IMAGE-BASED PATIENT-SPECIFIC DOSIMETRY FOR RADIONUCLIDE THERAPY - - PowerPoint PPT Presentation

image based patient specific dosimetry for radionuclide
SMART_READER_LITE
LIVE PREVIEW

IMAGE-BASED PATIENT-SPECIFIC DOSIMETRY FOR RADIONUCLIDE THERAPY - - PowerPoint PPT Presentation

IMAGE-BASED PATIENT-SPECIFIC DOSIMETRY FOR RADIONUCLIDE THERAPY Michael Ljungberg, PhD, Katarina Sjgreen-Gleisner, PhD International Symposium on Standards, Applications and Quality Assurance in Medical Radiation Dosimetry 9-12 November


slide-1
SLIDE 1

Medical Radiation Physics, Clinical Sciences Lund, Lund University

Michael Ljungberg, PhD, Katarina Sjögreen-Gleisner, PhD

IMAGE-BASED PATIENT-SPECIFIC DOSIMETRY FOR RADIONUCLIDE THERAPY

International Symposium on Standards, Applications and Quality Assurance in Medical Radiation Dosimetry 9-12 November 2010 Vienna, Austria

slide-2
SLIDE 2

Medical Radiation Physics, Clinical Sciences Lund, Lund University

BASIC DOSIMETRY

Absorbed dose is mean imparted energy in a mass element! Conditions:

  • A radiation source volume somewhere
  • A target volume somewhere
  • The intensity and characteristics of the radiation will be changed on its way

toward the target volume!

Common equation for dosimetry in Nuclear Medicine

  • A is the total number of disintegrations (cumulated activity)
  • S describes the energy, emitted from the source volume, absorbed in the

target volume per mass unit and disintegration.

D = A S %

~

slide-3
SLIDE 3

Medical Radiation Physics, Clinical Sciences Lund, Lund University

THE ”MIRD” EQUATION

{

D A S A A n E m S S φ = ⋅ = ⋅Δ⋅Φ = ⋅ ⋅ ⋅ % % % 1 4 2 4 3

( ) ( ) ( )

k h

r r r r r r

i T S i T S i h i i h i i T T

n E m D m A A φ φ ← = ⋅ = ⋅ ← ← Δ

∑ ∑

% %

i i i i i

n E Δ = Δ =

∑ ∑

n is the number of particles emitted per transition E is the mean energy per particle S denotes the source T denotes the target

( ) ( )

S

r r 1 r r 0 ; T S

np T np T S

φ φ ← = ← = ≠

( )

T S

r r 1

i

φ ≤ ← ≤

Photons Electrons Mean energy per disintegration Absorbed Fraction

slide-4
SLIDE 4

Medical Radiation Physics, Clinical Sciences Lund, Lund University

APPLICATIONS OF DOSIMETRY IN NUCLEAR MEDICINE

Dosimetry for Diagnostic Nuclear Medicine

  • Estimate risk for cancerogenic effects and hereditary changes
  • Low activities / gamma-radiation
  • Individuals is not in focus
  • Populations
  • Specific for the study but not for the individual patient

Dosimetry for therapy with radionuclides

  • Primary aim is to treat a disease with radiation
  • High activities / charged particles
  • The individual is in focus
  • Study specific as well as patient specific
slide-5
SLIDE 5

Medical Radiation Physics, Clinical Sciences Lund, Lund University

DOSIMETRY FOR TREATMENT

Activity (A)

  • Measurement with a ’diagnostic tracer amount ’ for kinetic and dosimetry

calculations

  • Preferably made with a scintillations camera or SPECT/PET
  • Use Gy/MBq to predict activity needed to delived an prescribed absorbed

dose to the target

Geometry (S)

  • The more accurate the geometry is - the better.
  • Patient-specific geometry from a CT study
  • Sometimes difficult to segment target volumes in 3D

General Goal: To determine absorbed dose for individuals

slide-6
SLIDE 6

Medical Radiation Physics, Clinical Sciences Lund, Lund University

WHY IS RADIONUCLIDE DOSIMETRY DIFFICULT?

