Now, we wish to view the integrated images. If you are using a colour console or an monochrome X terminal, enter the following to create a MATLAB figure window and display the image obtained by rectangular integration:
figure; viewimage (PETsummed1);(Note that explicitly creating the window via the
figure
command is not really necessary. However, if you already had a
figure window with a plot or image that you did not want to lose,
MATLAB would have overwritten the existing figure when you called
viewimage. It is generally good practice to call
figure
before creating new plots or images, unless you are
sure you wish to obliterate whatever was previously in the current
figure window.)
You may want to title this window:
title ('yates_19445, slice 8, rectangular integration');
Now do the same thing for PETsummed2
. You will probably want
to move one or both figure windows so that you can see both images
simultaneously. Clearly, the two images are similar--but if we want
a more objective comparison, we can calculate the ratio of the two
images on a pixel-by-pixel basis and view this:
ratio = PETsummed1 ./ PETsummed2; figure; viewimage (ratio);
Note the use of the ./
operator to specify an
element-by-element operation on two matrices. The resulting image
shows that most of the pixels in PETsummed1
and
PETsummed2
appear fairly close. However, because some pixels
have such large values compared to 1, any small variations around 1
are swamped. One way to deal with this is simply to clip the image
at certain points: for example, set all points below -10 to -10, and
all points above 10 to 10. This is accomplished as follows:
clipneg = find (ratio < -10); ratio (clipneg) = ones (size (clipneg)) * (-10); clippos = find (ratio > 10); ratio (clippos) = ones (size (clippos)) * (10);
If you wish to know how many points were clipped at either end,
type size (clipneg)
or size (clippos)
.
If you now viewimage (ratio), it should be clear that the overall ratio between the two integrated images is close to 1 inside the head.