to operate without strong fluctuations and doubles the dose for the imaged object as
the additional second scan is needed.
Generally, when incident photons collide with electrons of a medium, they give
up some of their energy, and these Compton scattered photons, deflected out of their
original direction of motion, have a distinct angular and energy distribution. In fact,
this dependence can be exploited for scatter removal by determining the different
spectro-spatial behaviors of the scattered and fluorescent X-rays [22]. The X-ray
fluorescence signal has a fixed energy and is distributed uniformly in the 4π angular
space. The Compton scatter instead is anisotropic: backscatter X-rays (Θ % 180
)
have lower energies than forward-scatter X-rays (Θ % 0
). Thus, the two different
types of events can be separated by placing the detector in an optimized position.
Monte Carlo simulations with gold nanoparticles have shown that distinct angles can
be found in which directions the overlap between the Compton and fluorescence
photons are avoided or much reduced for different excitation spectrums [22]. In this
study, the detection of the K β lines of gold with Θ > 110
proved to be a better
configuration than the full 4π coverage. This means that contrary to intuition, the
optimized configuration simultaneously reduces the number of detectors and
improves the imaging result.
By optimizing the fluorescence data acquisition geometry, the interfering Compton scatter X-rays can be reduced by orders of magnitude, which pushes the limit of
detectable gold concentrations to 0.001% by weight (the corresponding molar
concentrations of a mono-atomic gold solution is 50 μM, and the molar concentrations for nanoparticle solutions would be nM). By optimally placing the detectors
with respect to these results, an order of magnitude increase in sensitivity is
achievable.
In order to apply the principle of XFCT to a living organism, one must take the
attenuation of X-rays by the surrounding tissues into account. Generally, elements
such as gadolinium (Z ¼ 64, K α1 ¼ 42 keV), gold (Z ¼ 79, K α1 ¼ 69 keV), barium
(Z ¼ 56, K α1 ¼ 32 keV), and iodine (Z ¼ 53, K α1 ¼ 29 keV) can be imaged at a depth
of ~20 cm, making XFCT imaging of many elements a viable choice. As L-shell
fluorescent photons are much stronger attenuated by tissue than K-shell fluorescent
X-rays, the attenuation correction is more important in L-shell XFCT imaging.
Nevertheless, suitable attenuation correction is required for quantitative imaging
for all X-ray energies.
The iterative image reconstruction commonly used in XFCT imaging requires a
system matrix. The attenuation of incoming and outgoing X-rays needs to be
incorporated correctly into the system matrix with different linear attenuation coefficients for each XF energy [15]. Hence, the position of each fluorescent X-ray on the
detector must be determined to calculate the probability of XF absorption along this
specific line. The attenuation must be integrated into each system matrix element as
each system matrix element represents the detection probability of one detector pixel
with one image pixel/voxel. For each system matrix element, the attenuation is
calculated by summing all line segments weighted with the attenuation coefficient
obtained from the CT image. For both the line connecting to the detector pixel (with
the attenuation coefficient for the XF energy) and the line created by the pencil beam
X-Ray Excited Fluorescent Materials for Medical Application
133
the additional second scan is needed.
Generally, when incident photons collide with electrons of a medium, they give
up some of their energy, and these Compton scattered photons, deflected out of their
original direction of motion, have a distinct angular and energy distribution. In fact,
this dependence can be exploited for scatter removal by determining the different
spectro-spatial behaviors of the scattered and fluorescent X-rays [22]. The X-ray
fluorescence signal has a fixed energy and is distributed uniformly in the 4π angular
space. The Compton scatter instead is anisotropic: backscatter X-rays (Θ % 180
)
have lower energies than forward-scatter X-rays (Θ % 0
). Thus, the two different
types of events can be separated by placing the detector in an optimized position.
Monte Carlo simulations with gold nanoparticles have shown that distinct angles can
be found in which directions the overlap between the Compton and fluorescence
photons are avoided or much reduced for different excitation spectrums [22]. In this
study, the detection of the K β lines of gold with Θ > 110
proved to be a better
configuration than the full 4π coverage. This means that contrary to intuition, the
optimized configuration simultaneously reduces the number of detectors and
improves the imaging result.
By optimizing the fluorescence data acquisition geometry, the interfering Compton scatter X-rays can be reduced by orders of magnitude, which pushes the limit of
detectable gold concentrations to 0.001% by weight (the corresponding molar
concentrations of a mono-atomic gold solution is 50 μM, and the molar concentrations for nanoparticle solutions would be nM). By optimally placing the detectors
with respect to these results, an order of magnitude increase in sensitivity is
achievable.
In order to apply the principle of XFCT to a living organism, one must take the
attenuation of X-rays by the surrounding tissues into account. Generally, elements
such as gadolinium (Z ¼ 64, K α1 ¼ 42 keV), gold (Z ¼ 79, K α1 ¼ 69 keV), barium
(Z ¼ 56, K α1 ¼ 32 keV), and iodine (Z ¼ 53, K α1 ¼ 29 keV) can be imaged at a depth
of ~20 cm, making XFCT imaging of many elements a viable choice. As L-shell
fluorescent photons are much stronger attenuated by tissue than K-shell fluorescent
X-rays, the attenuation correction is more important in L-shell XFCT imaging.
Nevertheless, suitable attenuation correction is required for quantitative imaging
for all X-ray energies.
The iterative image reconstruction commonly used in XFCT imaging requires a
system matrix. The attenuation of incoming and outgoing X-rays needs to be
incorporated correctly into the system matrix with different linear attenuation coefficients for each XF energy [15]. Hence, the position of each fluorescent X-ray on the
detector must be determined to calculate the probability of XF absorption along this
specific line. The attenuation must be integrated into each system matrix element as
each system matrix element represents the detection probability of one detector pixel
with one image pixel/voxel. For each system matrix element, the attenuation is
calculated by summing all line segments weighted with the attenuation coefficient
obtained from the CT image. For both the line connecting to the detector pixel (with
the attenuation coefficient for the XF energy) and the line created by the pencil beam
X-Ray Excited Fluorescent Materials for Medical Application
133