Development of X-ray phase-contrast imaging techniques for medical diagnostics

The X-Ray phase-contrast techniques are innovative imaging methods
allowing overtaking the limitations of classic radiology. In
addition to the differential X-ray absorption on which standard
radiology relies, in phase-contrast imaging the contrast is given
by the effects of the refraction of X-rays inside the tissues. The
combination of phase-contrast with quantitative computer tomography
(CT) allows for a highly accurate reconstruction of the tissues’
index of refraction. Thanks to the high sensitivity of the method,
tomographic images can be obtained at clinically compatible dose.
For all these reasons phase-contrast imaging is a very promising
approach, which can potentially revolutionize diagnostic X-Ray
imaging. Several techniques are classified under the name of X-Ray
phase-contrast imaging. This Thesis focused on the so-called
analyzer-based imaging (ABI) method. ABI uses a perfect crystal,
placed between the sample and the detector, to visualize the phase
effects occurred within the sample. The quantitative reconstruction
of the refraction index from CT data is not trivial and before this
Thesis work it was documented only for small size objects. This
Thesis has focused on two main scientific problems: (1) the
development of theoretical and calculation strategies to determine
the quantitative map of the refraction index of large biological
tissues/organs (>10 cm) using the ABI technique; and (2) the
preparation of accurate and efficient tools to estimate and
simulate the dose deposited in CT imaging of large samples. For the
determination of the refraction index, two CT geometries were
considered and studied: the out-of-plane and the in-plane
configurations. The first one, the most used in the works reported
in the literature, foresees that the rotation axis of the sample
occurs in a plane parallel to that of the sensitivity of the
analyzer crystal; while, in the second CT geometry, the rotation
axis is perpendicular to that plane. The theoretical study,
technical design and experimental implementation of the in-plane
geometry have been main tasks of this Thesis. A first experiment
has been performed in order to compare the results obtained with
in-plane quantitative phase contrast CT with the absorption-based
CT ones. An improved accuracy and a better agreement with the
theoretical density values have been obtained by exploiting the
refraction effect while keeping the dose to sample low. A second
campaign of experiments has been performed on large human breasts
to investigate the efficiency of the in-plane and out-of-plane CT
geometries and the performances of the associated image
reconstruction procedures. The same experimental conditions were
also studied by numerical simulations and the results were
compared. This analysis shows that the in-plane geometry allows
producing more accurate quantitative three dimensional maps of the
index of refraction, while the out-of-plane case is preferable for
qualitative investigations. A study for developing advanced
procedures for improving the quality of the obtained CT images has
been also conducted. As a result, a two-step procedure has been
tested and identified: first the noise level of the experimental
images is reduced by applying a wavelet decomposition algorithm and
then a deconvolution procedure. The obtained images show an
enhanced sharpness of the interfaces and of the object edges and
high signal to noise ratio values are preserved. The second problem
of this Thesis was to find strategies to calculate, in a fast way,
the delivered dose in CT imaging of complex biological samples. For
this purpose an acceleration method to speed-up the convergence of
Monte Carlo simulations based on the Track Length Estimator method
has been computed and included in the open-source software GATE.
Results show that this method can lead to the same accuracy of
conventional Monte Carlo methods while reducing the required
computation time of up to two orders of magnitude, with the respect
to the considered geometry. A database of dose curves for the case
of monochromatic breast CT has been produced: it allows for a quick
estimation of the delivered dose. A way to choose the best energy
and the optimal photon flux was also proposed, which leads to a
significant reduction of the delivered dose without any loss in
terms of image quality. Most of the experimental and data
reconstruction methods developed within this Thesis work can be
applied also to other phase-contrast techniques. This Thesis shows
that high resolution three dimensional diagnostic imaging of large
and complex biological organs can, in principle, be performed at
clinical compatible doses; this is the most significant
contribution of the Thesis towards the clinical implementation of
phase-contrast CT.