Eros Pedroni Paul Scherrer Institute SWITZERLAND Zurzach 10.01 · E. Pedroni Center for Proton Radiation Therapy - Paul Scherrer Institute - Proton Therapy Winter School 10 -01-2010

E. Pedroni Center for Proton Radiation Therapy - Paul Scherrer Institute - Proton Therapy Winter School 10 -01-2010 1

Zurzach10.01.2010

Active scanning beams: 1. Modulating delivery

Eros Pedroni

Paul Scherrer InstituteSWITZERLAND


1. Beam delivery options 2. Scanning technology: experience with Gantry 1

3. Dose precision issues 4. Practical advantages of scanning

5. The organ motion problem – link to the next talk “new developments”

of

Wednesday 6. Questions

Summary


1. BEAM DELIVERY OPTIONS FOR SCANNING


•

Typical “physical”

scanning beam–

x= ±3 mm θ= ±10 mrad δp/p

= ±0.5%

1.1 The basic element –

the proton “pencil”

beam

•

Common

to all delivery methods–

Pencil beams with zero phase space and given initial energy

1 mm sigma

3 mm sigma

10 mm sigma

Monte Carlo

Scanning

Wobbling

After integration in the lateral direction the differences in depth disappear

•

Difference

between beam delivery methods?–

The “cumulated”

phase space

at

nozzle exit

(beam formation)–

The time structure

of the beam

and beam delivery sequence•

Which is relevant for

organ motion errors


1.2 Options for beam spreading in the

lateral direction

•

Scattering (static) –

Constant particle fluence

(homogeneous dose field)

•

Single scattering (good penumbra –

but low efficiency)•

Double scattering (higher efficiency –

less good penumbra)

•

Contoured 2nd scatter (or double contoured to compensate energy mixing)–

Field size and depth dependent solutions

•

Uniform scanning (dynamic) •

Circular (wobbling) -

rectangular–

spiral –

BEV shapes

•

Conformal scanning (dynamic)–

Modulated particle fluence

(for 3d-conformation and IMPT)

•

Beam motion

magnetic•

Or through patient table motion

–

Performance issues:•

Precision of the dose shaping

•

Scanning speed –> repainting capability –> to reduce organ motion

sensitivity


Lateral scanning options

•

Spot scanning–

Switch beam OFF in between spots

–

PSI Gantry 1–

Let beam ON in-between spots

–

GSI raster scanning•

Continuous scanning–

Magnet-driven scan

•

Dose shaping by changing the magnet speed (at constant dose rate)

–

Time-driven scan•

The beam moves with maximum speed –

the dose is painted by

modulating the beam intensity–

PSI Gantry 2

(feasibility ?)

Homogeneous energy layer

Non-homogeneous energy layer


1.2 Options of beam formation in depth (Bragg peak)

•

Rotating wheel (very fast -

200 ms / cycle)–

Fast SOBP mixing (quasi-static)

•

Must be designed for each –

energy -

SOBP extent -

field size–

Not of interest in combination with scanning

•

Ridge filter –

(miniature structure blurred by angular confusion)–

SOBP energy mix (static)

•

Must be designed for each -

energy -

SOBP extent -

field size–

Concern –

angular confusion of the beam -

penumbra

•

Range shifter

(dynamic –

30 ms) -

> see PSI Gantry 1–

Plates (regular or digital) -

moving wedges -

water column

•

Range steps or continuous–

Concern –

beam spot broadening in the air gap

to the patient

•

Variable energy of the beam

(dynamic) -> see PSI Gantry 2–

Range steps

–

Concerns –

very steep Bragg peaks at the lowest energies•

Best in combination with a ridge filter or preabsorber?


•

Range shifter

-> PSI Gantry 1–

Variable amount of material in the beam

–

Water-equivalent?•

Material, geometry, density of the plates

•

With zero air gapPencil beam size invariant with range

–

With non-zero air gap

-

problematic the •

Spot broadening due to MCS in the RS

•

Dynamic beam energy

-> PSI G2–

With degrader

and beam line

•

See “future developments talk”


1.3 Relation between proton energy and fluence

for conformal dose shaping

Depth

Starting example : hat boxUniform proton fluenceFixed SOBP

Target

Gray shading = density of proton stops

Patient

Collimator

100%(0%)

100%(80-50%)

