Biomedical Engineering Reference
In-Depth Information
The main limitation to the spatial integrity of CT images lies in the mechan-
ical elements of the imaging process, i.e., movement of the radiation source
and detectors, as well as of the patient couch. Therefore, the following spe-
cific mechanisms can compromise the geometric integrity of images:
1.
Error in the slice angle relative to the scanning axis due to uncer-
tainty in the gantry angle and
or table bending due to the patient
weight;
2.
Error in the slice separation due to fluctuations in the couch speed
relative to the prescribed value;
3.
Error in the dimensions of the field of view due to imperfections
and
or changes in the physical dimensions of the tube detector
assembly.
Zylka and Wischmann devised a method to measure and compensate for
all of the above sources of distortion in CT images.
10
By modeling the distor-
tions as a second (or higher) degree polynomial and using an “N-shaped”
localization device fixed to the table, the authors were able to observe distor-
tions of the order of 2 mm, which depended on the gantry tilt angle. These
will also vary depending on the mechanical properties of each scanner and
the subject's body weight.
5.3
Spatial Inaccuracies in Magnetic Resonance Imaging
From the point of view of geometric distortions, MRI differs greatly from CT
in at least three respects: first, the image acquisition process of MRI is funda-
mentally different, starting with the fact that it does not rely on any mechan-
ical device. Second, the problem of geometric distortions is much more
important in MRI, due to the fact that these are essentially unavoidable and
can be quite large (5 mm or more).
6,11,12
Third, MR images obtained using dif-
ferent data acquisition schemes (conventional 2D and 3D Fourier transform
imaging, echo-planar imaging, spiral imaging) and hardware (e.g., birdcage,
surface coils) can be subject to substentially different image distortion effects.
Therefore, the importance of the problem of image distortions was recognized
in the early days of the application of MRI
11,13
and has been the subject of
intense investigation over the past 20 years, resulting in a relatively large body
of literature. There have been major advances in this field in the past decade, in
particular improved main field homogeneity and the use of stronger readout
gradients, which has significantly helped to reduce the magnitude of the prob-
lem in most standard applications. However, the problem of image distortion
has not yet been resolved for all types of image acquisition sequences and
applications.
