psd7 conference, 13/09/05 monte carlo simulation of a coded aperture imaging with dedicated gamma...

PSD7 Conference, 13/09/05

Monte Carlo Simulation of a Coded Aperture Imaging with Dedicated

Gamma Camera System for Scintimammography

M. Alnafea1*, K. Wells1, N.M. Spyrou1 & M. Guy2

1School of Electronics & Physical Sciences, University of Surrey, Guildford, Surrey, UK

2Deptment of Nuclear Medicine, Royal Surrey County Hospital, Guildford, Surrey, UK

*E-mail:[email protected]


Outline Aim and Motivations

Geometry and Methods of Simulation

Validation

Coded Aperture CdZnTe Results

Conclusions


Breast Cancer is a leading cause of deaths in women due to cancer.

Mammography is gold standard for screening (post- menopausal) women.

Why are alternative imaging modalities needed ?

Mammography has limitation in terms of sensitivity (ability to see abnormality ~ 90 %), specificity (ability to determine if lesion is benign or malignant ~ 93 %)1.

Scintimammography can be used as an adjunct imaging modality to improve specificity -> minimise invasive secondary examinations.

This study focus on the applications of Modified Uniformly Redundant Array (MURA) coded aperture imaging methods using CdZnTe gamma camera without collimator for use in breast tumour imaging

1. Tabár L, et al. (2000) Radiol Clin North Am 38:625.651.

MOTIVATION & AIM


Why CZT camera?

Higher charge carrier statistics leading to superior energy resolution -> Better scatter rejection

Compact and light weight Have a high stopping power due to their high

mass density (5.8g/cm3) and high effective atomic No. of 49.6 (Cd0.9 :48, Zn0.1:30, Te:52) -> high efficiency

Efficient radiation absorption, good linearity, high stability, high sensitivity and wide dynamic range.


The Concept of Coded Aperture Imaging

Techniques originally used in Astronomical imaging

The detector response is

If G exists, then so we can recover O

through post processing

Contribution of background term can be minimized.

The concept of the CAI system. The CA encodes the photon source location and strength information in the detector. The original source distribution may be determined by correlating the recorded image with a decoded pattern.

)1(NAOD

)2(GDI

GA


Methods of Simulation MCNPX (Version 2.4.0) code is usedMCNPX (Version 2.4.0) code is used

Point-like (99mTc isotropic emitting 140 keV) sources are located Point-like (99mTc isotropic emitting 140 keV) sources are located inside the breast, with or without a background activity. inside the breast, with or without a background activity.

Photons are tracked on an event by event basis through scattering Photons are tracked on an event by event basis through scattering media, CA and CdZnTe camera, subject to Compton and photoelectric media, CA and CdZnTe camera, subject to Compton and photoelectric interactions.interactions.

The recorded (x, y) spatial information is blurred by sampling a The recorded (x, y) spatial information is blurred by sampling a Gaussian Gaussian probability distribution functionprobability distribution function with FWHM=1.58mm. with FWHM=1.58mm.

The energy deposition process is also subject to Gaussian broadening The energy deposition process is also subject to Gaussian broadening with energy dependent FWHM values from experimental data. with energy dependent FWHM values from experimental data.

Consider limited statistics expected in a clinically realistic imaging Consider limited statistics expected in a clinically realistic imaging situationsituation


VALIDATION OF MONTE CARLO SIMULATION

The developed simulation has been validated by comparison with experimental results.

Comparison of the spatial resolution for 99mTc of an infinitesimally small point source in air in terms of FWHM between simulated and experimental CZT camera data (using a LEHR collimator).

0

1

2

3

4

5

6

7

0 1 2 3 4 5 6 7 8 9 10

Distance from Collimator (cm)

Res

olut

ion

(mm

FW

HM

)

Brzymialkiewicz et. al. Data

Our Simulated Data

Brzymialkiewicz et. al., IEEE Trans. Med. Imaging 24 (2005) 868


Geometry of Point source in air

Simulated history ~0.5M


Simulation Results Point source in air


Simulation Results Point source in Tissue Equivalent

Material

A diagonal profile through the decoded images

0.00E+00

1.00E+03

2.00E+03

3.00E+03

4.00E+03

5.00E+03

6.00E+03

1 15 29 43 57 71 85 99 113 127 141 155

No. of Pixe ls

Yie

ld A

rbit

rary

Un

itPoint source in air

Point source in TEM

Simulated history ~0.5M


Simulation Results 10mm diameter lesion in cold and hot

TEM

History Simulated

~7.7M photons TBR 100:1


Simulation Results 10mm diameter lesion in air, cold and

hot TEM

History Simulated ~7.7M photons from each lesion

0.00E+00

2.00E+03

4.00E+03

6.00E+03

8.00E+03

1.00E+04

1.20E+04

1.40E+04

1 21 41 61 81 101 121 141 161

No. of Pixels

Yie

ld (

a.u

.)

Tumour in air

Tumour in TEMTumour in hot TEM


Simulation ResultsLesion Contrast

0

0.05

0.1

0.15

0.2

0.25

0 20 40 60 80 100 120

Tissue-to-background-ratio

Co

ntr

as

t 10 mm diameter lesion

visibility limit Lesion contrast was used Lesion contrast was used

to quantify the lesion to quantify the lesion visibility.visibility.

The contrast The contrast CC is defined is defined by normalizing the signal by normalizing the signal to the background counts to the background counts with the following formula:with the following formula:

The signal was obtained The signal was obtained using a ROI defined by the using a ROI defined by the FWHM that is calculated FWHM that is calculated by taking a profile through by taking a profile through the centre of the detected the centre of the detected tumor after a Gaussian tumor after a Gaussian curve (fit) obtained to the curve (fit) obtained to the data points. data points.

)1()(

B

BTc

0

2

4

6

8

10

12

0.00E+00 2.00E+06 4.00E+06 6.00E+06 8.00E+06 1.00E+07

No. of photons

Co

ntr

as

t

Point source in TEM

Point source in air


First Simulation Results of 3D Phantom Geometry

With cardiac shielding (1.5 mm tungsten)

Demonstrates that not only cardiac shielding required, but also from non-specific uptake in the torso. Ideally image breast using a radio-opaque

shielded plate


Summary

MURAs are attractive because displacing the gamma camera away from the breast allows access to a larger FoV (lymph nodes etc). Imaging point-like lesions [objects] in medicine akin to imaging stellar points [objects] in astronomy.

The Monte Carlo Simulation provides a realistic computational tool for modelling a coded aperture camera.

For this simple geometry, the 50% open fraction of the MURA mask can detect point like objects extremely well, both in air and in TEM. However when object assumes a finite size, AND non-specific uptake coupled with limited statistics present -> correlation noise artefacts present.

Future work focussed on developing better shielding geometry and developing a cross-correlation background subtraction method to reduce background correlation artefacts.

psd7 conference, 13/09/05 monte carlo simulation of a coded aperture imaging with dedicated gamma...

Documents