Fixed range modulation =unnecessarydose delivery

BEAM


3d or 4d-shaped dose conformal scanning

Intensity modulation within energy-layerDelivery spot by spot

Collimator compensator scattering-wobbling

Collimator

OAR

Compensator

Stacking of energy-layersVariable range modulation

OAR

Compensator

MLC Shape shrinkingUniformScanning

Scanning:Freedom to deliverany pattern

UnnecessaryDose delivery


2. PRACTICAL DEVELOPMENT OF SCANNING THE PSI EXPERIENCE WITH GANTRY 1


2.1 PSI Scanning

•

Development started in 1989–

With limited resources

–

Compromises due to the limited space of the area

–

Parasitic use of the PSI beam•

With very long shut-downs


The decision to build Gantry 1 was taken in 1992

•

Until May 2008 …the only gantry with scanning (1st patient in 1996)–

Magnetic scanning started before the last bending magnet

•

parallel scanning

(but only along one magnetic axis) •

gantry radius

reduced to only 2m

–

Eccentric mounting of the patient table on the gantry front wheel•

Patient moves away from the floor when treating with beam from below (a drawback)

α rotation

φ rotationβ rotationUpstream scanning –

eccentric design


If I could do it again …

… Eccentric compact Gantry 2

•

Eccentric mounting as with Gantry 1 (R = 2m)–

But with rotation only on one side (0°

180°)

-

as with the new Gantry 2

–

With a counter-rotation–

False floor underneath the patient table moving with the gantry

360° 180°


X Sweeper magnet most often used

Y Range shifter 2nd

loop

–

Gaussian pencil beam of 3 mm sigma–

Cartesian scanning (infinite SSD)

–

“Step and shoot”

–

spot on a 5 mm grid

Z Patient table slowest loop

Time

Spot-Dose Monitor + Fast Kicker

The sequence of the elements of scanning:

Discrete pencil beam scanning

2.1 Scanning on the PSI Gantry 1

–

Weak point: transverse scanning by moving patient table–

Slow motion ( no repainting possible)–

We can treat only non moving targets

–

Head, spinal chord and low pelvis


Dimensional considerations for scanning

•

Reference size of 1 liter (very often much less, our max value 4

liters)–

Assumed beam size: 3 mm sigma (see next section)

–

Derived grid size: 5 mm

(21 spot lateral -

23 in depth)•

21 x 21 x 23 ~ 10’000 spots/liter

•

21 x 21 x 10 cm = 44 m path length–

Assumed treatment time: 1.5 minute

beam-ON time

•

10 ms/spot in average–

Due to the non-uniform spot weights (for a uniform SOBP)

•

Most distal spots: 60 ms•

Most proximal: 3 ms

(with beam spot weight optimization: we accept spots >= 0.5 ms)•

Treatment roughly proportional to -

VOLUME (the moving time)

and to BEV SURFACE (beam ON time of the high weighted distal Bragg peaks)•

Required intensity: 0.2 nA proton current–

beam ON time of 1 Min to deliver 1 Gy

in1 Liter


Beam monitoring

•

Transmission ionization chambers (M1

and M2) –

proton flux–

5 (10) mm

in air –

2kV

voltage

–

response time < 100 μs

–> precision of controlling the dose •

1% of the mean spot time

•

(switching time of the kicker 50 μs)•

Delay of the current measurement and of the kicker subtracted in

advance from

preset -> 0.2% of mean spot time•

Strip-monitor chamber –

4 mm strips

–

Measure position and width of the beam after the delivery of each spot

–

Position resolution: < 0.5 mm–

Charge collection time ~ 0.8 ms

•

Wait 1 ms before reading scalers

at the end of the spot


The reasons to switch off the beam in-between spots

•

Avoid dose uncertainties when stepping with beam ON to the next spot (transients)–

Dose errors (sweeper power supplies delay and non linearities)

–

Errors in checking the beam position with the strip monitor (time resolution of 0.8 ms at 1 cm/ms => 8 mm error)

•

Poor quality of the beam until 2006 (before COMET) –

Beam splitting

( 0.5% intensity) from the 2 mA

beam of the PSI ring

cyclotron and degrading it from 590 MeV down to 100-200 MeV•

After any repair -> bad vacuum -

> intensity spikes

•

Check monitor units precisely at the end of spot •

Conservative strategy –

Perform all calculations at the end of a static spot

–

Start next spot only when previous delivery shown to be correct -> overall dead time of 5 ms in-between spots

•

Best approach to control the dose with repainting

Beam IC

SweeperHall probe


•

Sweeper magnets–

30 ms for a full sweep of 20 cm

–

For a 5 mm step•

Time to move the beam and to stabilize: 3 ms

•

< time to check previous spot: 5 ms

(Gantry 2 2ms)•

Range shifter–

40 plates (80 pneumatic valves)