External Therapy

  • Well-defined source and intensity
  • Turn-on and off
  • Energy usually evenly distributed within a volume element
  • High dose-rate

Radionuclide therapy

  • Injection of the source
  • Cannot turn the source on and off!
  • Need to measure the activity distribution in time and space
  • Imaging systems have limitations (spatial resolution, noise,attenuation ..)
  • Localization in the tissues and cells generally heterogeneous
  • Biokinetic may vary with patients
  • Low dose-rate
slide-7
SLIDE 7

Medical Radiation Physics, Clinical Sciences Lund, Lund University

THE DIFFICULTIES IN ACTIVITY MEASUREMENTS

A.

No patient motion and perfect camera resolution

B.

Patient respiration and heart beating

C.

Normal system resolution and patient movements

D.

Photon attenuation

E.

Photon attenuation and scatter

F.

Realistic noise level A B C D E F

Monte Carlo simulated images

slide-8
SLIDE 8

Medical Radiation Physics, Clinical Sciences Lund, Lund University

2D DOSIMETRY – PRINCIPLES

Dosimetry based on tools developed for diagnostic dosimetry

  • Activity from Planar WB measurements

(Geometrical-Mean)

  • S-values calculated from analytical

phantoms

  • Assumes homogenous activity i organ
  • Calculate mean absorbed dose in
  • rgans
  • Correction for differences in organ

masses relative to reference phantom!

A S

slide-9
SLIDE 9

Medical Radiation Physics, Clinical Sciences Lund, Lund University

AUC Intensity [counts] Time [h]

BIOKINETICS TO OBTAIN CUMULATED ACTIVITY

slide-10
SLIDE 10

Medical Radiation Physics, Clinical Sciences Lund, Lund University

DEVELOPMENT OF MORE REALISTIC PHANTOMS

May lead to more ’patient specific’ dosimetry

Example: The NCAT phantom by P Segars, Duke University

slide-11
SLIDE 11

Medical Radiation Physics, Clinical Sciences Lund, Lund University

WHY IMAGE‐BASED 3D DOSIMETRY

Dosimetry based on two-dimensional (2D) whole-body imaging has known limitations.

  • Contribution from overlapping structures
  • Attenuation correction is 2D
  • Scatter correction
  • Source thickness correction
  • Background/overlap correction

3D images provides information of the absorbed dose on a voxel level

  • Heterogenity
  • Corrections more accurate
  • Patient-specific anatomy
  • SPECT/CT on different time points – biokinetics on voxel level

Hybrid SPECT/WB method common compromise!

  • One Quantitative SPECT measurement to nomalise a kinetic curve obtained

from multiple WB measurements

slide-12
SLIDE 12

Medical Radiation Physics, Clinical Sciences Lund, Lund University

3D DOSIMETRY

Multiple registered SPECT/CT or PET/CT studies Correction for

  • Photon attenuation
  • Scattered radiation
  • Collimator resolution
  • Septal penetration
  • Partial-Volume Effect

3D dose calculation from

  • Dose kernels
  • Monte Carlo method

Evaluated as

  • Dose/volume histograms
  • Relate to biological effect

Function Imaging

SPECT/PET

Function Imaging

SPECT/PET

Image Registration Image Registration Anatomical Imaging

CT

Anatomical Imaging

CT

Correction for

Attenuation and Scatter Collimator Response Septal Penetration Partial-Volume Effects

Correction for

Attenuation and Scatter Collimator Response Septal Penetration Partial-Volume Effects

Dose Calculation by

Dose Kernels Monte Carlo

Dose Calculation by

Dose Kernels Monte Carlo

Segmentation DV Histogram Segmentation DV Histogram Obtain biokinetics Obtain biokinetics Image Reconstruction Image Reconstruction

slide-13
SLIDE 13

Medical Radiation Physics, Clinical Sciences Lund, Lund University

3D DOSIMETRY

Modern SPECT/CT (PET/CT) systems makes life easier Correction for

  • Photon attenuation
  • Scattered radiation
  • Collimator resolution
  • Septal penetration
  • Partial-Volume Effect

3D dose calculation from

  • Dose kernels
  • Monte Carlo method

Evaluated as

  • Dose/volume histogram
  • Relate to biological effect

Correction for

Attenuation and Scatter Collimator Response Septal Penetration Partial-Volume Effects

Correction for

Attenuation and Scatter Collimator Response Septal Penetration Partial-Volume Effects

