Proton CTWill finding a challenging solution solve a challenging problem?

Nigel M Allinson, MBE, ScD, FIET, CEng

Proton Radiotherapy Verification and Dosimetry Applications

Funded by

STFC Global Challenge Network+ in Advanced Radiotherapy – 22 April 2016

• University of Lincoln • University of Birmingham • University of Liverpool • University of Surrey • University of Warwick• University of Cape Town

• National Research Foundation (NRF) - iThemba LABS, SA• University Hospital Birmingham NHS Foundation Trust • University Hospital Coventry and Warwickshire NHS Trust • United Lincolnshire Hospitals NHS Trust • The Christie NHS Foundation Trust

• ISDI: Image Sensor Design and Innovation Ltd• aSpect Systems GmbH

• Elekta AB (Publ)• Advanced Oncotherapy Plc

Protons

PhotonsNever stop

Do stop, just not quite certain where

The essence of the advantage and the problem

Partners

Proton ruler

X-ray CT ruler

Physical ruler

Density of electrons

Proton stopping power

Real distance

Why we need proton imaging?

If beam passes through 20 cm of tissue, then Bragg peak could be anywhere within +/- 7 mm. Can prohibit treatment of tumour adjacent to spinal cord

Range uncertainties

Things change! Planning CT and after 5 weeks of treatment

Relative Stopping Power (wrt water)

Relative Electron Density (wrt water)

mean excitation potential

relativistic correction

Weakly energy dependent (via relativistic )Depends on material composition (via material I-values)

β

Bethe equation

≠ m + c

• Simple look-up-table

• Stoichiometric calibration

• Dual-energy CT

• Photon counting CT

• Proton CT

• Cone-beam CT

• Ultrasound

• MRI

• Proton CT

CT calibration Image guidance

Every proton matters

Follow the herd - use statistics

• Interaction of energetic protons chiefly through Coulomb interactions with the outer-shell electrons – Multiple Coulomb Scattering. Such energy losses are statistical processes

• Fluctuation in the proton range – range power

• Fuctuation in the proton direction – lateral power• Fraction of proton undergo non-elastic nuclear

interactions – attenuating power

Protons vs. photons

“From where we stand, the rain seems random. If we could stand somewhere else, we would see the order in it.”

Tony Hillerman

+ + xxx x

Proton tracker pair (proximal) Proton tracker pair (distal) Residual energy-resolving detector(Range Telescope)

Estimate entry point

Estimate exit pointEstimate maximum likely path

The principle

Repeat lots of times ….

Category Parameter Value

Proton beamEnergy

$200MeV (head)

$250MeV (body)

Fluxa $3000 protons cm22 s22

Imaging dose Maximum absorbed doseb ,20mGy

Image qualitySpatial resolution, s �1mm

Relative stopping-power accuracy ,1%

TimeData acquisition time ,10min

Reconstruction time ,10min

aQuoted figure based on the scenario of 1-mm voxels and 180 projections, a target of 100 protons passing through a voxel per projection6 and a 10-minacquisition.bQuoted figure based on a crude calculation of comparable stochastic risk to typical X-ray CT head scans (�40mGy7,8), assuming a proton radiationweighting factor twice that of photons.9

Working specification for practical proton CT

100 – 300 MeV protons

First Proximal Strip Camera

Second Proximal Strip Camera First Distal Strip Camera

Second Distal Strip Camera

Residual energy-Range detector

(Range Telescope)

Sets of 3 strip sensors Multiple (20-30) layers of CMOS imagers, silicon strip

sensors, or mixtureRecord incident

trajectoryRecord exit trajectory

Record residual energy

Published patent: WO2015/189603

The instrument

“One of the most complex medical imaging instruments ever conceived”

Loma Linda (USA)PRIMA (Italy)

Fermi Lab (USA)

An aside …

… other players

100 – 300 MeV protons

100 – 300 MeV protons

150 – 350 MeV protons

Quality Assurance Mode

Patient Imaging Mode

Treatment Monitoring Mode

Beam current = 10 - 100 nA

Beam current = 10 - 100 nA

Beam current = 0.1 - 1 nA

0 – 50 MeV protons

Operational Modes

14

Range Telescope 24 layers CMOS images - each:20 um epi700 um substrate200 um pitch (512 x 512 pixels)1 mm perspex

Strip Trackers 4 banks of three 10 cm x 10 cm Silicon strip detectors100 um pitch150 um thickness

