Simulation Study of Muon Scattering For Tomography Reconstruction - PowerPoint Presentation

Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra, A. Banerjee, S. White, S. Waweru, R. Hoch K. Gnanvo M. Hohlmann Florida Institute of Technology 10/27/2009 1 IEEE NSS-MIC 2009, Orlando, FL

Co-ordinates Where are we?10/27/2009 2 IEEE NSS-MIC 2009, Orlando, FL

Cosmic Ray-generated Muons more massive cousin of electron produced by cosmic ray decay at rate 1 /cm 2/minhighly penetrating, long half-life affected by Coulomb force 10/27/2009 3 IEEE NSS-MIC 2009, Orlando, FL

Muon Tomography Concept 10/27/2009 4 IEEE NSS-MIC 2009, Orlando, FL

Muon Scattering Scattering angle Scattering function distribution: Approx. Normal (Bethe 1953) Heavy tail over Gaussian 10/27/2009 5 IEEE NSS-MIC 2009, Orlando, FL

Cosmic-ray generated Muon generated by proton and upper atmosphere’s interaction median at about 3 Gev Peaks at about 30 degree10/27/2009 6 IEEE NSS-MIC 2009, Orlando, FL

Physics behind Models Emission tomography: SPECTPETMRI Transmission tomography X-raySome Optical Reflection Ultra Sound Total Internal Reflection Fluoroscopy (TIRF) Scattering/ Diffusionmuon tomography some Optical10/27/2009 7 CS Seminar, FIT

Reconstruction Algorithms Point of Closest Approach (POCA) Purely geometry based Estimates where each muon is scattered Max-Likelihood Expectation Maximization Introduced by Schultz et al. (at LANL) More physics based than POCA Estimates Scattering density ( λ ) per voxel 10/27/2009 8 IEEE NSS-MIC 2009, Orlando, FL

POCA Concept Incoming ray Emerging ray POCA 3D Three GEM detector-array above and three below: http://www.lnf.infn.it/esperimenti/imagem/ 10/27/2009 9 IEEE NSS-MIC 2009, Orlando, FL

POCA Result ≡ processed-Sinogram Al Fe Pb U W Θ 40cmx40cmx20cm Blocks (Al, Fe, Pb, W, U) Unit: mm 10/27/2009 10 IEEE NSS-MIC 2009, Orlando, FL

POCA Discussion Pro’s Fast and efficient Accurate for simple scenario’s Con’s No Physics: multi-scattering ignored Deterministic Unscattered tracks are not used 10/27/2009 11 IEEE NSS-MIC 2009, Orlando, FL

ML-EM System Matrix Voxels following POCA track L T Dynamically built for each data set 10/27/2009 12 IEEE NSS-MIC 2009, Orlando, FL

ML-EM Algorithm (adapted from Schultz et al., TNS 2007, & Tech Reports LANL) gather data: ( ΔΘ , Δ, p): scattering angles, linear displacements, momentums estimate track-parameters (L, T) for all muons initialize λ (arbitrary small non-zero number) for each iteration k=1 to I (or, until λ stabilizes) for each muon-track i=1 to M Compute C ij (2) for each voxel j=1 to N // M j is # tracks (5) return λ 10/27/2009 13 IEEE NSS-MIC 2009, Orlando, FL

Experiment GEANT4 simulation with partial physics for scattering (Gas Electron Multiplier detector is being built) 10/27/2009 14 IEEE NSS-MIC 2009, Orlando, FL

MLEM Reconstruction Very slow for complex scenario A computing cluster (Grid Tier 3) is available, but our programs run on a single node Reconstruction used smart data structure for speed and better memory usage [In ‘Next Generation Applied Intelligence’ (Springer Lecture Series in Computational Intelligence: 214), pp. 225-231, June 2009.]10/27/2009 15 IEEE NSS-MIC 2009, Orlando, FL

Slabbing Concept10/27/2009 IEEE NSS-MIC 2009, Orlando, FL16 Slabbing Slice

“Slabbing” studies with POCA: Filtered tracks DOCA (distance of closest approach) Ev: 10Mil, Vertical stack: Al-Fe-W: 50cm50cm20cm, Vert. Sep: 10cmSlab size: 3 cm 10/27/2009 17 IEEE NSS-MIC 2009, Orlando, FL

