The Use of Computed The Use of Computed Tomography Images in Tomography Images in Monte Carlo Treatment Monte Carlo Treatment

Planning Planning

Magdalena BazalovaMagdalena BazalovaPhD defensePhD defense

AcknowledgementsAcknowledgements

• Frank Verhaegen, PhDFrank Verhaegen, PhD• Luc Beaulieu, PhDLuc Beaulieu, PhD• Jean-FranJean-François Carrier, PhDçois Carrier, PhD• Christophe Furstoss, PhDChristophe Furstoss, PhD• Eric Vigneault, MDEric Vigneault, MD• Robin van GilsRobin van Gils• McGill Medical Physics Unit staff and McGill Medical Physics Unit staff and

studentsstudents

• Natural Sciences and Engineering Natural Sciences and Engineering Research Council CanadaResearch Council Canada

Radiation therapyRadiation therapy

• The purpose of radiotherapy is to kill tumor The purpose of radiotherapy is to kill tumor cells by delivering a prescribed dose to the cells by delivering a prescribed dose to the tumor while sparing the healthy tissue.tumor while sparing the healthy tissue.

• Treatment planning is an important step of Treatment planning is an important step of radiotherapy and has to be performed radiotherapy and has to be performed carefully.carefully.

• The Monte Carlo method is the most The Monte Carlo method is the most accurate technique to calculate the accurate technique to calculate the delivered dose during treatment assuming delivered dose during treatment assuming the treatment machine model and the the treatment machine model and the patient anatomy are well known. patient anatomy are well known.

MotivationMotivationThe link between computed tomography The link between computed tomography

(CT) and Monte Carlo treatment planning (CT) and Monte Carlo treatment planning (MCTP)(MCTP)

MC geometryCT number

(HU)

Mass densityρ

Material(bone,tissue)

PhD?

Mass density calibration curve

y = 1.03E-03x + 1.03E+00

y = 8.84E-04x + 9.83E-01

y = 5.78E-04x + 1.05E+00

0.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

1.600

1.800

-1000 -500 0 500 1000

HU

mas

s de

nsity

[g/c

m3 ]

Mass density calibration curve

0.000

0.200

0.400

0.600

0.800

1.000

1.200

1.400

1.600

1.800

-1000 -500 0 500 1000

HU

mas

s de

nsity

[g/c

m3 ]

The conventional approach: (The conventional approach: (ρρ,HU) ,HU) density and tissue assignment for density and tissue assignment for

MCTPMCTP

bonesoft bone

tissueadipose

lung

Metal artifact reductionMetal artifact reduction Dual-energy CT-based tissue segmentationDual-energy CT-based tissue segmentation

• Correction of CT artifacts and its influence Correction of CT artifacts and its influence on Monte Carlo dose calculationson Monte Carlo dose calculations

M. Bazalova, S. Palefsky, L.Beaulieu and F. Verhaegen M. Bazalova, S. Palefsky, L.Beaulieu and F. Verhaegen Med. Phys.Med. Phys. 3434 2119-2132, 2007 2119-2132, 2007

- cubic spline sinogram interpolation correction method on - cubic spline sinogram interpolation correction method on phantoms with steel cylinders and on a patient, MC dose phantoms with steel cylinders and on a patient, MC dose calculations performedcalculations performed

• Monte Carlo dose calculation for phantoms Monte Carlo dose calculation for phantoms with real hip prostheseswith real hip prosthesesM. Bazalova, C. Coolens, F. Cury, P. Childs, L. Beaulieu and F. M. Bazalova, C. Coolens, F. Cury, P. Childs, L. Beaulieu and F. Verhaegen Verhaegen J. Phys. Conf. SeriesJ. Phys. Conf. Series 102102 2008 2008

- the correction method used on phantoms with real hip - the correction method used on phantoms with real hip prostheses, MC dose calculations performedprostheses, MC dose calculations performed

CT metal streaking artifactsCT metal streaking artifacts

corr

ecte

d i

mag

es

corr

ecte

d i

mag

es

Correction resultsCorrection resultsori

gin

al

CT

im

ag

es

ori

gin

al

CT

im

ag

es

head phantomshead phantoms pelvic phantomspelvic phantoms

Dose calculation resultsDose calculation results

18 MV photon18 MV photon

18 MeV electron18 MeV electron

exactexact

exactexact

differences from the exact differences from the exact distribution fordistribution fororiginal geom.original geom. corrected corrected

geom.geom.

