There are several types of image noise that can affect
the imaging process, but the ultimate source is the random,
statistical noise [7].The author showed that this type of noise
can be avoided by increasing the X-ray exposure.
Compton scattering is a phenomenon, which arises from
the fact that when X-ray photons pass through a matter, they
are deviated from their original direction. These photons are
then useless for the reconstruction. Artifacts in the direction
of highest attenuation are the strongest artifacts [8].
Partial volume effect is present when a voxel is partially
filledwith certain substance.When the voxel is reconstructed,
it represents the weighted average of the attenuations for all
substances in that particular voxel. Decreasing slice thickness
can decrease the partial volume effect. Several algorithms
have been proposed for the correction of these artifacts [9].
Although these methods gave promising results in computer
simulations, they failed when applied to clinical data.
Motion during the scanning process causes inconsistencies
in the measurements. There are several types of motion
that can occur in such conditions: discrete motion, pulsating
motion, and continuous motion. All of these cause different
types of artifacts. If the acquisition speed is increased, the
probability of motion artifacts is decreased. An algorithm
which tries to solve this issue is presented by Glover and Pelc
[10].
The presence of metal objects leads to the amplification
of the effects of listed phenomena.Metal objects are removed
from the patient, if possible, but usually this is not the case
in modern medical practice. Different methods have been
proposed for metal artifact reduction in the literature. The
easiest solution for metal artifact avoidance would be to
use less attenuating materials, such as titanium. Another
possibility is to use X-ray beams with higher energy, but this
approach does not give very good results and also the patient
is exposed to a larger dose of radiation.Therefore, numerous
techniques have been proposed aimed at the elimination or
reduction of the effects caused by metallic objects. These
approaches can be divided into two groups: implicit and
explicit methods. Implicit methods try to suppress artifacts
without the need to apply algorithmicmathematicalmethods
formetal-artifact correction.Their drawback is that they have
limited applicability. On the other hand, explicit methods
are more general and they represent the main focus of
researchers. An extensive list of existing techniques and
methods can be found in [11].
According to the authors, explicit techniques can be
grouped into four categories (Figure 1): corrections in the
sinogram domain, corrections in the image domain, iterative
reconstruction algorithms, and hybrid sinogram correction.
Corrections in the sinogram domain can be further divided
into interpolation-based and noninterpolation-based sinogram
correction techniques.
Some of the interpolation-based algorithms are described
in [10, 12, 13]. The basic idea consists in finding projection
bins which are influenced by the metal in the raw data
and then replacing them with values calculated using the
linear interpolation method.The affected projection bins are
usually found with simple thresholding techniques. After the
correction is performed in the sinogram domain, data is
back-projected in order to reconstruct the image.
Noninterpolation-based algorithms use different
approaches instead of interpolation, such as Monte Carlo
simulation [14].
There are several types of image noise that can affect
the imaging process, but the ultimate source is the random,
statistical noise [7].The author showed that this type of noise
can be avoided by increasing the X-ray exposure.
Compton scattering is a phenomenon, which arises from
the fact that when X-ray photons pass through a matter, they
are deviated from their original direction. These photons are
then useless for the reconstruction. Artifacts in the direction
of highest attenuation are the strongest artifacts [8].
Partial volume effect is present when a voxel is partially
filledwith certain substance.When the voxel is reconstructed,
it represents the weighted average of the attenuations for all
substances in that particular voxel. Decreasing slice thickness
can decrease the partial volume effect. Several algorithms
have been proposed for the correction of these artifacts [9].
Although these methods gave promising results in computer
simulations, they failed when applied to clinical data.
Motion during the scanning process causes inconsistencies
in the measurements. There are several types of motion
that can occur in such conditions: discrete motion, pulsating
motion, and continuous motion. All of these cause different
types of artifacts. If the acquisition speed is increased, the
probability of motion artifacts is decreased. An algorithm
which tries to solve this issue is presented by Glover and Pelc
[10].
The presence of metal objects leads to the amplification
of the effects of listed phenomena.Metal objects are removed
from the patient, if possible, but usually this is not the case
in modern medical practice. Different methods have been
proposed for metal artifact reduction in the literature. The
easiest solution for metal artifact avoidance would be to
use less attenuating materials, such as titanium. Another
possibility is to use X-ray beams with higher energy, but this
approach does not give very good results and also the patient
is exposed to a larger dose of radiation.Therefore, numerous
techniques have been proposed aimed at the elimination or
reduction of the effects caused by metallic objects. These
approaches can be divided into two groups: implicit and
explicit methods. Implicit methods try to suppress artifacts
without the need to apply algorithmicmathematicalmethods
formetal-artifact correction.Their drawback is that they have
limited applicability. On the other hand, explicit methods
are more general and they represent the main focus of
researchers. An extensive list of existing techniques and
methods can be found in [11].
According to the authors, explicit techniques can be
grouped into four categories (Figure 1): corrections in the
sinogram domain, corrections in the image domain, iterative
reconstruction algorithms, and hybrid sinogram correction.
Corrections in the sinogram domain can be further divided
into interpolation-based and noninterpolation-based sinogram
correction techniques.
Some of the interpolation-based algorithms are described
in [10, 12, 13]. The basic idea consists in finding projection
bins which are influenced by the metal in the raw data
and then replacing them with values calculated using the
linear interpolation method.The affected projection bins are
usually found with simple thresholding techniques. After the
correction is performed in the sinogram domain, data is
back-projected in order to reconstruct the image.
Noninterpolation-based algorithms use different
approaches instead of interpolation, such as Monte Carlo
simulation [14].
การแปล กรุณารอสักครู่..