3D-RD Imaging-Based Di...

3D-RD Imaging-D i tDosimetry

George Sgouros, Ph.D.g g ,

Russell H. Morgan Dept of Rg p& Radiological Science

Johns Hopkins University, SBaltimore MD

-Based

adiology gy

School of Medicine

Patient-Specifp• Patient's AnatomyPatient s Anatomy

- CT/MRI

• Patient's Activity Dis- SPECT/PETSPECT/PET

• Spatial distribution o- non-uniform activity di- absorbed dose "imageg- dose volume-histogram

fic Dosimetryy

stribution

of absorbed dosestributions"

ms

Flux GD, Webb S, Ott RJ, Chittenden SJ, Thomintralesional radionuclide therapy using matheJ Nucl Med 1997;38:1059 –1066

Sgouros G, Barest G, Thekkumthala J, et al. Tretherapy: three-dimensional dosimetry for nonuMed 1990;31:1884 –1891

Ljungberg M, Sjogreen K, Liu X, et al. A 3-dimebased on quantitative SPECT for radionuclide tsimulation. J Nucl Med. 2002;43:1101–1109.J Nucl Med. 1997;38:1059 1066.

Kolbert KS, Sgouros G, Scott AM, et al. Implemthree-dimensional internal dosimetry. J Nucl M

Med. 1990;31:1884 1891.

Sgouros G, Chiu S, Pentlow KS, et al. Three-dimtreatment planning. J Nucl Med. 1993;34:1595–

simulation. J Nucl Med. 2002;43:1101 1109.

Sgouros G, Kolbert KS. The three-dimensional In: Zaidi H, Sgouros G, eds. Therapeutic ApplicMedicine Philadelphia PA: Institute of PhysicsErdi AK, Yorke ED, Loew MH, et al. Use of the fadose calculation in radionuclide therapy. Med P

Liu A Williams LE Wong JY Raubitschek AA

Kolbert KS, Sgouros G, Scott AM, et al. Dose-vdose distribution in 3-dimensional internal dosP124.

Medicine. Philadelphia, PA: Institute of Physics

McKay E. A software tool for specifying voxel mBiother Radiopharm. 2003;18:379 –392.Liu A, Williams LE, Wong JY, Raubitschek AA. method (MAVSK) for internal beta dosimetry. N

Clairand I, Ricard M, Gouriou J, Di Paola M, Aubased software for internal radionuclide dosim

Giap HB, Macey DJ, Bayouth JE, Boyer AL. Valtechnique for internal dosimetry. Phys Med Bio

Giap HB Macey DJ Podoloff DA Development

Descalle MA, Hartmann Siantar CL, Dauffy L, etsimulations to targeted radionuclide therapy. C

Sgouros G Squeri S Ballangrud AM et al Patibased software for internal radionuclide dosim

Bolch WE, Bouchet LG, Robertson JS, et al. MInonuniform activity distributions–radionuclide 1999;40:11S 36S

Giap HB, Macey DJ, Podoloff DA. Developmenttreatment planning system for radioimmunothe

Furhang EE, Chui CS, Sgouros G. A Monte CarPhys 1996;23:1523 1529

Sgouros G, Squeri S, Ballangrud AM, et al. PatiHodgkin’s lymphoma patients treated with 131Idose-response. J Nucl Med. 2003;44:260 –268.

Ljungberg M Frey E Sjogreen K et al 3D abso1999;40:11S–36S.

Yoriyaz H, dos Santos A, Stabin MG, Cabezas Rphantom calculated with the MCNP-4B code. M

Phys. 1996;23:1523–1529.

Tagesson M, Ljungberg M, Strand SE. A Monte-distributions to absorbed dose distributions inA t O l 1996 35 367 372

Ljungberg M, Frey E, Sjogreen K, et al. 3D absoevaluation for 111-In/90-Y therapy using Monte Radiopharm. 2003;18:99 –107.

Sgo ros G Kolbert KS Sheikh A et al PatientYoriyaz H, Stabin MG, dos SA. Monte Carlo MCestimates for patient-specific dosimetry. J Nuc

Acta Oncol. 1996;35:367–372.