TreatmentNozzle

Compensator

TreatmentCollimator

Phantom 75 mm diameter

SuSi

University of Birmingham BlueBEAR HPC cluster

and GridPP

Proton displacement in Range Telescope

-1010

-910

-810

-710

-610

-510

-410

-310

-210

-110

x [mm]-100 -50 0 50 100

y [m

m]

-100

-50

0

50

100

rangetelcapacitors10Doserangetelcapacitors10Dose

Radiation exposure simulations

Modelling

• Birmingham MC40 Cyclotron – up to 36 MeV

• Clatterbridge NHS – up to 60 MeV

• iThemba LABS, South Africa – up to 192 MeV

Available proton sources

• Measures the directions of individual protons as they enter the patient and as they exit – consists of assemblies of high-speed silicon strip sensors

• Radiation-hard technology developed for the Large Hadron Collider at CERN by the University of Liverpool, and employed in the discovery of the Higgs Boson

• Counts at over 100 million times per second

• Recorded over 25 million prtons per second

• Custom read-out chips (ASIC)

Double-ended 150μm thick n-in-p technology – 1024 strips at 90.8 um pitch

128 channels - 2 thresholds (programmable at 6 bits)

256

Prio

rity

en

cod

er12

8 : 7

SYN

C

Ch

ann

el s

can

/res

etlo

gic

chAddress(6:0)

thr2 thr2_out

dataValid

chAddress_out(6:0)

dataValid_out

clk_in

eventClear_in

chan

nel

Gat

e

0

1

2

125

126

127

2

Imaging mode • Every proton detected

Treatment mode • Known fraction of protons

detected• Profile histograms to provide

sufficient information for on-treatment monitoring

Proton trackers

v - outputs

u - outputs

x - o

utpu

ts

A B C D

A B C D

Image reconstructions: Pac-man collimator (29 MeV Birmingham)

Four proton trackers – iThemba Proton Therapy Vault

Published patent: WO2015/189601

Our very first pCT

3 rotated strip assemblies per camera• Reduce ambiguities• Cope with treatment level fluxes

Proton trackers

2015, Pa

~180 Gbit/s

2x CL Cable

8x CL Cable

Sync to Range Telescope

Beam Clk

4x Housing

12x Hybrids

12xCamera Boards

12x Stiffener

4xMux Boards

4xHV Unit

Tracking detectors

Proton trackers: major piece of engineering

Range telescope: even more major piece of engineering

2015, Page 37

~120 Gbit/s

24 Cable Interfaces

8x CL Cable 8x CL Cable

4x CL Cable + 1x PWR Sync from Strip System

1x Housing

48x Imager Halves

24xCamera Boards

24x Stiffener

8xMux Boards

2xMother Boards

Strip only RT Raw data rate ~ 360 Gbit/s

CMOS only RT Raw data rate ~120 Gbit/s

Published patent: WO2015/189602

Proton CT reconstruction: getting the data is only half the problem • Tomographic reconstruction relies on straight rays

• Assumptions of tomography are only weakly violated but there are important consequences

G. Poludniowski, N. Allinson and P. Evans, “Proton computed tomography reconstruction using a backprojection-then-filtering approach” Phys Med Biol. 59:7905-7918 (2014)

backprojection-then-filtering

Published patent: WO2016/016653

• Total analytic solution• Cope with non-linear paths• Correction for finite reconstruction volume• Incorporate differing most likely path algorithms• Computationally efficient

Combine all 180 images

Proton CT reconstruction

Sim

ulat

ion

Sim

ulat

ion

And now, in 3-dimensions …

University of LincolnGrainne Riley Chris WalthamMichela Esposito

University of BirminghamPhil AllportDavid ParkeTony Price

University of LiverpoolJon TaylorGianluigi Casse, Tony Smith, Ilya Tsurin

University of SurreyPhil Evans

University of WarwickSam ManagerJon Duffy

Karolinska University Hospital, SwedenGavin Poludniowski

University Hospital Birmingham NHS Foundation Trust Stuart Green

University Hospital Coventry and Warwickshire NHS TrustSpyros Manolopoulos

iThemba LABS, SAJaime Nieto-Camero

ISDIThalis Anaxagoras Andre FantPrzemyslaw Gasiorek Michael Koeberle

aSpectMarcus Verhoeven Daniel Welzig Daniel Schöne Frank Lauba

Ack

now

ledg

emen

ts

Thank you &

Any questions