HIT Betriebs GmbH am Universitätsklinikum Heidelberg mit beschränkter Haftung www.med.uni-heidelberg.de/hit

Imaging techniques for in-vivo

treatment verification

Katia Parodi, Ph.D.

Heidelberg Ion Therapy Centre, Heidelberg, Germany

Previously: Massachusetts General Hospital and Harvard Medical School, USA,

and Research Center Dresden-Rossendorf, Germany

Hadron Therapy Workshop 2011

Erice, Italy, May 25th, 2011

Massachusetts General Hospital

and Harvard Medical School

The physical advantages of ion beams The finite range with the characteristic “Bragg-peak”

Depth in water (cm)

Peak-region

Photons

Bragg-Peak

Protons 12C-ions

Plateau-region

Treatment uncertainties in ion beam therapy

Difference TP / delivery

Daily setup variations

Internal organ motion

Anatomical / physiological changes

TPS dose calculation errors

Inhomogeneities, metallic implants

Daily practice of compromising

dose conformality for safe delivery

Conversion HU in ion range

CT artifacts

After Enghardt 2005

Accounting for uncertainties

in the clinical practice

Current approach:

Opposed fields,

overshooting

Protons

Desirable approach:

Different beam angles and

no overshooting

A. Trofimov et al, MGH

? In-vivo verification

Positron-Emission-Tomography (PET)

1) b+-decay A(Z,N) A(Z-1,N+1) + e+ + ne

2) Moderation of e+ in medium (typically few mm in tissue)

e+

Imaging of b+-activity

e-

Eg = 511 keV

g

g

180°

3) Annihilation into 2 opposite g-rays (511 keV each)

Coincidence

processor Image reconstruction

4) Coincident detection and processing

Detector

b+-emitter

or prior to irradiation with same radioactive beam (planned at HIMAC, Japan)

HIMAC, Japan

PET imaging for verification of ion therapy

Injected to the patient via irradiation using primary b+-radioactive ions like 19Ne (T1/2 17s), 11C (T1/2 20 min) and 10C (T1/2 19 s)

Low dose exposure prior to therapy with stable beam (pioneered in 70s at LBL, USA)

In-situ, non-invasive detection of b+-activity

19Ne from 20Ne

LBL, USA

Not (yet?) in clinical routine use

Llacer et al, Nucl. Sci. Appl. 3 (1998)111; Kitagawa et al, Rev. Sci. Instrum 77 (2006)

11C

12C ions in PMMA (A) (D)

PET imaging for verification of ion therapy In-situ, non-invasive detection of b+-activity Formed as by-product of irradiation in nuclear fragmentation reactions

(11C [T1/2 20 min], 15O [T1/2 2 min], …)

Schardt et al, Rev Mod Phys 2010; Parodi et al, IEEE TNS 2005; Enghardt, … Parodi … Nucl Instrum Meth A 2004

g-emission

12C

11C, 10C

15O, 11C, ...

nf ≈0

A(r) D(r)

Dose-guidance from PET surrogate

by comparing measured b+-activity

with expectation as done at GSI

g-emission g-emission (“prompt”)

≠ annihilation

11C 11B+ e+ + ne

T1/2

Eg = 511 keV

<~180°

e+ e-

Annihilation g-rays

g

g

In-beam PET for scanned 12C therapy at GSI

Planned

dose

Once

Measured

b+-Activity

For every fraction

(typically 20 d @ 1Gy)

MC calculated

b+-Activity

Verification of

Beam range

Lateral position

In case of deviation

Timely reaction

Enghardt, … Parodi … Nucl Instrum Meth A 2004; Parodi et al Nucl Instrum Meth A 2005

> 400 patients

Time in s

Beam off (PET signal)

Beam on (noise)

In-beam PET for scanned 12C therapy at GSI

1998

Prediction

Measurement

Since 1999

Prediction

Measurement

Extraction of ion range in-vivo

Validation of the physical beam model of treatment planning

1. Precision measurements:

Range of 12C-Ions in tissue

(D. Schardt et al. GSI) 2. Modification:

R = R(HU)

(E. Rietzel et al. GSI)

Original-CT Modified CT

Fast PET recalculation

b+-activity: prediction b+-activity: measurem.

