Biomedical Engineering Reference
In-Depth Information
FIGURE 7.4
Contrast enhancement: Normal volunteer aged 63 years. Registered T1-weighted RF spoiled
images before (a) and after (b) intravenous gadolinium DTPA with difference image [(b)
(a)] shown in (c). Enhancement is seen in the meninges (arrows), vascular layer of the scalp,
skin, veins, sinuses, and nasal mucosa in (c).
changes in regions where the anatomy is complex and where there are vari-
able partial volume effects (such as in boundary regions). Difference images
are also useful for recognizing signal changes in tissues with intensities at the
extremities of the image intensity range. Whereas on source images these are
at the top or bottom of the gray scale, on difference images they are referred
to the neutral (zero change) level. Global changes in signal intensity may only
become obvious on difference images where the brain has a nonzero signal in
relation to the noise on the image outside the scalp.
Image registration can be useful in ensuring that pre- and postcontrast
images are precisely matched. In most circumstances it is unlikely that any
structure in the brain will significantly change in site, shape, or size in the
time between the pre- and postscans, so differences on registered images are
almost always due to pure changes in signal intensity (Figure 7.4).
Normal enhancement occurs in tissues such as the meninges or cortical veins.
These areas are subject to considerable partial volume effect from adjacent tis-
sues, such as brain and CSF, producing variable baseline intensity values. By
performing accurate registration it is possible to ensure the same set of partial
volume effects on pre- and postcontrast images. As a result, more uniform
enhancement of meninges, cortical veins, and ependyma is seen on difference
images than on conventional images. It is possible to perform double subtrac-
tions (i.e., the difference between the contrast enhancement with time from
one examination to the next).
On T1-weighted images, the facts that the contrast agent normally only gives
positive enhancement and that there should be no change in brain site, shape,
or size, provide an additional check on the fidelity of the registration proce-
dure. Under these circumstances, immediate post- minus pre-enhancement
difference images should only show positive changes, whereas a misregistra-
tion artifact typically shows both positive and negative effects. In some cases
we have found that the image differences introduced by the contrast agent
