagnosis, proton therapy treatment plan,adaptive treatment, positioningverification, etc.September 21-25, 2009Proton Computed Tomography4

Motivation: Range UncertaintiesSeptember 21-25, 2009Proton Computed Tomography5

OverviewSeptember 21-25, 2009Proton Computed Tomography6

Advantages and Benefits Practically eliminates range uncertainties,therefore allowing very accurate andprecise proton treatment plans Provides fast patient positioningverification and adaptive treatments, ifnecessary Achieves a reduced dose necessary forimaging relative to XCT Provides a quantification of the rangeuncertainties as a function of tumor site,type, etc. that will be useful to any protontherapy facility in operation that lacks apCT system.September 21-25, 2009Proton Computed Tomography7

History of pCT (1)September 21-25, 2009Proton Computed Tomography8

History of pCT (2) In the late 1970s, Ken Hanson (LANL) andKramer et al. (ANL) experimentally explored theadvantages of pCT and proton radiography They pointed out the dose reduction w.r.t. XCTand the problem of limited spatial resolution dueto proton scattering In the 1990s Ron Martin (ANL) proposed buildinga proton CT system using a scanning beamproton gantry During the 1990s Uwe Schneider (PSI) furtherdeveloped the idea of proton radiography as atool for quality control in proton therapy In the late 1990s Piotr Zygmanski (MGH) HarvardCyclotron tested a cone beam CT system withprotons pCT Collaboration (2003- )September 21-25, 2009Proton Computed Tomography9

The Fundamentals (1)The Bethe-Bloch formula gives the mean energy loss rateof protons in a mediumdE ( r) ηe ( r ) F (I ( r) , E ( r ))dl· µ¶ 2212me cβ (E)2ln βF (I ( r ) , E ( r )) K 2(E)2β (E)I ( r ) 1 β (E)Z LZ E(L)dEηe ( r ) dl F (I ( r ) , E ( r ))0E0 I( r ) 7 Iwater rhs is a numerical integration bi , i 1, m.Discretize ηe ( r) over some basis functions (typically voxels) xi , i 1, n.Each proton will determine a linear equation in the variables xiA x bSeptember 21-25, 2009Proton Computed Tomography10

The Fundamentals (2)A x bDetermined byproton pathsNot known exactlydue to MCSElectron densitiesFrom measurementsDetermined by averageproton energy loss over pathsNot known exactlydue to energy lossstragglingA(ξ) x(ξ) b(ξ)Random realization with ξ drawn from a probability distributionSeptember 21-25, 2009Proton Computed Tomography11

The Most Likely PathDeterministic systemsknownTransfer mapcomputedfixedSeptember 21-25, 2009Proton Computed Tomography12

The Most Likely PathStochastic systemsknownTransfer mapknowncomputedSeptember 21-25, 2009Proton Computed Tomography13

Spatial Resolution (1)measure ofof ourour abilityability toto predictpredict eacheach individualindividual proton’sproton’s trajectorytrajectoryAA measureinside thethe objectobject toto bebe imagedimaged isis thethe spatialspatial resolutionresolutioninsideMultiple Coulomb Scattering (MCS)r·µ ¶ M eV13.6 cll1 0.038 lnσ (l, E) β (E) p (E) XXExample: 200 MeV protons in20cm of water haveσ 39mrad - σlat 3.5mmUse constrains to reduce uncertainty: position, direction, energySeptember 21-25, 2009Proton Computed Tomography14

Spatial Resolution (2) Developed new formalism to include energy as a constraintEquivalently, it fixes the trajectory lengthExample: 2 protons, with exactly the same incoming energy,position, direction, and outgoing position and direction – butdifferent outgoing energyPrevious MLP formalism gives the same MLP, new one is differentdue to the different path lengthsAlso implies improved spatial resolution – difficult to computeanalyticallySeptember 21-25, 2009Proton Computed Tomography15

Electron Density Resolution (1)A(ξ) x(ξ) b(ξ)x fromx(ξ)ismeasure ofof ourour abilityability toto predictpredict from thethe randomrandom vectorvector isAA measurethe electronelectron densitydensity resolutionresolutiontheDefinition:Definition:σx sPni 02σxi (ξ)nσhEout i kAσx g vF ( hEout i) mvSeptember 21-25, 2009Proton Computed Tomography16

Electron Density Resolution (2)September 21-25, 2009Proton Computed Tomography17

Reconstruction MethodsProjection MethodsBasic property: To reach any goal that is related to thewhole family of sets by performing projections onto theindividual sets.Basic ability: To handle huge-size problems whosedimensions are beyond the capabilities of current, moresophisticated, methods.September 21-25, 2009Proton Computed Tomography18

Algebraic Reconstruction TechniquexkH4H5H3H2H1September 21-25, 2009x k 1Proton Computed Tomography19

Block Iterative ProjectionB1 (1, 2,3)B2 (4,5, 6)H6xkH5H4x k 1H3H2H1September 21-25, 2009Proton Computed Tomography20

String AveragingI1 (1,3,5, 6)I 2 (2)I 3 (6, 4)xkH6H5H4x k 1H3H2H1September 21-25, 2009Proton Computed Tomography21

Hardware Implementation Desktop/laptopCompute ClustersGP-GPUGP-GPU clustersHybrid: multi-core CPUs GP-GPUs (clusters)Speedup of the rel. electron densitycalculation when performed with aNVIDIA GTX280 GPU relative to aIntel Q6600 quad core CPU (ScottMcAllister, Master’s Thesis, Cal StateSB, 2009)September 21-25, 2009Proton Computed Tomography22

GEANT4 SimulationsSeptember 21-25, 2009Proton Computed Tomography23

Reconstruction Results FromSimulations1 cycle5 cycles10 cyclesBICAVSeptember 21-25, 2009DROPProton Computed TomographyOS-SARTCARP24

Reconstruction Results FromReal DatapCT prototypeprototypepCTpCT phantomphantompCTSeptember 21-25, 2009Proton Computed Tomography25

The pCT System Prototype SchematicReady by early 2010September 21-25, 2009Proton Computed Tomography26

System ComponentsSeptember 21-25, 2009Proton Computed Tomography27

Next PhaseSeptember 21-25, 2009Proton Computed Tomography28

Summary pCT is a new medical imaging method that willgreatly benefit proton therapy in general and willoffer a low-dose diagnostic imaging modality The project is truly interdisciplinary involvingphysics, mathematics, computer science andengineering The NIU-LLUMC-SCIPP Collaboration is wellunderway; the first pCT prototype systemcapable of imaging head-sized objects will beready by early 2010 Further work is necessary towards a fully clinicaloperation-ready pCT systemSeptember 21-25, 2009Proton Computed Tomography29