> 5% < 2%

> 10%

< 2%

Artifact reduction for a patient Artifact reduction for a patient with bilateral hip prostheseswith bilateral hip prostheses

DVH of the prostateDVH of the prostate

• 20% of target voxels receive zero 20% of target voxels receive zero dose in the original geometrydose in the original geometry

• due to the incorrect assignment of due to the incorrect assignment of some voxels to air some voxels to air

• in MC dose calculation, dose to air is in MC dose calculation, dose to air is set to zeroset to zero

tissue segmentationtissue segmentation

original geometry

corrected geometry

Real hip prosthesesReal hip prostheses

Ti-alloyTi-alloy Stainless Stainless steelsteel

Co-Cr-Mo Co-Cr-Mo alloyalloy

ρρ=4.48 g/cm=4.48 g/cm33 ρρ=6.45 g/cm=6.45 g/cm33 ρρ=8.20 g/cm=8.20 g/cm33

Scatter and beam hardening as causes Scatter and beam hardening as causes of metal artifacts: a MC simulations of metal artifacts: a MC simulations

studystudyw

ith

w

ith

b

eam

hard

en

ing

beam

hard

en

ing

with scatterwith scatter without scatterwithout scatter

HU = -80HU = -80 HU = -39HU = -39

HU = -78HU = -78 HU = 3HU = 3

wit

hou

t w

ith

ou

t b

eam

hard

en

ing

beam

hard

en

ing

CT metal streaking artifacts: CT metal streaking artifacts: conclusionsconclusions

• Sinogram interpolation correction algorithm for Sinogram interpolation correction algorithm for metal streaking artifacts improves image metal streaking artifacts improves image quality and makes tissue segmentation and MC quality and makes tissue segmentation and MC dose calculations more accuratedose calculations more accurate

• Tested with three common hip prosthesis Tested with three common hip prosthesis materialsmaterials

• Patient study showed significant differences in Patient study showed significant differences in DVH of the target after artifact correction is DVH of the target after artifact correction is donedone

• Beam hardening has a minor effect on metal Beam hardening has a minor effect on metal streaking artifacts compared to scatterstreaking artifacts compared to scatter

• In MCTP, omitting metal streaking artifact In MCTP, omitting metal streaking artifact correction leads to large dose calculation errorscorrection leads to large dose calculation errors

Monte Carlo simulations of a Monte Carlo simulations of a CT x-ray tubeCT x-ray tube

• M. Bazalova and F. Verhaegen M. Bazalova and F. Verhaegen Phys. Med. Phys. Med. BiolBiol. . 5252 5945-5955, 2007 5945-5955, 2007

•The model was used in the thesis for dual-energy CT material extraction. The model was used in the thesis for dual-energy CT material extraction. •It can be used for scatter correction when metal streaking artifacts are present. It can be used for scatter correction when metal streaking artifacts are present.

CT x-ray tube simulation CT x-ray tube simulation resultsresults

CT x-ray tube MC simulation: CT x-ray tube MC simulation: conclusionsconclusions

• A Monte Carlo model of a CT x-ray tube A Monte Carlo model of a CT x-ray tube was developed and validated with half-was developed and validated with half-value layer and spectral measurements value layer and spectral measurements using a CdTe detector.using a CdTe detector.

• 100 and 140 kVp beams with no 100 and 140 kVp beams with no additional filtration and with 9 mm added additional filtration and with 9 mm added aluminum foil were tested and a very aluminum foil were tested and a very good agreement was found.good agreement was found.

• The modeled spectra were used for dual-The modeled spectra were used for dual-energy CT-based material extraction, the energy CT-based material extraction, the next part of the PhD. project.next part of the PhD. project.

Dual-energy CT imaging Dual-energy CT imaging • Dual-energy material extraction (DECT) is Dual-energy material extraction (DECT) is

based onbased on– taking CT images at two tube voltages taking CT images at two tube voltages

(100 kVp and 140 kVp)(100 kVp and 140 kVp)– parameterization of the linear parameterization of the linear

attenuation coefficientattenuation coefficient

• Results in the electron density (Results in the electron density (ρρee) and ) and the atomic number (the atomic number (ZZ)) values of each values of each voxelvoxel

Material segmentation for MC Material segmentation for MC dose calculationsdose calculations

1 tissue type in single- energy CT?

1 tissue type in single-energy CT

DECT for MCDCDECT for MCDC• Tissue segmentation in Monte Carlo Tissue segmentation in Monte Carlo

treatment planning: a simulation treatment planning: a simulation study using dual-energy CT imagesstudy using dual-energy CT images

M. Bazalova, J.-F. Carrier, L. Beaulieu and F. Verhaegen M. Bazalova, J.-F. Carrier, L. Beaulieu and F. Verhaegen Radioth. Oncol.Radioth. Oncol. 8686 93-98, 2008 93-98, 2008

• Dual-energy CT-based material Dual-energy CT-based material extraction for tissue segmentation in extraction for tissue segmentation in Monte Carlo dose calculationsMonte Carlo dose calculations

