Compton Camera with Position-Sensitive Silicon Detectors

Andrej Studen

Outline:● Motivation● Basic Principles● New Compton camera prototype● Image reconstruction● Conclusion

Supervisor: Professor Marko Mikuž

Doctoral thesis

University of LjubljanaFaculty of mathematics and physics

Warning

This is a presentation of a doctoral thesis. If your plans for today do not include listening to a presentation of a doctoral thesis, this would be a perfect time to leave.

MotivationIn short:Compton camera could substantially improve SPECT imaging.

With explanation:

Single Photon Emission Computed Tomography (SPECT) is a diagnostic tool in nuclear medicine (brain scan, whole body tumor search,...)

Selected substance is labeled by radio-active sources emitting g-rays and injected into a patient.

A gamma camera is required to trace the substance within the body of a patient.

Traditional cameras: Mechanically collimated or Anger cameras.

Anger camera

Hal Anger (1953)

(sold by Siemens, ELSCINT, GE)

Inverse coupling (trade-off) between resolution and efficiency limits SPECT.

Many possibilities for alternative detection methods.

Compton camera● Compton camera replaces the mechanical collimator with a special

scattering detector where Compton scattering occurs.● Collimation is based on kinematics of Compton scattering.

ANGER COMPTON

Compton camera● Compton camera replaces the mechanical collimator with a special

scattering detector where Compton scattering occurs.● Collimation is based on kinematics of Compton scattering.

ANGER COMPTON

Compton camera (cont'd)Compton imaging is a successful tool in astronomy (COMPTEL, MEGA,...), where photon energies exceed 1 MeV.

The photon energy range in SPECT(140.5-511 keV) introduces harsh requirements for Compton camera.

This work continues the efforts in the field, represented by prototype C-SPRINT, constructed in 1999 at University of Michigan.

COMPTEL 1991-2000

C-SPRINT 1999

Basic Principles: Compton scattering

Kinematics of Compton scattering relates the scattering angle q of the photon to the energy Ee of the Compton electron.

sin2/2=mc2E/E eE−E eCompton equation

Random movement of bound electron prior to scattering broadens the scattering angles - Doppler broadening (indicated by the blue arrows).

Division of Compton camera in sub-detectors

Compton camera consists of two sub-detectors:

● Scatterer measures the position of Compton scattering and energy Ee of the Compton electron

● Absorber measures the impact position of the scattered photon.

sin2/2=mc2E/E eE−E eCompton equation

Possible directions of incoming photons form a cone with axis along the scattered photon track, apex on the interaction point in the scatterer and opening angle equal to the scattering angle.

Position reconstruction

Intersection of cones formed for consecutive events reproduces the position of the source.

Position reconstruction

Requirements: Angular resolutionAngular resolution of the reconstruction is due to three contributions: ● Doppler broadening (inherent to the method).● Energy resolution of the scatterer (through Compton equation).● Geometrical factor related to spatial resolution of sub-detectors.

Task: reduce contributions of geometry and energy resolution below Doppler broadening:● Geometry => simple,● Energy => the harsh

requirement of SPECT, goal 1 keV FWHM.

astr

onom

y

Requirements: Efficiency & Background

Efficiency important in SPECT: collateral damage by radiation uptake.

Management of radiation dose absorbed by the patient. => Efficient setup required.

Plenty of background gamma radiation expected:● large activity used (0.1 GBq scale),● uptake in tissue surrounding the imaged object.

Steps taken to ensure signal data is recorded:● maximize timing correlation between sub-detectors => timing resolution

required,● shield the absorber from direct hits => adjust geometry,● reject events based on sum of energies in both sub-detectors => measure

energy in the absorber.

New Compton camera prototypeScatterer:● based on recent development in

silicon technology.● resolution of 1 keV FWHM● optimize timing.

● Absorber (borrowed from existing SPRINT setup)

● DAQ● mechanics

Advantages of silicon as scattering material:

● mature production and processing yielding:● excellent spatial resolution,● excellent inherent energy

resolution at room temperature,

● low Doppler broadening,● high Compton to photo-

absorption ratio making efficiency grow with sensor thicknes with no signal loss.