Hypothesis on the reason for the deviation from the treatment plan

Dose recalculation

Modified CT Original-CT

Original-CT Modified CT

Interactive CT manipulation

New CT CT after

PET findings

Parodi Ph.D. Thesis, 2004; Enghardt, Parodi et al, Radiother Oncol, 2004

Indirect PET-guided dose quantification

Indirect estimation of 12C dose deviation from in-beam PET

g-emission

12C

Parodi et al PMB 2002, Parodi et al IEEE TNS 2005

PET monitoring of proton therapy?

Ap ~ 3 A12C

at same range and dose

(but ~102 lower than in nuclear

medicine PET for typical

therapuetic ion doses)

12C ions

11C, 10C

15O, 11C, ...

In-beam phantom (PMMA)

experiments at GSI

(A) (D)

protons

15O, 11C, ...

(A) (D)

Proton

(Projectile fragmentation only for Z>1)

g-emission

Offline PET/CT for scattered p therapy at MGH

Passive beam delivery at MGH Boston

Proton Irradiation… 14-20 min elapsed…

PET/CT @ MGH Radiology

Offline PET/CT

Only long-lived isotopes

(11C: T1/2 20 min

15O: T1/2 2 min)

Full ring tomograph

CT for co-registration

Phys. MC PET

Bq/ml

Clinical case of clival chordoma

Field 1: 0.87 Gy, DT1 ~ 26 min

Field 2: 0.87 Gy, DT2 ~ 16 min

Parodi et al Int J Rad Oncol Biol Phys 2007

Offline PET/CT for scattered p therapy at MGH

TPS

Field 1

Field 2

mGy

PET/CT Meas.

Bq/ml

MC dose

mGy

Range monitoring: possible in well co-registered low perfused tissues

Challenges: washout, S/N, and (extra-cranial sites) motion, registration

MC PET + washout

Bq/ml

# of patients Dose / field [GyE]

head 12 0.9-3

eye 1 10

C-spine 3 0.6-2.5

T-spine 1 1.8

L-spine 2 0.9-2

sacrum 2 1-2

prostate 2 2

TOTAL 23 0.6-10

Offline PET/CT clinical experience at MGH

Parodi et al, IJROBP 68, 2007; Knopf, Parodi et al, PMB 54, 2009; Knopf, Parodi et al, IJROBP 72, 2011

Direct PET-guided dose quantification Mathematical formalism towards dose deconvolution in proton therapy

Planned dose Filter-PET

11C activity

+ washout

Convolution with filter functions (cross-section dependent)

Deconvolution would enable direct dose quantification from measured

PET images, but issue of statistical noise, washout, motion

Parodi and Bortfeld PMB 2006; Parodi et al AAPM 2006; Attanasi, … Parodi … IEEE 2009

Need for improved imaging strategies

MC-PET

(offline PET/CT scan)

b+-activity

+ washout

Short delay DT improves S/N, reduces washout

Short scan time tmeas minimizes motion artifacts

and maximizes patient throughput

Towards better imaging strategies

Shakirin, … Parodi … PMB 2011; Parodi et al IJROB 2008; Parodi et al NIMA 2005

But optimal solution depends on

Development and integration efforts

Patient throughput in treatment room

Beam macro- and micro-structure

Worldwide active research on novel

dedicated in-beam PET scanners

Time in s

Beam on (noise)

Beam off (PET signal)

-200 -190 -180

Single g-RF time correlation experiments at GSI

Random correction failure due to „prompt“

(sub-ns) radiation correlated with RF (problem even worse for cyclotron)

Dedicated data acquisition needed (Enghardt, Crespo, Parodi, Pawelke, patented)

Distance between two opposing detector heads of 30 - 100 cm

Icentric rotating of 0 -360 deg.

