Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M....

28
Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 06/23/22 1 Decision Sciences, San Diego, April 2010 Presented at NSS-MIC 2009 Orlando

Transcript of Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M....

Page 1: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Simulation Study of Muon Scattering For Tomography Reconstruction

D. Mitra

A. Banerjee

K. Gnanvo

M. Hohlmann

Florida Institute of Technology

04/21/23 1Decision Sciences, San Diego, April 2010

Presented at NSS-MIC 2009 Orlando

Page 2: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Muon ScatteringMuon Scattering

Scattering angleScattering angle Scattering function Scattering function

distribution: Approx. Normaldistribution: Approx. Normal (Bethe 1953)(Bethe 1953)

Lrad

H

cp

MeV

15

rad

radLp

L115

2

0

Heavy tail over Gaussian

04/21/23 2Decision Sciences, San Diego, April 2010

milirad 2 /cm

Page 3: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Types of Tomography• Emission tomography:

• SPECT• PET• MRI

• Transmission tomography• X-ray• Some Optical

• Reflection• UltraSound• Total Internal Reflection Fluoroscopy (TIRF)

• Scattering/ DiffusionMuon tomography• Some Optical (IR) tomography

04/21/23 3Decision Sciences, San Diego, April 2010

Page 4: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Experiment

• GEANT4 simulation with partial physics for scattering

• Large array of Gas Electron Multiplier (GEM)

detector is being built • IEEE NSS-MIC’09 Orlando Poster# N13-246

04/21/23 4Decision Sciences, San Diego, April 2010

Page 5: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Reconstruction Algorithms

Point of Closest Approach (POCA) Purely geometry based Estimates where each muon is scattered

Max-Likelihood Expectation Maximization for Muon Tomography

Introduced by Schultz et al. (at LANL) More physics based-model than POCA Estimates Scattering density (λλ) per voxel

04/21/23 5Decision Sciences, San Diego, April 2010

Page 6: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POCA Concept

Incoming ray

Emerging ray

POCA

3D

Three detector-array above and three below

04/21/23 6Decision Sciences, San Diego, April 2010

Page 7: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POCA Result ≡ processed-Sinogram?

AlFe

Pb

UW

Θ

40cmx40cmx20cm Blocks (Al, Fe, Pb, W, U)

Unit: mm

04/21/23 7Decision Sciences, San Diego, April 2010

Page 8: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POCAPOCA

Pro’sPro’s Fast and efficientFast and efficient Accurate for simple Accurate for simple

scenario’sscenario’s Con’sCon’s

No Physics: multi-No Physics: multi-scattering ignoredscattering ignored

DeterministicDeterministic

Unscattered tracks Unscattered tracks are not usedare not used

04/21/23 8Decision Sciences, San Diego, April 2010

Page 9: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

ML-EM System MatrixML-EM System Matrix

Voxels following POCA track

x

L

T

Dynamically built for each data set

04/21/23 9Decision Sciences, San Diego, April 2010

Page 10: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

ML-EM AlgorithmML-EM Algorithm(adapted from Schultz et al., TNS 2007, & Tech Reports LANL)(adapted from Schultz et al., TNS 2007, & Tech Reports LANL)

(1)(1) gather data: (gather data: (ΔΘΔΘ, , ΔΔ, p): scattering angles, linear displacements, , p): scattering angles, linear displacements, momentum valuesmomentum values

(2)(2) estimate track-parameters (L, T) for all muonsestimate track-parameters (L, T) for all muons

(3)(3) initialize initialize λλ (arbitrary small non-zero number, or…) (arbitrary small non-zero number, or…)

(4)(4) for each iteration k=1 to I (or, until for each iteration k=1 to I (or, until λλ stabilizes) stabilizes)

(1)(1) for each muon-track i=1 to Mfor each muon-track i=1 to M

Compute CCompute Cijij

(2) for each voxel j=1 to N(2) for each voxel j=1 to N

// M// Mjj is # tracks is # tracks

(5) return (5) return λλ

0:

2 1)(

ijLi

ijold

jold

jnew

j CMj

04/21/23 10Decision Sciences, San Diego, April 2010

Page 11: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

ML-EM Reconstruction

• Slow for complex scenario

• Our implementation used some 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.]

