Biomedical Engineering Reference
In-Depth Information
2.
Interleaved pixel or checkerboard fusion, in which each pixel dis-
plays an intensity purely from one of the images being fused, with
the represented image alternating in a regular interleaved pattern. This
can be on a pixel-by-pixel basis, or in larger groups of pixels, forming
a blocked appearance (e.g., Chapter 11, Color Figure 11.2). This
method has the advantage of allowing extra colors to be displayed
on older computers which only have eight bit frame buffers. It also
provides a rapid means of manipulating color maps to fade one or
another image in and out.
3.
Dynamic alternating display, in which the computer switches rap-
idly from one image to the other either automatically or under user
control.
4.
Split view displays place two images in the same location on the
screen with a movable dividing line, so that one image is seen on
one side of the line and the other is seen on the other side.
5.
Subtraction images is a simple method that is particularly useful
in serial studies of the same subject. Subtraction of one registered
image from another provides a direct display of change, which can
then be related back to the source images. Subtraction images can
be viewed separately (see Chapter 7) or overlaid on a source image,
for example in color.
6.
Displaying images in a standard space (e.g., an atlas such as that due
to Talairach
16
) can help in comparison of images from different
subjects. The data can be displayed directly, segmented features of
interest can be positioned within the standard space, or features
can be referred to simply by their coordinate locations within the
standard space (see Chapter 14 for examples).
7.
Segmenting a surface from one modality, and generating a ren-
dered surface in which intensity is derived from the surface orien-
tation but hue comes from a second set of registered images.
When nonrigid registration methods have been used, critical information
may reside not just in the final images produced or difference images derived
from them, but also in the deformations required to transform one image into
the coordinates of the other. An example of this is a study of growth patterns
in the brain during childhood, where brain images from subjects acquired
serially over several years were aligned using transformations that included
spatial distortions. The pattern of distortions required to make an early scan
match a later one then reveals the pattern of growth in the intervening time.
17
Display of deformation fields is a major challenge because the deformations
are generally spatial distributions of displacements that may vary in magni-
tude and direction from point to point in the image space. Such vector distri-
butions, or in some cases tensor distributions (e.g., strain maps), are not
easily displayed on flat screens designed to display variations in intensity
