'' Light scattering artefacts in

1
HUNTING FOR..

• .. Light scattering artefacts in
• a funnel phantom using
• optical CT

Stephen Bosi, Saxby Brown, Sarvenaz Sarabipour,
Yves De Deene and Clive Baldock Institute of
Medical Physics University of Sydney and
University of Ghent
2
(No Transcript)
3
(No Transcript)
4
Laser Optical CT scanner
SLOOOW!
5
Cone-beam Optical CT
light panel
FAAAST! but..
dosimeter
camera
diffuser
aquarium
filter wheel
6
Vista Optical CT Scanner Modus Medical Devices
Inc.
7
Now the ACTUAL point of the work...

• The brief Test the suitability of Vista scanner
  for polymer gels

Test exposed gelComplications?depth
dose,polymer saturation etc..
8
Faking itOldham Gelatin Finger Phantom

• Use test tube to make finger-shaped cavity

Gelatine food-dye tosimulate exposure
9
Oldham Gelatin Finger Phantom Shortcomings
Food-dye diffuses...
10
Oldham Gelatin Finger Phantom Shortcomings
Food-dye absorbs light (Beer-Lambert law) I I0
e-xD

• BUT
• Polymer gel scatters light
• B-L breaks down for large conc
• Blurred image boundaries

11
Polymer Gel Dosimeters

• Optically scattering

• Opalescent (Tyndall scattering) scatter
  bluish-white, transmission reddish?
  colloidal particles radii lt 1 mm

12
Simulating Polymer Gels
Must use colloidal, light scattering medium

• Colloidal suspensions when added to water
• e.g. Absinthe, Ouzo, Araq, Pernod, milk

OR "DettolTM"
13
Layered Dettol Finger Phantom

• Mix antiseptic with melted gelatinconcentration
  determines turbidity
• Backfill Oldham-style finger cavity with turbid
  gel layers

14
Layered Dettol Finger Phantom

• More Stable!

15
Testing Scanner

• Images scanned
• Feldkamp back projection reconstruction
  (VistaReconTM)
• Viewed (VistaViewTM)

16
Optical density profiles

• Pixel brightness 1000 OD (Optical density)
• Hi OD slices exhibit dishing artefact

17
Optical density profiles

• Profiles taken from the reconstruction. (D OD)
• Plots labels are Dettol conc. in mg/g

18
Optical density v conc.

• ImageJ to analyse image pixel statstics
• Plot mean OD vs Dettol conc.

• Saturated polymer? Clearly not!

19
Spectroscopic Optical Density

• Dettol gels in cuvettes

• ODs measured spectrometrically

20
Compare Optical Density

• Saturation "dose" depends on path length?can't
  use OD-dose lookup table to fix it

21
Dettol Phantom Compared to PAGAT

• PAGAT gels in cuvettes

• Expose to Co-60 ?

• spectrometric ODs

22
Compare Antiseptic Phantom with PAGAT

• Dose/Conc. equivalence curve for Co-60 ?
• Calibration OpCT phantoms?

23
Optical Saturation

• nearly opaque region scattered stray light
• "Transmission" too high - OD low conc. indep.
  i.e. saturation

24
Modelling Optical Saturation 1

• Assume stray light constant over width of each
  slice
• Assume transmitted light obeys Beer-Lambert
• Light detected stray transmitted

25
Modelling Optical Saturation 2

• Artefact completely explained by stray light
• Obeys Beer-Lambert? Sure does!

26
Babushka Finger Phantom

• Finger within a finger
• Outer finger in layers

27
OD profiles from reconstruction

• Outer finger dishing
• Inner finger OD too high

28
Modelling Babushka 1

• Same model as finger (fit to projection
  snapshots)
• Fit good - stray light explains all

29
Modelling Babushka 2

• Raw reconstruction overestimates OD in central
  pith

30
Modelling Babushka 3

• Fitted OD in outer shells obeys B-L law ( slope)

31
The Uniform OD Funnel Phantom
32
OD profiles from the reconstruction
Dishing AND doming!?
33
Modelling funnel projections

• "Stray light Beer-Lambert" model again
• Background deliberately higher OD now!... model
  now includes stray light in baseline ref.
• OH NO! Check out the Penumbras -(

34
Reconstructing fitted projections
Synthetic fitted projections,no
penumbra Cupping, no doming
35
Adding the penumbra...
Synthetic fitted projections, penumbra added
manually Penumbra explains the doming!
36
Conclusions

• Must characterise scanner using artificial
  phantom to avoid confusing optical, radiation or
  polymer effects
• Scattered stray light accounts for all our
  artefacts
• Can't use OD vs dose lookup table to correct
  scattering induced non-linearities
• Improvement by including model of stray light in
  reconstruction algorithm?

37
(No Transcript)