M. Bazalova, J.-F. Carrier, L. Beaulieu and F. Verhaegen M. Bazalova, J.-F. Carrier, L. Beaulieu and F. Verhaegen Phys. Med. Biol.Phys. Med. Biol. 5353 2439-2456, 2008 2439-2456, 2008

• Practical aspects of dual-energy CTPractical aspects of dual-energy CT

DECT: Monte Carlo simulationsDECT: Monte Carlo simulations

•BEAM and EGSnrc/DOSXYZnrc codeBEAM and EGSnrc/DOSXYZnrc code•Picker PQ5000 Picker PQ5000

• soft spectrasoft spectra 100 and 140 kVp100 and 140 kVp• hard spectrahard spectra added 9mm Al filteradded 9mm Al filter 100 and 140 kVp100 and 140 kVp

ZZ (5.740,14.141)(5.740,14.141)

ρρee (0.292,1.692) (0.292,1.692)

SOFT BEAMS

HARDBEAMS

MC simulation resultsMC simulation results

In order to minimize beam hardening effects, hard beams have to be used.In order to minimize beam hardening effects, hard beams have to be used.

DECT material extraction for tissue DECT material extraction for tissue segmentation in Monte Carlo dose segmentation in Monte Carlo dose

calculationscalculations• solid water phantom with RMI cylindrical inserts, scans solid water phantom with RMI cylindrical inserts, scans

taken at 100 and 140 kVp with a 9 mm Al filter, taken at 100 and 140 kVp with a 9 mm Al filter, ZZ and and ρρee extractedextracted

1.6 %

2.8 %

exact geometry

CT

DECT

ρe

Z

Material segmentation using Material segmentation using DECTDECT

250 kVp

18 MeV

Dose calculation resultsDose calculation results

17 %17 %

anterior 250 kVpanterior 250 kVp

<1 %<1 %

6 %6 %

lateral 18 MeVlateral 18 MeV

<1 %<1 %

SingleE-material segmentation • dose calculation errors are the largest in the soft bone tissue equivalent material irradiated by the 250 kVp photon beam, up to 17%!• the largest dose difference for the 18 MeV electron beam is found to be in the polyethylene cylinder, being 6%

DualE-material segmentation • all dose differences in all cylinders are below 1%

Practical aspects of DECTPractical aspects of DECT

• Streaking artifact reduction Streaking artifact reduction observed in the RMI phantom studyobserved in the RMI phantom study– US phantom with brachytherapy seedsUS phantom with brachytherapy seeds– permanent implant prostate patientpermanent implant prostate patient

125125I dose calculations with I dose calculations with MCNPXMCNPX

Ddual/Dsingle

70%

dose distribution

ρdual

ρsingle

tissues: dual

tissues: single

MC dose calculation resultsMC dose calculation resultsρDECTρCT

No significant differences for this particular patient.No significant differences for this particular patient.

Dual-energy CT: Dual-energy CT: conclusionsconclusions

• In order to minimize beam hardening effects, In order to minimize beam hardening effects, dual-energy CT (DECT) should be done with hard dual-energy CT (DECT) should be done with hard beamsbeams

• DECT material extraction was successful for a set DECT material extraction was successful for a set of tissue-equivalent materials with a wide range of tissue-equivalent materials with a wide range of densities and atomic numbers using 100 kVp of densities and atomic numbers using 100 kVp and 140 kVpand 140 kVp

• DECT results in more accurate MC dose DECT results in more accurate MC dose calculation results, especially for low-calculation results, especially for low-ρρee high- high-ZZ materials and orthovoltage and kilovoltage beamsmaterials and orthovoltage and kilovoltage beams

• Patient DECT tissue segmentation is significantly Patient DECT tissue segmentation is significantly influenced by image noise and patient motion influenced by image noise and patient motion

• DECT has the potential to reduce streaking artifacts DECT has the potential to reduce streaking artifacts from brachytherapy seeds, the effect on MCDC should from brachytherapy seeds, the effect on MCDC should be studied with a larger group of patients be studied with a larger group of patients

Overall conclusionsOverall conclusions• The link between computed tomography The link between computed tomography

images and Monte Carlo geometry files was images and Monte Carlo geometry files was strengthened by means of metal streaking strengthened by means of metal streaking artifact reduction and dual-energy CT-based artifact reduction and dual-energy CT-based material extraction.material extraction.

• The dose delivered to patients during The dose delivered to patients during radiotherapy can be therefore calculated more radiotherapy can be therefore calculated more accurately with Monte Carlo techniques.accurately with Monte Carlo techniques.

Future workFuture work• Scatter correction for metal artifact removal.Scatter correction for metal artifact removal.• Noise reduction in DECT images.Noise reduction in DECT images.• MC dose calculations for more permanent MC dose calculations for more permanent

implant brachytherapy patients.implant brachytherapy patients.

Thank you!Thank you!