Image Processing Reference
In-Depth Information
corrected prospectively using phantoms with known reference intensities
and reference points. Retrospectively, intensity inhomogeneities can be
corrected with the methods presented in this chapter. It is, though, very
difficult if not impossible to correct the spatial distortions retrospectively.
2.
Imperfections in the gradient coils and abnormal currents through the
gradient coils often lead to effects similar to those created by
B
0
inhomogeneities and thus need to be corrected similarly.
3.
Radio frequency (RF) coil-related effects are caused by coil imperfec-
tions, by nonuniform sensitivity of the coil (e.g., in surface coils) or by
ferromagnetic objects in the imaged object (e.g., amalgam dental fillings,
orthopedic implants). The main effect is a variation of the intensity values
across the MR image. This type of inhomogeneity is often scan-specific
and cannot be corrected prospectively using phantoms, but rather needs
to be corrected retrospectively.
4.
Poor RF penetration due to absorption of the RF signal leads to darker
inner regions of the imaged objects compared to the outer regions. This
effect can only be corrected retrospectively.
5.
Minor intensity inhomogeneities of the same tissue might also be a
true observation of the subject's anatomy. In general, none of the human
organs or tissue types are fully homogeneous. Retrospective corrections
of these subject-specific inhomogeneities can be beneficial and desired
in order to enhance homogeneity within the tissue. On the other hand,
the separation of inhomogeneity effects from true tissue contrast can
be difficult at the border of two tissues and can lead to a subsequent
incorrect segmentation[6].
6.
Any of the sources listed in the preceding text can cause slowly varying
intra- and interslice intensity inhomogeneities. Especially early magnetic
resonance (MR) acquisition protocols, but also contemporary
MR-angiography (MRA) sequences, often produce additional interslice
intensity inhomogeneities that do not vary slowly, but rather in step-like
fashion (“venetian blind artifact”). It is noteworthy that only a small set
of “modern” methods can deal with this artifact (e.g., [7-10]).
7.
Intensity variations in images of the same subject taken at different
time points, as well as variations in images of different subjects, can
be regarded as a special case of intensity inhomogeneity. Some of the
methods presented in this chapter can correct both intrasubject as well
as intersubject intensity inhomogeneity between images.
Severe intensity distortions due mainly to RF coil effects can lead to misleading
visual interpretations (see
Figure 5.1a
). More subtle distortions that are barely
visible are commonly less relevant for visual interpretation, but can still severely
corrupt segmentation and registration methods (see Figure 5.1b). The task of
retrospectively correcting both severe and subtle intensity inhomogeneities in
MR scans has been extensively researched and discussed in the medical imaging
research over the past 15 yr. This chapter gives an overview of the most
Search WWH ::
Custom Search