Dose Calculation by

Dose Kernels Monte Carlo

Dose Calculation by

Dose Kernels Monte Carlo

Segmentation DV Histogram Segmentation DV Histogram Obtain biokinetics Obtain biokinetics Image Reconstruction Image Reconstruction Hybrid SPECT/CT Hybrid SPECT/CT

slide-14
SLIDE 14

Medical Radiation Physics, Clinical Sciences Lund, Lund University

3D DOSIMETRY

Today Quantification is made by iterative methods Include correction for

  • Photon attenuation
  • Scattered radiation
  • Collimator resolution
  • Septal penetration
  • Partial-Volume Effect

3D dose calculation from

  • Dose kernels
  • Monte Carlo method

Evaluated as

  • Dose/volume histogram
  • Relate to biological effect

Iterative Methods preferrable since they allow for correction of

Attenuation and Scatter Collimator Response Septal Penetration Backscatter

Iterative Methods preferrable since they allow for correction of

Attenuation and Scatter Collimator Response Septal Penetration Backscatter

Dose Calculation by

Dose Kernels Monte Carlo

Dose Calculation by

Dose Kernels Monte Carlo

Segmentation DV Histogram Segmentation DV Histogram Obtain biokinetics Obtain biokinetics Hybrid SPECT/CT Hybrid SPECT/CT

slide-15
SLIDE 15

Medical Radiation Physics, Clinical Sciences Lund, Lund University

New Image Estimate Forward Projection Estimated Projections Comparing step Error projection Backproject Error Update step Measured Projections Initial Image Estimate More angles? Image space Projection space

PRINCIPLES OF THE ML-EM ALGORITHM

yes

More Iterations

Exit

yes

Ratio

slide-16
SLIDE 16

Medical Radiation Physics, Clinical Sciences Lund, Lund University

ABSORBED DOSE CALCULATIONS FROM SPECT/PET IMAGES

Dose Calculation

Absorbed Dose Rate Activity Density

( ) ( )

T S T S

r r r r

i S i i i T

D A n E m φ ← ← = ⋅∑

Same ”MIRD” Equation!!!

Source is one voxel in the SPECT/PET image set Target is one voxel in the density image set S values not pre-tabulated but calculated when needed Patient-specific geometry

slide-17
SLIDE 17

Medical Radiation Physics, Clinical Sciences Lund, Lund University

POINT‐DOSE KERNELS – 3D

Describes specific absorbed fraction as function of radial distance from a point source. Derived for homogeneous media (H2 O) using Monte Carlo calculations.

  • photons
  • mono-energetic electrons
  • β-particles
  • Radionuclides
slide-18
SLIDE 18

Medical Radiation Physics, Clinical Sciences Lund, Lund University

FULL MONTE CARLO CALCULATION – 3D

Energy deposition in 3D Takes into account heterogenities in sources and tissues May take considerable of CPU time ”Public Domain” programs

  • EGS4,EGSnrc
  • MCNPX
  • Geant4
  • Penelope
  • ........

90‐Y MCNP4 Simulation

slide-19
SLIDE 19

Medical Radiation Physics, Clinical Sciences Lund, Lund University

3D DOSIMETRY BASED ON SPECT/PET

Absorbed dose calculation from local deposit energy, dose kernels or full Monte Carlo are on a voxel level Segmentation of

  • rgan volumes is not critical
slide-20
SLIDE 20

Medical Radiation Physics, Clinical Sciences Lund, Lund University

3D-ID

slide-21
SLIDE 21

Medical Radiation Physics, Clinical Sciences Lund, Lund University

RMDP (ROYAL MARSDEN)

slide-22
SLIDE 22

Medical Radiation Physics, Clinical Sciences Lund, Lund University

THE LUNDADOSE

Include also methods for 3D SPECT/CT based dosimetry

slide-23
SLIDE 23

Medical Radiation Physics, Clinical Sciences Lund, Lund University

Medical Radiation Physics, Lund Katarina Sjögreen-Gleisner, Michael Ljungberg, Karin Wingårdh, David Minarik, Sven-Erik Strand, Tomas Ohlsson Department of Oncology, Lund Ola Lindén, Jan Tennvall

EXAMPLE: 3D BASED DOSIMETRIC STUDY IN LUND

High Dose Zevalin™ antibodies Non-Hodkin’s Lymphoma Dosimetry study - 111In Therapy study – 90Y Bremstrahlung Imaging Bone-marrow support Administered activity based on absorbed dose (Gy) Maximum Tolerated Dose (MTD)

slide-24
SLIDE 24

Medical Radiation Physics, Clinical Sciences Lund, Lund University

PROCEDURE

Seven multiple 111In SPECT/CT studies and WB studies SPECT Reconstruction with OSEM iterative reconstruction algorithm