•

4.5 mm thick each + one half plate •

Water equivalent arrangement

–

Dead time 50 ms

(30 ms for motion)•

Patient table–

Moves in steps of 5 mm

•

1-2 seconds

dead time per step•

Acceleration and decelaration

•

Smooth motion for patient comfort

Scanning motion devices


The price to pay: the overall scanning dead time

•

Gantry 1–

Beam off after delivery of each spot

•

Sweeper dead time 10’000 x 5 ms = 50 s •

Range shifter 21x21x 50 ms= 22 s

•

Patient table 21 x 1s = 21 s–

1.5 minutes beam off vs. 1.5 minute beam on

•

Duty factor of our present discrete scanning low 50%•

Precision of dose delivery very good

–

Dose reproducibility of ~ 0.2%

•

Future: exploring more efficient solutions with Gantry 2–

Painting of lines instead of spots (check of dose delivery at the end of a line)

–

-> see “future developments”

presentation


•

Beam delivery–

Small systematic errors in the sweeper calibration

–

Relative beam position precision of±

0.2 mm from one spot to next

•

Target motion during beam delivery (patient)

•

Statistical error–

Gets reduced by

•

sqrt( N*m*k)•

N= number of fraction

•

m= number of fields•

k= repainting number

•

Practical limit for Gantry 1–

Motions < ±

1-2 mm (including full fractionation N=30)

The required precision of placing the spots

Organ motion error is a function of ratio Motion size / pencil beam size


•

The reproducibility of the beam tunes–

Automatic set-up of the beam energy without retuning the beam during treatments(cyclic ramping of the gantry magnets)

•

The beam appears at the correct position within 1.5 mm•

Position correction at the end of the first spot–

Correction allowed if within ±1.5 mm

–

All further spots•

Position deviation within ±

0.5 mm

of the expected value–

Interlock if deviation > 1.5 mm

•

On-line and off-line analysis of delivered spots–

Analysis of logged data

U sweeper position

Beam position error

1 mm

The precision of setting the beam (beam tunes)


3. DOSE SHAPING PRECISION OF SCANNING


3.1 Distal fall-off

Desirable Δp/pAt highest energy ~0.2% FWHM At lowest energy ~2%

FWHM

Momentum band dependence

•

Function of energy (linear with range)–

Range straggling

(σR/R ~1%) unavoidable

–

Momentum band

avoidable

Energy dependence

Very sharp Bragg peaksReason to use a pre-absorber

at low

energies (for variable beam energy)

302010 cm


Distal fall-off

•

Scanning with variable energy–

In principle superior to scattering

•

No material in the beam (no compensator –

no range shifter)•

Minimalst

energy for the required range (physical limit)

•

Less range straggling•

Sharper distal fall-off

•

In practice the advantage compared to scattering is probably very marginal


3.2 Lateral fall-off

•

Governed by the spot size within the patient (dose spot) –

… at the Bragg peak (the major effect)

•

Derived from –

The beam broadening due to the Multiple Coulomb Scattering in the patient

•

Given by the physics –

unavoidable•

In doesn't help to use a beam size which is much smaller than this effect

–

The beam size in air

(size of the beam at the exit of the nozzle)•

Given by the beam delivery system by design

–

Accelerator -

beam line -

and the material in the nozzle

–

The smaller the dose spot -> the sharper the lateral fall-off•

Sigma lateral fall-off ~=~ sigma dose spot


Optimization of the beam size (spot size) •

Beam in vacuum–

Provide a beam source with small phase space

•

Beam in air–

Bring the vacuum very close to the patient

•

Beam monitors (material in the nozzle)–

Place material close to the nozzle exit

–

Reduce amount of material

(of low atomic number)–

Reduce air gap to the patient !!!!

Size = θ

x Distance θ

= F(E) x sqrt

(Material

/ S)

Beam pipe

•

Range shifter or pre-absorber –

Reduce air gap to the patient !!!!

Nozzle

•

Patient –

MCS–

You can do nothing about this


The air gap problem when using a range shifter

119 MeV 177 MeV 214 MeV

Gap(cm) Gap(cm)

FWH

M (c

m)

FWH

M (c

m)

0 0 0

10

10 10

20

2020

3030

–

Beam blow-up due to MCS in the range shifter in front of the patient–

The reason with Gantry 2 to go for variable beam energy

–

Strategy of positioning beam modifiers in the beam•

Either very close

to the patient

Small air gap

•

Or very far

Loss of intensity•

But not in-between

The worse you can do

–

Similar problem with scattering –

air gap to compensator

–

lateral penumbra

Gap(cm)

FWH

M (c

m)


Patient MCS

•

MCS in the patient PMCS•

Depends linearly with range–

Rule of thumb 1 cm FWHM at 20 cm

1.5 cm FWHM at 30 cm

•

Choice of the beam size (BS) in air –

Sum in quadrature

of the beam size in air (BS) and the MCS in the patient (Pmcs)