POClust Algorithm: clustering POCA points 10/27/2009IEEE NSS-MIC 2009, Orlando, FL 18 Input: Geant4 output (list of all muon tracks and associated parameters) 1. For each Muon track { 2. Calculate the POCA pt P and scattering-angle (Vossman/3-D regression). 3. if ( P lies outside container) continue; 4. Normalize the scattering angle (angle*p/3GeV). 5. C = Find-nearest-cluster-to-the (POCA pt P ); 6. Update-cluster C for the new pt P ; 7. After a pre-fixed number of tracks remove sporadic-clusters; 8. Merge close clusters with each-other } 9. Update λ (scattering density) of each cluster C using straight tracks passing through C Output: A volume of interest (VOI)

Scenario 1 Geometry Five 40cmx40xcmx20cm Boxes 10/27/2009 19 IEEE NSS-MIC 2009, Orlando, FL

POClust Results Medium: Air U,W,Pb,Fe,Al Size: 40X40X20cm 10/27/2009 20 IEEE NSS-MIC 2009, Orlando, FL

Three target scenario geometries10/27/2009 IEEE NSS-MIC 2009, Orlando, FL21 Al-Fe-W: 40cm*40cm*20cm 1m separation gap Al-Fe-W: 40cm*40cm*20cm 10cm separation gapAlFe W Al FeW

POClust Results: vertical clutter Medium: Vacuum Al-Fe-W: 100cm vertical gap 10/27/2009 22 IEEE NSS-MIC 2009, Orlando, FL Al-Fe-W Size: 50X50X20cm Separation: 10cm

POClust Results: Reverse Vertical Clutter U-Pb -Al Size:40X40X20cmGap:10cm Medium: Vacuum 10/27/2009 23 IEEE NSS-MIC 2009, Orlando, FL

POClust Results U inside Pb boxU size: 10X10X10cm Pb Box: 200X200X200 cm Thickness(Pb box): 10cm Medium: Vacuum 10/27/2009 24 IEEE NSS-MIC 2009, Orlando, FL

Why POClust & Not just POCA visualization? Quantitate: ROC Analyses Improve other Reconstruction algorithms with a Volume of Interest (VOI) orRegions of Interest (ROI) Why any reconstruction at all?POCA visualization is very noisy in a complex realistic scenario 10/27/2009 25 IEEE NSS-MIC 2009, Orlando, FL

Additional works with POClust Clustering provides Volumes of Interest (VOI) inside the container: Run MLEM over only VOI’s for better precision and efficiency Slabbing, followed by Clustering Clustering first, and then sub-dividing regions into variable-sized hierarchical voxel tree, followed by MLEM Clusters expanded over voxels Automated cluster-parameter selection by optimization 10/27/2009 26 IEEE NSS-MIC 2009, Orlando, FL

POClust as a pre-processor 10/27/2009IEEE NSS-MIC 2009, Orlando, FL 27 Volume of Interest reduces after Clustering: A minimum bounding box (235cm X 235cm X 45cm) Initial Volume of Interest (400cm X 400cm X 300cm)

Scenario: 5 targets VOI : 400X400X300 cm3 Iterations: 50EM after pre-processing with POClust Targets: Uranium (100,100,0), Tangsten (-100, 100, 0) W U

Scenario : U, W, Pb, Al, Fe placed horizontally Important Points: IGNORE ALL VOXELS OUTSIDE ROI EM COMPUTATION DONE ONLY INSIDE ROI Iterations Actual Volume (400 X 400 X 300 cm) Time taken (seconds) Clustered Volume ( 235 X 235 X 45 cm ) Time taken (seconds) 100 113.5 21.5 60 99.54 20.2 50 95.6 19.5 30 84.48 17.4 10 79.27 16.0 Here, Total Volume = 400 X 400 X 300 cm Voxel Size= 5 X 5 X 5 cm #Voxels = 384000 After Clustering, VOI reduces, #Voxels = 18330 Results From EM over POClust generated VOI

A human in muon!Not on moon, again, yet … 10/27/2009IEEE NSS-MIC 2009, Orlando, FL 31 Twenty million tracks In air background 130cmx10cmx10cm Ca slab inside150cmx30cmx30cm H2O slab GEANT4 Phantom

10/27/2009 IEEE NSS-MIC 2009, Orlando, FL 32 Thanks!Debasis Mitradmitra@cs.fit.edu Acknowledgement: Department of Homeland Security Domestic Nuclear Detection Office