+

emphasis

The scattererScatterer module consists of:● silicon pad sensor :

4.8 x 1.2 cm2, up to 1 mm thick, 256 pads, 32 x 8 array, designed by W. Dulinski & produced by SINTEF

● associated front-end electronics:

pair of VATAGP3, made by Ideas, A.S.A., Norway

The scatterer consisted of several identical modules (efficiency grows linearly with total thickness).

VATAGP3

STACK

The scatterer: DAQ system

Receive trigger, read, digitize & store values.● Distribution board: daisy-chain the modules● Intermediate board: analog signals for VATAGP3-s, sensor bias● VME board: digital signals for VATAGP3-s, ADC, PC

communication● PC: data storage and processing

The scatterer: Energy resolutionImportant for angular resolution of prototype.VATAGP3 provides a slow shaped (3 ms) CR-RC output for precise measurement of Compton electron energy. Noise (= error) : electronic + (inherent) silicon, 99mTc (126 e)2 : (120 e)2 + (37 e)2

Pedestal noise of module with 0.5 mm thick sensor

Electronic noise dominates: signal must be collected in a single pad, → pad size (1.4 mm) exceeds expected Compton electron range (< 0.7 mm). Spatial resolution of the scatterer equals to pad size (1.4 mm FWHM).

The scatterer: Energy resolution

Thick sensor desired - larger efficiency:● 0.5 and 1 mm sensor tested, exceeding

standard thickness (0.3 mm).● tests performed with 241Am

(Eg= 59.5 keV).

● performance of both sensor types close to design goal.

0.5 mm

1 mm

The scatterer: Timing resolutionImportant for proper matching of interactions in scatterer and absorber.

VATAGP3 trigger: fast (150 ns) CR-RC shaper and constant level discriminator.

Shaping time > signal collection time (1 mm Si) => 150 ns minimum.

150 ns time-walk

Measurement (varying threshold at fixed charge injection):● jitter: 17 – 50 ns● time-walk: 150 ns (dominant)

Possible gain with alternative triggering methods (e.g. CFD) : reduced time-walk (at the expense of jitter).

The absorber

Borrowed from SPRINT.

3 layers of associated electronics

20 Hamamatsu R980 PMTs

44 NaI(Tl) bars, each 3 mm wide, 15 cm long and 1.27 cm thick

The absorber: DAQ system

Generate trigger from Energy sum; collect and digitize maximum value of shaped outputs of 20 PMT-s.Based on VME, PC stores & manipulates data.

The absorber: Spatial resolutionModules were calibrated on a rectangular (5 x 5 mm2) mesh. Position was determined using the centroid method and distortion was corrected with the known position of the source.

Corrected resolution below 4 mm FWHM across the sensitive area.

variance [mm]

The absorber: Energy resolutionSum of energies recorded in scatterer and absorber is used for separation of true events from random background.

Energy resolution of the sum dominated by energy resolution of NaI(Tl), which is much lower than in silicon.

The absorber: Timing resolutionThe timing resolution measured using 22Na positron source: ● one photon absorbed by the plastic scintillator (timing reference),● second photon absorbed in SPRINT.

Timing resolution of absorber is 24 ns FWHM; small compared to 150 ns resolution of the scatterer.

Mechanics

Custom designed gantry: ● up to 5 scatterer modules; at the center● 3 absorber modules form a semi-ring (reduced background)● optimized for on-axis source● good efficiency for scattering angles around 1 radian where imaging is

optimal

EfficiencyComposed of combined probabilities :● that emitted photon hits the scatterer (solid angle, 1.6 %

@ 5 cm)● of Compton scattering (P(q), Klein-Nishina)● of geometrical acceptance of scattered photons (a(q))● that scattered photon interacts in absorber (ea)

photon absorberenergy [keV] efficiency [%]

140.5 100250 60511 10

=a1∫0

a P d

Total efficiency (e):

1.6 ‰ for 99mTc @ 5 cm, 4.5 mm scatterer

Oskar Klein

The data throughput of the prototype was saturated by the speed of the DAQ system electronics at 50 Hz.