Furhang EE, Chui CS, Kolbert KS, Larson SM, Sdosimetry method for patient-specific internal e

Sgouros G, Kolbert KS, Sheikh A, et al. Patienttherapy using 124I PET and 3-dimensional-inte2004;45:1366 –1372.

as R. Three-dimensional dosimetry for ematical modeling and multimodality imaging. eatment planning for internal radionuclide

uniformly distributed radionuclides. J Nucl nsional absorbed dose calculation method therapy: evaluation for 131I using Monte Carlo

mentation and evaluation of patient-specific Med. 1997;38:301–308.

mensional dosimetry for radioimmunotherapy –1601.

internal dosimetry software package, 3D-ID. cations of Monte Carlo Calculations in Nuclear s; 2002ast Hartley transform for three-dimensional Phys. 1998;25:2226–2233.

Monte Carlo-assisted voxel source kernel

volume histogram representation of patient simetry [abstract]. J Nucl Med. 1994;35:P123–

s; 2002.

models for dosimetry estimation. Cancer

Monte Carlo assisted voxel source kernel Nucl Med Biol. 1998;25:423– 433.

bert B. DOSE3D: EGS4 Monte Carlo code-metry J Nucl Med 1999;40:1517–1523

idation of a dose-point kernel convolution ol. 1995;40:365–381.

t of a SPECT-based three dimensional

t al. Application of MINERVA Monte Carlo Cancer Biother Radiopharm. 2003;18:71–79.

ient-specific 3-dimensional dosimetry in non-metry. J Nucl Med. 1999;40:1517–1523.

RD pamphlet no. 17: the dosimetry of S values at the voxel level. J Nucl Med.

t of a SPECT-based three dimensional erapy. J Nucl Med. 1995;36:1885–1894.

rlo approach to patient-specific dosimetry. Med

ient-specific, 3-dimensional dosimetry in non-I-anti-B1 antibody: Assessment of tumor

orbed dose calculations based on SPECT:

R. Absorbed fractions in a voxel-based Med Phys. 2000;27:1555–1562.

-Carlo program converting activity a radionuclide treatment planning system.

orbed dose calculations based on SPECT: Carlo simulations. Cancer Biother

specific dosimetr for 131I th roid cancerNP-4B-based absorbed dose distribution l Med. 2001;42:662–669.Sgouros G. Implementation of a Monte Carlo emitter therapy. Med Phys. 1997;24:1163–1172.

-specific dosimetry for 131I thyroid cancer ernal dosimetry (3D-ID) software. J Nucl Med.

Sgouros, et al., JNM ‘90

Metho

• 3D-ID used for dose calculatio– ROI for each lesion.– mean, min, max, DVH– point-kernel method

K(r)

DO

SE rr

DISTANCE

DOSE = CADOSE = CA11 x K(rx K(r11) +) +

ds

ons

rr22

rr11rr33

+ CA+ CA22 x K(rx K(r22) + ...) + ...

3D3D--ID SystemID SystemyyInputInput:: Registered anatomic & funRegistered anatomic & fun

( CT/MR

( CT/MRSPEC

Convert images from a variety of sources into a common data formatformat.

Define regions-of -interest volumes by drawing contoursvolumes by drawing contours.

Select source and target gvolumes for dose calculation.

nctional imagesnctional images

I iI images,CT/PET images)

Calculate absorbed dose to target from source volume.

Create 3D dose distribution mapsmaps.

Generate mean dose, dose-volume histograms andvolume histograms and parametric images.

3D3D--ID FunctionID Function3D3D ID FunctionID Functionn Access Paneln Access Paneln Access Paneln Access Panel

3D-ID Methodology

Tumor DVHTumor DVH

Mean Dose:0.05 (Gy)

100

80

LUM

E

40

60

TUM

OR

VO

L

40

20

% O

F T

0.02 0.04 0.06 0.08DOSE per mCi (Gy)

0

Kolbert, et al., JNM ‘9

I-124 PET-based t

• Feasibility of 124I-PET-b• Fully 3-D calculation

– 3D kinetics based on mu– not planar imaging and o

• Evaluate dose uniformiEvaluate dose uniformi– DVH, min, max

images– images• correlate with response

hyroid dosimetry

ased 3-D dosimetry

ultiple PET scanspone SPECTty (or lack thereof)ty (or lack thereof)

e


Metho

• 15 patients w/ metastatic thyro– 3 - 4 124I-PET scans over 7 – treated with 131I

• PET scans were co-registered– MIAU (Multiple Image Analy– transmission studies for ana– 4-d data set (x,y,z,t) for eac( y )

