Hello Tomas,
We used SPM to solve this issue in single voxel MRS.
Basically it is simple; as soon as you have the coordinates/size of the
spectroscopy voxel counted in voxels of your T1-weighted scan, you can create
make using this information and cut out grey matter, white matter and CSF
from a segmentated volume.
Proportions of gm.wm.csf or brain.csf can then be calculated for the VOI.
Segmentation could for this purpose be done with the default settings in SPM2
(or 5b).
The tricky part of all this is transforming the coord. of the MRS-system to
your 3D volume - usually the scanner writes out the zero-ccordinates of
either the midpoint of each slices or the edge of a slice in the absolute
scanner coordinate system which is usually identical with the system in which
the MRS coords are given.
You then have to calculate the voxel in each dimension that is first and last
covered by your spectroscopy voxel. Without going into matlab or IDL to
process the images it may be difficult to do this. If the angle of the voxel
is not parallel to the slices it may even become a more complicated.
So, in brief: yes, SPM can deliver you well segmentated images, but to
transfer the coords of the MRS voxel into your 3D volume probably takes some
programming.
Another point is the calculation of the GM/WM/CSF ratios: we preferred a hard
segmentation and binarized the images, so each voxel is assigned one tissue
class - this is debatable and you might add up the probability maps to a
volume.
Best regards,
Philipp Saemann
Max-Planck-Institute of Psychiatry
NMR Study Group
Kraepelinstr 2-10
80804 Munich
|