Elena-
This is a tricky issue. As I understand it, you are using several template
images, and choose the one with the closest pitch to your CT scan for
mapping the lesion. One option is to use SPM to compute the normalization
parameters for each of your templates, and then use that spatial rotation
matrix for each corresponding lesion map. The trick is that this would
really require you to draw the lesion on each slice of your template
images. A sort of sloppy alternative would be to copy/paste your lesion
maps into the intermediate slices where you have not drawn a lesion, and
then use save the data with a wide smoothing function to try to remove the
stepping-effects of the edges.
However, I honestly think the best solution is to have a human expert map
all the lesions to a single template with the same orientation. Classical
CT scans have large inter-slice spacing, and it therefore you need a human
to identify the critical regions.
-chris
On Fri, 8 Feb 2008 18:14:18 +0000, Eleni Papageorgiou
<[log in to unmask]> wrote:
>Dear MRICRO-Users,
>
>I am trying to perform a subtraction analysis between 2 patient groups and
I
>have the following problem: Since in some patients we don’t have any
digital
>images of the patient MRI-scans, we draw the ROIs (strokes) by hand onto
the
>template brain of the MRICRO. However, we cannot make overlap images or
>subtract the ROIs of the patient groups, because the slices we used to draw
>the ROIs and their rotation are not identical in every subject (in order to
>fit the orientation of the actual MRI-slice).
>
>1.Is there a way to “interpolate” ROIs on slices that are
between those that
>have the ROI mapped?
>2.And how can we solve the problem of the variability in rotation
>coordinates across patients?
>
>Thanks for any suggestions,
>Elena Papageorgiou
>
>Centre for Ophthalmology
>University of Tuebingen
>Schleichstrasse 12-16
>72076 Tuebingen, Germany
>[log in to unmask]
>========================================================================
|