04/21/23 11Decision Sciences, San Diego, April 2010

Page 12: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POCA Result for a vertical clutter

04/21/23 12Decision Sciences, San Diego, April 2010

Page 13: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Slabbing Concept

04/21/23 Decision Sciences, San Diego, April 2010 13

Slabbing Slice3cm thick

Page 14: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

“Slabbing” studies with POCA: Filtered tracks with DOCA (distance of closest approach)

Ev: 10MilVertical stack: Al-Fe-W: 50cm50cm20cm, Vert. Sep: 10cm

Slab size: 3 cm

04/21/23 14Decision Sciences, San Diego, April 2010

Page 15: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POClust Algorithm: clustering POCA points

04/21/23 Decision Sciences, San Diego, April 2010 15

Input: Geant4 output (list of all muon tracks and associated parameters)

1. For each Muon track {1. For each Muon track { 2.2. Calculate the POCA pt Calculate the POCA pt P P and its scattering-angle and its scattering-angle 3. 3. if ( if (PP lies outside container) continue; lies outside container) continue; 4.4. Normalize the scattering angle (angle*p/3GeV). Normalize the scattering angle (angle*p/3GeV). 5.5. CC = Find-nearest-cluster-to-the (POCA pt = Find-nearest-cluster-to-the (POCA pt PP);); 6.6. Update-cluster Update-cluster CC for the new pt for the new pt PP; ; 7. After a pre-fixed number of tracks remove sporadic-clusters;7. After a pre-fixed number of tracks remove sporadic-clusters; 8. 8. Merge close clusters with each-other } Merge close clusters with each-other } 9. Update 9. Update λλ (scattering density) of each cluster (scattering density) of each cluster C C using straight using straight tracks passing through tracks passing through CC

Output: A volume of interest (VOI)

Page 16: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POClust essentials

04/21/23 Decision Sciences, San Diego, April 2010 16

• Not voxelized, uses raw POCA points

•Three types of parameters:• Scattering angle of POCA point

• Normalized “proximity” of the point to a cluster

• how the “quality” of a cluster is affected by the new poca point andmerger of points or clusters

• Real time algorithm: as data comes in

Page 17: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POClust Results

04/21/23 17Decision Sciences, San Diego, April 2010

G4 Phantom

Page 18: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Three target vertical clutter scenario

04/21/23 Decision Sciences, San Diego, April 2010 18

Al-Fe-W: 40cm*40cm*20cm 100cm gap

Al

Fe

W

AlFe

W

Page 19: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Three target vertical clutter scenario:Smaller gap

04/21/23 Decision Sciences, San Diego, April 2010 19

Al-Fe-W: 40cm*40cm*20cm 10cm gap

Al

Fe

W

Page 20: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POClust Results: Reverse Vertical Clutter

04/21/23 20Decision Sciences, San Diego, April 2010

Al

U

Pb

Page 21: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POClust Results

04/21/23 21Decision Sciences, San Diego, April 2010

Page 22: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Why POClust & Not just POCA visualization?

• Quantitate: ROC Analyses

• Improve other Reconstruction algorithms with a Volume of Interest (VOI) or

Regions of Interest (ROI)

• Why any reconstruction at all?POCA visualization is very noisy in a

complex realistic scenario

04/21/23 22Decision Sciences, San Diego, April 2010

Page 23: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Additional works with POClust

1. Clustering provides Volumes of Interest (VOI) inside the container: Run ML-EM over only VOI for better precision and efficiency

2. Slabbing, followed by Clustering

3. Clusters growing over variable-sized hierarchical voxel tree, followed by ML-EM

4. Automated cluster-parameter

selection by optimization

5. Use cluster λ λ values in a Maximum

A Posteriori –EM, as priors (Wang

& Qi: N07-6)

04/21/23 23Decision Sciences, San Diego, April 2010

Page 24: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

POClust as a pre-processor

04/21/23 Decision Sciences, San Diego, April 2010 24

Volume of Interest reduces after Clustering:

A minimum bounding box(235cm X 235cm X 45cm)

Initial Volume of Interest (400cm X 400cm X 300cm)

Page 25: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

Scenario: 5 targets VOI : 400X400X300 cm3

Iterations: 50

EM after pre-processing with POClust

Targets: Uranium (100,100,0), Tangsten (-100, 100, 0)

W

U

04/21/23 25

Page 26: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

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 cmVoxel Size= 5 X 5 X 5 cm#Voxels = 384000

After Clustering, VOI reduces, #Voxels = 18330

Results From EM over POClust generated VOI

04/21/23 26

Page 27: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

A human in muon! …not on moon,

again, yet …

04/21/23 Decision Sciences, San Diego, April 2010 27

Twenty million tracksIn air background130cmx10cmx10cm Ca slab inside150cmx30cmx30cm H2O slab

GEANT4 Phantom

Page 28: Simulation Study of Muon Scattering For Tomography Reconstruction D. Mitra A. Banerjee K. Gnanvo M. Hohlmann Florida Institute of Technology 12/4/20151Decision.

04/21/23 Decision Sciences, San Diego, April 2010 28

Thanks!

Debasis Mitra

[email protected]

Acknowledgement:Department of Homeland Security

National Science Foundation& many students at FIT