• Convert images to 131I effectiv– 124I effective → biological →g

zyxAtzyxA = ,,(),,,( 124131

ds

oid Ca.days (pre-therapy)

dysis Utility) Software Packageatomical landmarksch patientpe distribution

→ 131I effectivett eetz ⋅−⋅ ⋅⋅ 131124), λλ


Clinical Impleme• 15 patients w/ metastatic thy

3 4 124I PET 7 d- 3 - 4 124I-PET scans over 7 days- treated with 131I

PET scans were co registere• PET scans were co-registere- MIAU (Multiple Image Analysis- transmission studies for anato- transmission studies for anato- 4-d data set (x,y,z,t) for each p

• Convert images to 131I effectConvert images to I effect- 124I effective → biological → 131

zyxAtzyxA = ,,,(),,,( 124131

entationyroid Ca.

( th )s (pre-therapy)

ededs Utility) Software Packageomical landmarksomical landmarksatientive distributionive distribution1I effective

tt eet ⋅−⋅ ⋅⋅ 131124) λλ


Lesion identification dPt. 13 PET

L2

L1

L4

Selected coronal slic

“… with tracer avid disease. There are multiplestructures in particular in the left scapular reg

Selected coronal slic

structures, in particular in the left scapular regproximal femur, and the right mid-femoral diaplower thoracic region, and approximately .…”

database

MIAU

L3L3

L1

L2

L4

ces summed

e metastases visualized in the bony ion right lower lateral chest wall right hip and

ces coronal

ion, right lower lateral chest wall, right hip and physis. Vertebral mets are visualized in the


Results - isodose c

day: 0 1 2

contoursPt. 13

Isodose levels of 10,25,5050,75%, , ,


Results - Dose-volum

80

100

Absor

60

80

ume 1 3

2 3 3 3

min

40

% V

ol

3 3 4 1

0

20

00 100 200

Absorbed D

me histogram

rbed dose (Gy)

Pt. 13

186 43 407 51

221 33

n max mean

221 33 43 10

300 400 500Dose (Gy)

(104 mCi)


Lesion identification dPt. 16 PET

L5

L4

L5

“ h d diff bil t l l ti i

coronal sagitta

“ … showed diffuse bilateral lung activisuperior mediastinum, as well as near tconsistent with ….”

MIAUdatabase

L3

L2L1

MIAU

L4L4

L5

it ll t k i th k th

al summed coronal

ity, as well as uptake in the neck, the the left posterior chest wall area, all


Results - Dose-volum

100

80

me

40

60

% V

olum

20

40

%

00 100 2000 100 200

Absorbed D

me histogram

Absorbed dose (Gy)

Pt. 16

Absorbed dose (Gy)

1 1 2 3

min max mean187 26432 692 3

3 5 4 10 5 1

198 45261 90228 28

300 400 500(400 mCi)

300 400 500Dose (Gy)


Results - Summary

Mean, min, max calculated fo

Absorbed Dose (Gy)1

( y)(admin. activity, GBq)

Mean 1.2 (15)

540 (11) m

ors

1

ea (15) (11)

Min 0.3 (15)

50 (14) N

umbe

r of t

u

(15) (14)

Max 1.5 (15)

4000 (11)

or 56 different lesions

Frequency distribution2

8

0

4

6

0

2

Absorbed dose (Gy)

0 100 200 300 400 500 6000


3-D Radiobiolog(3D RD)(3D-RD)

• 3D-ID• Radiobiological mo• Dose rate differenc• Dose-rate differenc• Non-uniformity in ay• Density differences

gical Dosimetry

odelingcescesactivity distributionys

Prideaux et al. J Nucl Med ‘07

3-D Radiobiolog(3D-(3D

• Extension of 3D-Internal Dosimetry (3• Radiobiological modeling with tissue/

Biologically Effective Dose (BED) A t f d t i ti

g gused to get

• Accounts for dose rate variations• Reference value relates to dose rate

⎞⎛ ( )⎟⎟⎟

⎠

⎞

⎜⎜⎜

⎝

⎛∞

+= DGDBEDβ

α1

dwewDdttD

G wtt

)(

002 )( )(D 2 )( −−

•∞ •

∫∫=∞ μ

α and β are the tissue specific coefficients forα and β are the tissue specific coefficients for radiation damage; μ is repair constant

gical Dosimetry-RD)RD)D-ID) (Kolbert et al JNM ’97)