Angular resolution

Worst case - 99mTc:● Doppler broadening: 60 mrad

@ 1 rad● Energy resolution of the

scatterer: 45-60 mrad @ 1 rad● Geometry contribution: 50

mrad @ 1 rad● Total: 105-125 mrad @ 1 rad● Increasing Eg helps!

Eg Doppler E scat (1.5 keV) Geometry Total[keV] [mrad] [mrad] [mrad] [mrad]140.5 60 60 50 125250 40 30 50 70511 30 10 50 60

optimal range

DAQ system

Joint scatterer and absorber DAQ systems.

SYSTEM TRIGGER: timing coincidence ofany SPRINT AND scatterer.

Event combined in VME, stored and manipulated by PC.

Timing resolution

Triggers from scatterer and absorber combined.

Timing resolution dominated by the scatterer.

Low energy (low angle) Compton electrons cut by the coincidence window.

low-energyComptonelectrons

Image reconstruction

To determine the performance of the prototype, simple back-projectionwas used instead of elaborate techniques (EM/LML).

Reduce the problem to 2D by reconstruction on a focal plane; multiple planes give 3D information.

For each event:● construct the cone of possible

directions of initial photon,● determine and properly weight pixels

hit by the cone

Image reconstructionImage recorded with 2 out of 3 absorber modules: y direction lacks resolution.

Representative profile along X:

99mTc point source; 1 x 1 mm2

Image reconstructionAngular resolution is determined as spatial resolution divided by the source-scatterer separation. Performance matches the prediction.

Best resolution (170 mrad FWHM) for lowest detected energies of Compton electrons.

Conclusion: Prototype evaluation

for this Anger(99m)Tc prototype camera

resolution [mm] 8 – 11 (@ 5 cm) 5 –151.6 0.05 – 0.4efficiency [‰]

average factor of 50 required for cone reconstruction (Compton) to break even with line reconstruction (Anger)

Performance comparable

simple CC prototype, worst conditions (99mTc), state-of-the art Anger camera

● Resolution:● optimized geometry● elaborate reconstruction techniques (in

C-SPRINT factor 2 improvement).

Compton camera is a possible tool of the future.

for C-SPRINT this(99m) Tc prototyperesolution 160 170

[mrad] 100* 80** 50***

0.001 1.6

efficiency [‰]

* equivalent geometry assumed** with EM/LML*** assumed both equiv. geom. and EM/LML

better (due to improved scatterer)

qualitative

qualitative

● Improved efficiency:● larger scatterer sensors, ● more scatterer modules.

● Shift to larger Eg

Possible improvements of the prototype:

Prostate probeA possible future application; prostate probe.Aimed at detection of prostate cancer, second most common cancer in men, which develops in 1/6 of population.

● finger-thick probe inserted intra-rectally, ● viewed by a large absorber surrounding the patient.

The probe relies on:● near-angle view of the prostate,● convenient radio-tracer used (111mIn, 250 and 171 keV).

Simulation predicts:● spatial resolution and efficiency far better than in Anger cameras,● possible detection of primary prostate tumors (only secondary

detectable with Anger camera) .

Prostate probe

Prostate probe conventional SPECT

Concluding remarks

● Compton camera offers an alternative to mechanically collimated imagers, removing the resolution-efficiency trade-off.

● Harsh requirements for SPECT applications.

● Developed position-sensitive silicon sensors with associated readout electronics meet those requirements (1 keV FWHM resolution, up to 1 mm thickness).

● Simple prototype with developed scatterer and conventional absorber, data reconstructed with a simple back-projection, was found comparable to contemporary Anger cameras.

● Possible improvements and shift to other radio-tracers can make Compton camera the tool of the future.

● A possible application is a prostate probe which could directly image primary tumors, allowing for early disease diagnostics and treatment.