Simulating CT imaging with GATE - Indico

Simulating CT gimaging with GATE

F.Cassol BrunnerimXgam group, http://imxgam.in2p3.fr

Centre de Physique des Particules de Marseille (France)

Simulating CT imaging with GATE

PIXSCAN, scanner for small animalsGATE simulationdata simulation comparisondata – simulation comparisonConclusionOutlooks

F.Cassol Brunner (CPPM/IN2P3) 212th Geant4 Collaboration Workshp, 13/9 2007

PIXSCAN

Single photon counting scanner for small animals g p gbased on hybrid pixels

Low noiseEnergy thresholdVery high signal dynamicVery fast acquisition


PIXSCAN prototypepixels: 330 x 330 µm²pixels: 330 x 330 µm²8 tiled modules (9˚ tilt )total size: 6.8 x 6.5 cm²total size: 6.8 x 6.5 cm206 x 200 pixelssensor: Silicon


SIMULATIONSource: Mo anode, 50 kV tube, 0.5 mA Filter: 6 mm Al

700

Spectre de la source

500

600

200

300

400

0

100

200


Energie (keV)10 15 20 25 30 35 40 45 50 55 600

Energy [keV]

SIMULATIONDetector geometry: tiled modules


SIMULATIONDigitalization: E > 15 keVDead pixels added afterwards (~25%)Dead pixels added afterwards (~25%)


SIMULATION

Output: 360 projections as for data

CPU Time:CPU Time:360 jobs x 2 days: 1000 counts/pixels

Data:PIXSCAN data

Data:3000-6000 counts/pixels


Data – Simulation comparison

Spatial resolutionImage noiseGeometry calibrationGeometry calibrationCT linearityTomography's distortions


Spatial resolutionPVC phantom R 12 mmR = 12 mm

derivation DFT


Spatial resolution: DATASpatial resolution: DATA

LSF FWHM ~ 500 µm µ

Wh b d?Why so bad?


Spatial resolution: FLAT modulesSpatial resolution: FLAT modules

FWHM ~ 160 µm Raw datadi tl t t d

µdirectly reconstructed


Spatial resolution: TILED modulesSpatial resolution: TILED modules

FWHM ~ 160 µm

FWHM ~ 350 µm

µ

AddedFWHM ~ 350 µm image rebinning


Spatial resolution: + DEAD PIXELSSpatial resolution: + DEAD PIXELS

FWHM ~ 160 µmµ

FWHM ~ 350 µm

FWHM ~500 µm

FWHM ~ 350 µm

AddedFWHM ~500 µm pixels interpolation


For 3000 ph/pixel:Noise

For 3000 ph/pixel:Data = 7.1± 0.1 %GATE = 7 1± 0 4 %GATE = 7.1± 0.4 %

GGATE

simmulation

100% ×= cyl

CTCTNoise

σ


− aircyl CTCT

Geometry

GATE used to validate our calibration procedure

Several parameters to be estimated

GATE used to validate our calibration procedure


GeometryGeometryGATE i t C lib ti t tGATE input

xp = -33.990 mmyp = 34 712 mm

Calibration outputxp = -34.344 +/- 5.447 mmyp = 34 305 +/ 0 437 mmyp = -34.712 mm

zp = 284.000 mm alpha = 0zt = 135 000 mm

yp = -34.305 +/- 0.437 mmzp = 284.397 +/- 5.497 mmalpha = -0.0001 +/- 0.001 radzt = 135 11 +/ 2 585 mmzt = 135.000 mm zt = 135.11 +/- 2.585 mm


CT linearitySimulation of a linearity phantom:

Base = epoxy resin (1 15 g/cm³)Base = epoxy resin (1.15 g/cm³)

Inserts = calcium hydroxyapatite )()( 345 OHPOCa

Insert ρ HA g/cm³

ρ tot g/cm³g/cm g/cm

5 0.00 1.154 0.05 1.16

3 0.20 1.262 0.80 1.66


1 1.20 1.86data

CT linearity v

alue

0.14

/ ndf 2χ 0.0963 / 3

p0 0.003545± 0.06213

p1 0.002636± -0.001063

/ ndf 2χ 0.0963 / 3

p0 0.003545± 0.06213

p1 0.002636± -0.001063

CT Linearity

CT v

a

0.1

0.12PIXSCAN Data

GATE

0.08

0.1

0.04

0.06

Simulation ok but data are not linear

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20

0.02

Simulation ok, but data are not linear.Concentrations not correct?


]-1attenuation coefficient [cm0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

]-1attenuation coefficient [cm0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 20

Tomography's distortionsTomography s distortions

Simulation of a phantom after Defrise

Data Gate


Tomography's distortionsTomography s distortions v

alu

e0.09

0.1

Data

Phantom profile

Phantom profile well reproducedC

T v

0.05

0.06

0.07

0.08

0.09

Data a to p o e e ep oduced

0.01

0.02

0.03

0.04

mm0 5 10 15 20 25 30

mm0 5 10 15 20 25 300

val

ue

0.09

0.1

Phantom profile

CT

v

0.05

0.06

0.07

0.08

0.09

GATE

0.01

0.02

0.03

0.04


mm0 5 10 15 20 25 30

mm0 5 10 15 20 25 300

ConclusionGATE could reproduce several figures of merit of our CT scanner:

Spatial resolutionSpatial resolutionNoiseLinearity (for low densities)Tomography’s distortions

The geometry calibration procedure could be validated

Weak points of our image treatment could be identified (image rebinning)Weak points of our image treatment could be identified (image rebinning)


OutlooksGATE will help us in the following analysis:

Optimization of the next PIXSCAN geometry : new chip with pixels of 130 x 130 µm²

Dose estimationsContrast studies based on simulated animal phantomsContrast studies based on simulated animal phantomsDesign of a simultaneous CT-PET scanner


Simulating CT imaging with GATE - Indico

Documents