Position resolution of 1.6-2.1 mm FWHM

Detection area of 164.8×167.0 mm2

Novel PET systems for in-room imaging

Courtesy of T. Nishio NCC Kashiwa, Nishio et al IJROBP 2010

Dual-head scanner mounted on rotating gantry in Kashiwa, Japan

- Planar imaging starting immediately after end of irradiation (cyclotron)

- A(r) ≠ D(r): Daily measurement compared to reference activity (reproducibility check)

Planning dose daily activity Reference activity

Similar finding as for GSI (e.g., detection of anatomical changes)

- > 50 patients of H&N, Liver, Lung, Prostate and Brain from 2007/10

Closeby PET/CT at HIT Heidelberg

Newly installed PET/CT next to the treatment rooms

Tx room

Tx room

Tx room

PET/CT

Biograph mCT

Combs,…, Parodi MIRANDA clinical study; Visualization / Analysis tools within BMBF project DOTMOBI

Establishment of clinical workflow at HIT

Adaptation of MC to handle facility-, patient- and

plan-specific information for automated dose calculation

Extension / validation of MC

for calculation of

b+-emitter yields

Unholtz, …. Parodi, DGBMT 2010; Bauer, …., Parodi, PTCOG 2011

Sommerer et al, Rad Oncol 2010

supported by EU-project PARTNER

Merging MC utilities for

patient calculations

of dose and PET

GUI for automated simulation

data management,

visualization,

data exploration and analysis

Establishment of clinical workflow at HIT

Unholtz, …. Parodi, DGBMT 2010; Bauer, …., Parodi, PTCOG 2011

Implementation realized via

„SimInterface“

(Software developement within the BMBF DOTMOBI project)

Towards 4D PET-guided in-vivo verification

Static absorber

Dynamic wedge

Moving target

(PMMA)

Motion sensor

Dipole magnets

Parodi et al Med Phys 2009, in collaboration with GSI (Bert), FZD (Enghardt), SAG (Rietzel), patent pending

12C ion tracking experiment with time-resolved in-beam PET at GSI

Planned dose

Planned delivery of homogeneous extended dose

Irradiation to static (ref.) and moving (~3cm in 1.5s) target

Correlation of dynamic PET acquisition with motion

Target

Absorber

Wedges

Moving

platform

12C beam PET

Dipole magnets

Static absorber

Dynamic wedge

Moving target

(PMMA)

Motion sensor

Proof-of-principle of 4D in-beam PET

Static reference

3D PET (motion uncorrected) m

oti

on

Static reference

3D PET (motion uncorrected) 4D co-registered PET

Ongoing / planned investigations

Experiments at GSI / HIT to compare in-

beam vs offline PET for monitoring of

motion-mitigated ion beam delivery

(gating, tracking,…)

Explore benefits from advanced internal

motion sensors (ultrasound-based)

Extrapolation to clinical cases

Parodi et al Med Phys 2009; collaboration HIT/GSI/FZD/Siemens/Mediri funded by EU Project ENVISION

4D co-registered PET

mo

tio

n

Real-time prompt gamma imaging

Testa et al, Applied Physics Letters 93 (2008)

Carbon ions in PMMA

Protons in water

Experimentally verified correlation

between 90 angled prompt g profiles

and p / 12C ions ion range

g-emission nf ≈0

g-emission g-emission (“prompt”)

≠ annihilation

(Z>1) 12C or p

Promise of real time in-vivo range

verification insensitive to washout

Challenge of efficient detector solution

(Anger or Compton camera)

=> Next talks!