/tumor specific α, β, μ values are

Equivalent Uniform Dose (EUD)Accounts for non uniform absorbed

p , β, μ

• Accounts for non-uniform absorbed dose distribution

• Provides a single value that may be used to compare different dose distributionsto compare different dose distributions

• Reference value relates to uniform distribution

⎟⎟⎠

⎞⎜⎜⎝

⎛−= ∑

=

−N

i

BED

NeEUD

i

1ln1

α

α

Prideaux et al JNM ’07, Hobbs, et al Med Phys ’09Baechler, et al Med Phys ‘08

Individual kidneys absorbUsing 86Y-DOTATOC quUsing 86Y-DOTATOC qu

MIRDOSE3.1 for dos(assuming a standa

)

5

ose

(mG

y/M

Bq)

3

4

y ab

sorb

ed d

o

2

Kid

ney

1

Dosimetry # 24

0MALE FEMALE

bed dose of 90Y-DOTATOCuantitative imaging anduantitative imaging and simetric calculationsard kidney volume)

Therapy planning:activities delivering 27 Gy to the kidneys

T i it t

UNIVERSITE CATHOLIQUE DE LOUVAIN

Toxicity assessment

Correlation betwee

Importance of organ vo

and creatinine clearance lo

Standard kidney volumes

s/ye

ar 60

/

eara

nce

loss

asel

ine)

40

l

eatin

ine

cle

(% b

a

20 tii

l

20 30 40

Cre

0

C

Dosimetry # 25

Dose Standard volume (Gy)

20 30 40

en kidney dose (Gy)

lume in self irradiationy ( y)

oss/year (% baseline) N=18

Kidney volumes measured by CTt (70%)cortex (70%)

s/ye

ar 60R=0.54p= 0.02

eara

nce

loss

asel

ine)

40

eatin

ine

cle

(% b

20

0 10 20 30 40 50

Cre

0

Kidney dose CT volume (Gy)

0 10 20 30 40 50


Correlatioand creatininea d c eat e

r 60ce

loss

/yea

rne

)

60R=0.93p<0.0001

ne c

lear

anc

(% b

asel

in 40

Cre

atin

in 20

Biologic Ef

0 10 200

Dosimetry # 26

Barone R, Borson-Chazot F, Valkema R, et

on between BEDe clearance loss/yeare c ea a ce oss/yea

ffective Dose (Gy)

30 40 50 60


al. J Nucl Med. 2005 Jan;46 Suppl 1:99S-106S

Clinical ThyroCT

SPECTNon Uniform Density in Lungs

SPECT

•Non Uniform Activity Distribution

oid Case

Eff. Half-life

•Non Uniform ClearanceNon Uniform Clearance

Prideaux et al. J Nucl Med ‘07

Clinical Thyro005

0.03

0.04

0.05

Dose Volume

0.01

0.02

00 50 100

Dose (Gy)

M D 57 7 G 9 5

( y)

Tumor LunMean Dose = 57.7 Gy 9.5

Mean BED = 58.5 Gy 9.8Mean BED 58.5 Gy 9.8

EUD = 25.0 Gy 8.3

oid Case

Tumor

HistogramTumorLung

150 200y)

G

y)

ngGy

Gy dose rateGy

Gy uniformityPrideaux et al. J Nucl Med ‘07

3D-RD Clinical3D RD Clinical • Real time (1 week) 131I trea

year old girl with metastaticyear-old girl with metastaticthyroid cancer using patiendosimetry (3D-RD).

• Heavy lung involvement mpulmonary toxicity and con

• Other considerations: tumo• Other considerations: tumo• Patient had prior 131I for dia

significant quantities espec• Use 124I and PET/CT for do

ImplementationImplementationatment planning for an 11 c differentiated papillaryc differentiated papillary nt specific 3-dimensional

meant concern about cern for overdosing

or dose and brain toxicityor dose and brain toxicity agnostic and still retained cially in two brain tumors osimetric assessment

Hobbs, et al JNM ’09

MetMet• The patient received 92 MBq (2

Whole body PET/CT scans wer• Whole body PET/CT scans wer96 h.– 2D mode with tungsten septa in pl

Calibration with a standard measu– Calibration with a standard measu• 3D-RD calculation includes

– longitudinal co-registrationcompensation for different half live– compensation for different half-live

– EGS-based Monte Carlo simulatio• The dose rate results were fitted

dose per administered (131I) actdose per administered ( I) actscaled to MTD of 27 Gy to norm