  • Patient-specific correction for attenuation,

scatter and collimator response

Image Registration

  • CT-CT Based Non-Rigid Registration over multiple measurements.
  • CT-CT transformations applied to SPECT-SPECT

Voxel based Absorbed Dose Distribution

  • Local absorbed energy (90Y)

LundAdose Software – written in IDL

slide-25
SLIDE 25

Medical Radiation Physics, Clinical Sciences Lund, Lund University

ABSORBED DOSE IMAGES

90Y absorbed dose

estimated from multiple quantitative 111-In images Absorbed dose by numerical integration voxel-by-voxel after image registration Masses from CT images

slide-26
SLIDE 26

Medical Radiation Physics, Clinical Sciences Lund, Lund University

FUSED IMAGES

90Y absorbed dose

estimated from multiple quantitative 111-In images Absorbed dose by numerical integration voxel-by-voxel after image registration Masses from CT images

slide-27
SLIDE 27

Medical Radiation Physics, Clinical Sciences Lund, Lund University

REGIONS OF INTEREST

slide-28
SLIDE 28

Medical Radiation Physics, Clinical Sciences Lund, Lund University

DOSE‐VOLUME HISTOGRAM

Dose calculations

  • Average
  • Median
  • Average of 75%
  • r above
  • Average of 90%
  • r above

Gy / MBq Estimation of necessary therapy activity is based

  • n this value.
slide-29
SLIDE 29

Medical Radiation Physics, Clinical Sciences Lund, Lund University

RADIOBIOLOGICAL PARAMETERS

Iterative Methods preferrable since they allow for correction of

Attenuation and Scatter Collimator Response Septal Penetration Partial-Volume Effects

Iterative Methods preferrable since they allow for correction of

Attenuation and Scatter Collimator Response Septal Penetration Partial-Volume Effects

Dose Calculation by

Dose Kernels Monte Carlo

Dose Calculation by

Dose Kernels Monte Carlo

Segmentation DV Histogram Segmentation DV Histogram Obtain biokinetics Obtain biokinetics Radiobiological Parameters Hybrid SPECT/CT Hybrid SPECT/CT

slide-30
SLIDE 30

Medical Radiation Physics, Clinical Sciences Lund, Lund University

RADIOBIOLOGICAL PARAMETERS

RBE-weighted Dose

  • Consider differences in

Radiobiological effect for different types of radiation

Biologically Effective Dose

  • Accounts for dose rate

variations in radionuclide therapy

  • Based on Linear-Quadratic

Response Model

  • Include α/β and rate-of-repair
  • f sublethal damages.

Equivalent Uniform Dose

  • Spatially varying absorbed

dose distribution converted into an equivalent uniform absorbed dose that yield a biologic response similar to that expected from the nonuniform dose distribution.

Isoeffective Dose

  • Equivalent absorbed dose of

low-LET radiation that when delivered would produce the same clinical effects as the high-LET treatment.

"MIRD pamphlet No. 21: a generalized schema for radiopharmaceutical dosimetry--standardization of nomenclature." J Nucl Med 2009 50(3): 477-484.

slide-31
SLIDE 31

Medical Radiation Physics, Clinical Sciences Lund, Lund University

PROBLEMS WITH MACROSCOPIC MONTE CARLO

Even if a voxel‐based Monte Carlo dosimetry approach that is properly done is a improvement over organ‐based dosimetry – we still do not know the distribution of activity within the the target voxel!!!

slide-32
SLIDE 32

Medical Radiation Physics, Clinical Sciences Lund, Lund University

CONCLUSION

Dosimetry in Nuclear Medicine is a real challenge ☺ New hybrid SPECT/CT systems have significantly improved the accuracy in activity quantitation Dosimetry on SPECT voxel level possible Inherent spatial resolution problem may need connection to models describing activity distribution of small-scale levels Radiobiological parameters will complement the physical ’absorbed dose’

slide-33
SLIDE 33

Medical Radiation Physics, Clinical Sciences Lund, Lund University

THANK YOU FOR YOUR ATTENTION

Lund University, Sweden