Pre- (intra-?) treatment ion-based imaging Imaging residual range of high energy transmitted ion beams for

- Validation of CT-range calibration curve

- Assessment of range variations in motion cycle

- Low dose verification of patient position at the treatment site

Pioneered since the 60’s, but no routine clinical application yet

Issue of high energies required and Coulomb scattering (esp. for protons)

Several groups are now working on

small scale systems based on (single)

ion tracking and residual range

measurement via range telescopes or

thick energy detectors

Major efforts towards tomographic

imaging to eventually replace X-ray

CT for treatment planning

Proton radiograph of a phantom measured

with in-house developed prototype at PSI

Schneider et al 1990s, www.psi.ch (Paul Scherrer Institute, Switzerland)

Towards ion radiography / tomography at HIT

• Scanning 0°-180° in steps of 5° 12C pencil-beam 400 MeV/u

3.5 mm Gaussian FWHM

5 x 106 pps

• PMMA phantom D=160 mm

tissue equivalent rods d=28mm

• Multi-channel electrometer

electronics highly integrated

Proof-of-principle 12C

Heavy Ion Tomography

Rinaldi Ph.D. research at HIT/DKFZ (in collaboration with B. Voss, GSI); Voss et al GSI Report 2010, in press

• Simple 2D back-projected

reconstruction

Stack of ionization chambers

(Voss et al, GSI) with new

electronics

Post-treatment magnetic resonance imaging

Radiation induces fatty tissue replacement of vertebral bone marrow

Krejcarek et al IJROBP 2007

Pre-Tx MRI Planned dose Post-Tx MRI (1 month)

Investigations on using this signal as range indicator of TOTAL dose

delivery (not for single fraction) for population-based assessment

The time scales of in-vivo imaging

techniques for in-vivo range verification Time Irradiation

Pre-treatment (-DT ~ min)

Radioactive ion (RI) beams

Ion radiography / tomography

DT

In-beam „real-time“ (DT << ms)

Prompt gamma, emitted particles

In-beam „delayed“ (DT ms - min)

Positron-Emission-Tomography (PET)

Shortly after treatment (DT ~ 2 – 10 min)

In-room or nearby PET(/CT)

Long after treatment (DT ~ 10 – 20 min)

Offline PET(/CT)

Long after therapy (DT days - weeks )

Magnetic Resonance Imaging (MRI)

First or each fraction (Fx)

Each Fx

Each or

selected Fx

Selected Fx

Sum

of Fxs

Each Fx

Conclusion

Full clinical exploitation of ion therapy promises requires

In-vivo imaging of “surrogate” signal, e.g., from escaping secondary

radiation or physiological changes correlated to range / dose

Reliable computational tools for accurate modeling of the

“surrogate” signal in relation to the range / dose deposition

PET is a mature imaging technique for in-vivo treatment verification,

however technological / methodological improvements desirable

In-beam PET would be the method of choice but requires dedicated,

expensive instrumentation: industrial partners?

Alternative or complementary (time scales!) techniques based on

emitted / transmitted radiation or MRI under development / investigation

Outlook

R&D in modeling, detector development, exp. validation,

clinical integration, depending on beam production / delivery

(beam time structure, background radiation, …)

X-ray source

X-ray

imager

Ion gantry

Motion sensor

Transmission (ions, novel X-rays?)

(ions, novel X-

rays?)

Synergy between imaging for (real-time?) verification of

- Patient / tumour position (image-guided-radiotherapy)

PET Prompt g

- Range / dose delivery (towards dose-guided radiotherapy)

Thank you for your attention

The MC-modeling and in-vivo imaging research group at HIT:

J. Bauer, C. Kurz, A. Mairani (now CNAO), I. Rinaldi, F. Sommerer, D. Unholtz

The colleagues at HIT, Universitätsklinikum Heidelberg and DKFZ:

J. Debus, S. Combs, T. Haberer, O. Jäkel and Medical Physics Group

J. Engelke, M. Martisikova

Collaborators / former colleagues:

A. Ferrari, F. Cerutti, CERN Geneva

D. Schardt, C. Bert, B. Voss, GSI Darmstadt

T. Bortfeld, H. Paganetti, MGH Boston

W. Enghardt, F. Fiedler, K. Laube, HZDR Dresden

F. Attanasi, INFN Pisa

Funding:

FP7 EU Project PARTNER

FP7 EU Project ENVISION

BMBF Project DOT-MOBI

Acknowledgements

HIT Start of clinical operation on

15th November 2009

To date: > 400 patients treated