• Other methods (absorbed fractiLeeper) were used for comparisLeeper) were used for comparis

hodhod2.5 mCi) of 124Ire performed at 1 24 48 72 andre performed at 1, 24, 48, 72, and

laceured in counting wellured in counting well

eseson of 131I decay for each time point.d and an estimated absorbed tivity to lungs was obtained andtivity to lungs was obtained and mal lung ion with OLINDA and Benua-son using PET activity mapsson using PET activity maps


PET/CTPET/CTT imagesT images


PET-based thyrAbsorbed Dose Map

• Based on dosimetry analysis, patient waadministered 5.1 GBq so as not to excee25 27 G t l

• Example of on-line dosimetry calculation• MC started after first acquisition

25-27 Gy to lungs• Physician was thinking of 7.4 GBq• Absorbed dose to T. lobe lesion ≈ 325 Gy

• Calculation completed within 48 h of lastime-point

• Patient temporal lobe tumor has shrunk• Lung tumor dose 36 Gy • Equivalent uniform dose (EUD) = 11.6 Gy

Patient temporal lobe tumor has shrunk• family is very happy

oid dosimetryDose-Volume histograms

as ed n

yAb b d d di ib i f

t

y

Absorbed dose distribution for 5.1 GBq 131I administration

OLINDA-absoOLINDA abso• Residence times from lungs and

rest of body poolrest of body pool• Input into OLINDA for all phantoms• Phantom results as a function of

mass and fit• Input patient mass• Scale to 27 Gy MTD constraint• AA: 2.89 GBq (78 mCi) 89 G q ( 8 C )

orbed fractionorbed fraction


MethodologicaMethodologica

• What activity to adminisWhat activity to adminis• OLINDA: 2.9 GBq

3D RD 5 1 GB• 3D-RD: 5.1 GBq• Retrospective re-exami

al Comparisonal Comparison

ster?ster?

nation


OLINDAOLINDA • Patient lung mass greater g g

than typical – Tumor increases density

Higher mass means lower dose– Higher mass means lower dose for same activity

– Plot OLINDA phantom results as a function of lung massas a function of lung mass

– Input patient lung mass– Scale to 27 Gy MTD– AA: 5.18 GBq (140 mCi)

• Convergence of results!

reviewedreviewed


Lung/Tumor DLung/Tumor D• 27 Gy Constraint to “normal

l ti ” th th llung tissue” rather than lung VOI

• Define “normal lung”• Discriminate based on

activity uptake at 48 h (overlap with anatomical position density)position, density)

• BED uses numeric integration

/β t diff t l– α/β parameters different – less variation in the BED

• AA = 6.83 GBq (185 mCi)

DiscriminationDiscrimination


3D-RD for pe3D RD for pe• Feasibility of real time tre

3D-RD, patient-specific d• A higher recommended Ag

based method (with a higoutcome) was obtained. )

• Re-visitation of methods this case)this case).

• Further investigation of ludiscrimination in futurediscrimination in future

ediatric caseediatric caseeatment planning using dosimetry. AA than by an S-value yghly favorable clinical

led to convergence (for

ung/tumor


Salivary Gland 124

Collaboration with Prof. Andreas Bockisch, M.D. and Dr. Walter Jentzen, Ph.D.Uni ersit of D isb rg Essen

• Dose• 7 124I • CompUniversity of Duisburg-Essen,

Essen, Germany • Comp• Impac

~2 h ~7

4I-PET dosimetry-Response for Salivary gland toxicityPET scans over 4 dayspare w/ absorbed fraction calculationpare w/ absorbed fraction calculationct of spatial dose distribution

72 h

Absorbed dose map for doseAbsorbed dose map for dose contribution due to decays after the first 24 h

The Problem

“C l l t“Calculate energydensity (i.e., absoy ( ,particular organ o

ol me”volume”

defined

d itiy deposition orbed dose) in a )or tumor

The ProblemIndex of response an• absorbed dose, D(x,y• absorbed dose rate, Dabsorbed dose rate, D• Radiosensitivity, R(x,• proliferation rate, P(x• Criticality (importancCriticality (importanc

failure), C(x,y,z)

Response = F(D,D

definedd toxicity:y,z)DR(x,y,z)DR(x,y,z),y,z)x,y,z)ce in likely organce in likely organ

R,R,P,C); F ?

3D-RD Imaging-Based Di...

Documents

Transcript of 3D-RD Imaging-Based Di...