Optimization of in-line phase contrast particle image velocimetry using a laboratory X-ray source
thesis
posted on 2017-02-17, 00:07authored byNg, King Lun
Phase contrast particle image velocimetry using a laboratory X-ray source (phase contrast
X-ray PIV) is investigated by a synthetic model of a mouse lung. The lung is
modelled by densely-packed 75 micron air bubbles which are embedded within water exhibiting
a steady-state vortical flow. Phase contrast images of the bubbles are generated
under the paraxial approximation using a tungsten X-ray spectrum at 30.0 keV. Images
at a range of source-to-object and object-to-detector distances are generated, and used
as input into a simulated PIV measurement. The effects of source-size-induced penumbral
blurring, together with the finite dynamic range of the detector, are accounted
for in the simulation. Simulation results show that the bubbles appear as speckles
when phase contrast imaging is applied. The PIV measurement procedure involves
using the cross-correlation between temporally-sequential speckle images to estimate
the transverse displacement field for the transportation of the bubbles. Investigations
of the contrast level of the images and the corresponding correlation peaks suggest
that phase contrast X-ray PIV optimization is distinct from phase contrast imaging
optimization. The global error in the PIV reconstruction, for the set of simulations
that was performed, suggests that magnification is the key parameter for designing a
laboratory-based phase contrast X-ray PIV system. For the modelled system, phase
contrast X-ray PIV can be optimized in a system with magnification between 1.5 and
3. For large effective pixel size (> 20 micron), high geometric magnification (> 2.5)
is desired, while for large source system (> 40 micron), low magnification ( < 1.5) is
suggested instead. The model developed in this thesis can be applied to phase contrast
velocimetry optimization using a variety of laboratory X-ray sources. The high degreeof-
freedom of input parameters of this model is beneficial to design systems for clinical
phase contrast X-ray PIV of biological flows and tissue movements.