•

if we choose BS < 50%

Pmcs

we obtain sqrt( Pmcs^2 + BS^2) < 1.12

Pmcs–

A too small beam size -> needs smaller scan grid –> more spots -> more dead time

•

and is more sensitive to organ motion errors


“Edge enhancement capability”

of scanning

•

Delivery of separated spots–

Variable choice of intensity

–

The dose lateral fall

can be made to be similar to the fall-off of the original beam Gaussian

•

Uniform fluence

of spots–

The case of collimation

–

Gaussian

folded with step function = error-function

•

Max difference–

Factor 1.7

(1.4)

•

Scanning with optimization can produce a sharper lateral fall-

off as compared to scattering


Beam width at the Bragg peak as a function of the range

–

Assumed phase space: 6 mm FWHM

constant in vacuum at all energies•

Relevant for organ motion

errors

practical limit -

beam line

–

Nozzle material: as for

Gantry 1

practical limit -

monitoring–

MCS in the patient enhanced by factor a 1.4 in favor of scanning

•

Collimation improves precision only at ranges below 11 cm

Factor 1.4 higherdue to the error-

function

Scanning with collimationbetter

Scanning alone better

Idealized scattering(zero phase space)Realistic scanning


3.2 Lateral fall-off

•

Scanning with variable energy (with or without collimator)–

In principle superior to scattering

•

No material in the beam (no compensator –

no range shifter)–

No MCS propagating in the air gap –

sharp beam

–

Less angular confusion –

sharp shadow when using collimators


SCANNING ONLY APERTURE

CONTOURED APERTUREOAR-

SHIELDING BLOCK

Options for “collimated scanning”

Scanned field

Sensitive structure

LESS WEIGTH

BETTER VISIBILITYFOR BEV X-RAYS


Possible scanning patterns

Criteria

Precision (edge of the field)

Duty factor (beam-off time)

“Topological”

scan (repainting capability)

Spot scanning (G1)

Spot scanning++ (G2) Line scanning (G2)

Raster scanning (GSI)

Contoured scanning (G2)IM

IM


4. ADVANTAGES OF SCANNING


All done by software –

with minimal equipment

•

No need to use individualized hardware–

Avoid fabrication and mounting of patient specific equipment in the nozzle

•

Apply dose fields sequences in one go without personnel entering in the treatment room

•

To reduce treatment time•

All fields of IMPT are delivered in the same fraction

•

Most efficient use of the beam–

“All”

used protons reach the target

•

Minimal neutron background

(for the patient) by default•

Less activation of equipment

(for the personnel) by default

•

Flexibility to treat from small to very large fields without changing equipment–

“All done by the beam”

•

Sharper penumbra than scattering (scanning alone or collimated) (Gantry 2)–

Less material before the patient (no compensator)


Variable modulation of the range

•

For avoiding unnecessary 100% dose on the healthy tissues–

Especially relevant for large tumors


4D –

(dose-modulated fields)

•

Dose tailored geometrically in 3d•

Dose shaping within the target–

Used for

•

Intensity modulated therapy

IMPT•

Biological targeting

•

Competition with IMRT


Other possible scanning related “advantages”

•

Gantry design with “upstream scanning”–

Reduce the radius of the gantry

•

No additional radial distance for the spreading the beam–

Parallelism of scanning

–source at infinite distance

•

Simplify dosimetry –

treatment planning –

field patching -

collimation and compensation

•

Capability to simulate scattering (repainted BEV box scans)–

Scanning can simulate and improve scattering (variable range modulation)

•

The opposite is probably not true–

Provide backward compatibility with the more established techniques using one and the same nozzle (see future developments -> Gantry 2)

•

The only drawback compared to scattering–

The SENSITIVITY TO ORGAN MOTION ERRORS

–

A problem common to all dynamic beam delivery methods including IMRT


5. THE ORGAN MOTION PROBLEM


5.1 An unsolved problem: organ motion errors of scanning

•

Disturbance of the lateral dose fall-off (same problem for scattering and scanning)

•

Add safety margins or•

Reduce with Gating

or

Tracking

•

Disturbance of the dose homogeneity–

Scattering –

highly repainted -

insensitive

–

Single painted scanning -

very sensitive–

Repainted scanning

-

acceptable

•

Alone

–

for medium motion•

With Gating

or Tracking

–

for large motion

•

The experience of treating moving targets with scanning is still inexistent–

WE HAVE TO LEARN HOW TO DO THAT

•

Prospective solutions …

see next talk on future developments with Gantry 2


THANK YOU

A very exciting field …

Based on a beautiful idea

The next step –

scanning speed

Documents

Eros Pedroni Paul Scherrer Institute SWITZERLAND Zurzach 10.01 · E. Pedroni Center for Proton Radiation Therapy - Paul Scherrer Institute - Proton Therapy Winter School 10 -01-2010