2004 IEEE NSS

1
Characterisation of an Integrated X-Ray Breast
Imaging system based on a Large Area Sensor and a
Photon-Counting Spectroscopic Multi-element Array

• D.G. Darambara1, P. Sellin1, G. Maehlum2 and J.
  Hearne3
• 1Department of Physics, University of Surrey,
  Guildford, Surrey GU2 7XH, UK
• 2IDE AS, Veritasveien 9, N-1322 Hovik, Norway
• 3Moranlord, Instrumentations Control Products,
  Rudwick, West Sussex, UK
• Sponsored by The Wellcome Trust

2
Challenge

• In diagnostic x-ray imaging the determination of
  small size lesions in a low-contrast environment,
  while keeping the dose applied as low as possible

3
Coherent x-ray Scatter (CS)

• A considerable fraction of the total x-ray
  scattering process
• Peaks at precisely those energies where x-ray
  breast imaging is performed
• The x-ray diffraction cross-sections of various
  tissues quite different in a material-specific
  way for specific angles and photon energies
• Every molecular structure has its own scattering
  pattern
• The remarkable structure in the scatter intensity
  at small angles is the diffraction pattern of
  tissue arising from interference between
  radiation scattered from different electrons. The
  atomic, molecular and intermolecular e-
  distributions all affect the scatter distribution
• The degree of interference is dependent on the
  momentum transfer
• ? (1/?) sin(?/2)

4
X-ray Imaging

• The ability of x-ray diffraction to provide the
  molecular structure of the biological tissues
  could add a new dimension in x-ray imaging
  capable of tracking the molecular structural
  changes during disease progression and of
  improving the sensitivity to low contrast lesions
  without increasing the radiation dose
• Capability of tissue characterisation with the
  coherently scattered photons while imaging

5
Scatter Signatures

• Poletti et al (2002) Kidane et al (1999)
• (a) water, (b) glandular breast tissue
• (c) Adipose breast tissue, (d) water vs breast
  tissues
• Synchrotron studies Lewis et al (2000,
  Daresbury), Fernández et al (2002, ESRF)
• Johns et al (2002) scatter images better C and
  SNR than projection images for the same
  radiation dose

6
Experimental set-upConcept for simultaneous
measurements of the transmitted primary and the
forward scattered x-rays
7
Flat Panel Transmission Detector

• a-SiH
• Pixel size 127 ?m
• Active area 40 x 30 cm
• Array format 2304 x 3200
• ADC 14 bits
• Fill factor 57
• Scintillator Gadox
• Resolution 3.5 lp/mm at 10 MTF
• DQE(0) 35

• Hamamatsu CMOS
• Pixel size 50 ?m
• Active area 120 x 120 mm
• Array format 2240 x 2344
• ADC 12 bits
• Fill factor 80
• Scintillator CsI (200?m)
• Resolution 10 lp/mm at 10 MTF
• DQE(0) 60

8
Flat Panel Transmission Detector
Minimum Detectable Thickness (?m) Minimum Detectable Thickness (?m) Minimum Detectable Thickness (?m)
Diameter (mm) a-SiH CMOS F/S
0.13 1.6 - -
0.16 1.0 1.0 1.0
0.5 0.16 0.2 0.25
3.2 0.06 0.06 0.1

• CMOS ? selective readout ? direct addressable
  pixel readout (random access) ? capability of
  simultaneous measurements

9
Requirements of CS Detector

• Energy-resolved detector capable of giving us
  position info
• Direct detection and conversion of x-rays to
  electric charges
• Low noise vs a comparatively weak CS signal
• Clean discrimination between signal and noise
  by setting-up a threshold
• Operation in room temperature
• ?min ? ?max more photons vs increase the
  contrast
• developed a high-resistivity photon-counting
• spectroscopic multi-element 2D Si detector
• with multi-channel low-noise
• pulse-processing front-end electronics
• R.J. Leclair P.C. Johns (2002)

10
Coherent Scatter Sensor

• Optimised to measure E 12 45 keV
• Array of 6 x 21 pads 2 mm pitch
• Active area 12 mm x 42 mm
• High resistivity 1mm thick Si
• Stable operation in full depletion, corresponding
  to Vbias 150-300 V

11
Coherent Scatter Si Sensor

• Meets design specifications combining good Energy
  Resolution (1-3 keV), with Position Information
• V. good spectral response to x-ray energies of
  interest (17-45 keV)
• Resolves the Ka and K? peaks of Tb, Ba, Ag, Mo,
  which tie in with the known branching ratios
• Detection Efficiency 70 at 17 keV, 65 at 20
  keV and 17 at 45 keV
• Low-noise behaviour
• Operates at near room temperature without
  cryogenic cooling

12
Analysis and Visualisation of Data

• Data can be presented in raw ADC values and/or
  E in keV via a LabView-based programme
• Single spectral response, part of, and/or overall
  spectrum across all detector channels
• hit map, mean amplitude, FWHM across all channels
• peak centroid, E resolution measurements
• 2D intensity maps pulse height in ADC/E for each
  channel across the detector

13
Reference Scatter Signaturesa) PMMA, b) adipose,
c) polyethylene, d) nylon, e) water
14
Reference Scatter Signatures
15
Phantom Scatter Signature
16
Si-pad vs CS

• Excellent low-angle x-ray scatter performance
  with highly-resolved diffraction signatures at
  mammographic energies
• Capability of characterising breast tissues
• Capability of applying a mixture of angle and E
  dispersion
• Multi-element 2D array to collect multiple
  scatter signals simultaneously to maximise the
  content of diagnostic information and for shorter
  measurement times
• Multiple angle scatter collimation collect
  multiple scatter signatures at different angles
  (?min ? ?max)
• Capability of using circularly symmetric
  region-of-interest by placing Si-pad detectors
  at both sides of transmission line
• ?Multi-element Si-pad sensor lends itself well
  to CS

17
Transmission and T/D tissue images
18
Problems / Barriers

• Design and construction of a 2D multi-angle
  scatter collimators
• Contribution of multiple scattering (thick
  samples) be investigated and minimised
• Attenuation of transmitted and scattered beams
  (depth of scattering volume - thick samples)
• Dimensions of the system and experimental
  parameters (voltage current of source,
  scattering angle, primary beam diameter,
  dimensions of scatter collimator) ? accuracy of
  the scattering volume localisation and height of
  the measured diffraction peaks
• Low intensity scatter signal slow measurements
• Better data for breast tissue scattering
  cross-sections (synchrotron)
• More clear and convincing data for breast
  tissue discrimination from scatter signatures in
  a clinical setting

19
Conclusions

• Do the coherent scatter measurements of in-vitro
  breast specimens reflect the in-vivo situation
  where
• the various breast tissues are bathed in fluids
  such as blood and
• the lesion is typically surrounded by benign
  tissue that can produce scatter and other signals
  that will be difficult, if not impossible, to
  separate and localise independently from the
  small signal produced by malignancy?????