The brain masking merely assigns different weights to
different template voxels when doing the affine registration.
There have been a number of queries on the subject sent to
the SPM mailing list, particularly with respect to
normalising EPI data sets. e.g.,
http://www.mailbase.ac.uk/lists/spm/2000-02/0168.html
There is no detailed paper on the subject of weighting the
registration. The only place that I can remember briefly
mentioning it is in:
Ashburner J, Hutton C, Frackowiak RSJ, Johnsrude I, Price C
& Friston KJ (1998): "Identifying global anatomical
differences: deformation-based morphometry."
Human Brain Mapping. 6(5):348-357
I think that to do a quantitative study over time, then it may
be worth changing the strategy slightly. The registration to
the templates may vary, depending on what part of the images is
covered by your field of view. To get around this problem,
I would consider working with spatially normalised images, where
the same warps are used for all datasets:
1) Co-register the all images to the same reference image.
There is no need to reslice at this stage as the .mat files
will be incorporated in the next stage.
2) Estimate spatial normalisation parameters from
one of these images, and apply the same parameters to
all the individual images. Write the spatially normalised
images using a higher resolution to than the default (e.g.,
1x1x1mm or 1.5x1.5x1.5mm).
3) Segment the spatially normalised images.
Doing it this way should produce more consistent results, as the same
spatial transformation is used to overlay the prior probability maps
on to all the MR images.
All the best,
-John
| thanks a lot for the tips. I disabled the brain mask
| and the problem with the incorect segmentation in
| upper regions was fixed. What happens in the
| algorithm, if I disable the brain mask. I read the
| paper (Multimodal Image Coregistration and
| Partitioning ..) but I can't find a hint concerning a
| brain mask ? Does any more detailed paper exist ?
| I like to do a quantitative analyze of the segmented
| data over the time. That is why I do need preferably
| the same segmentation results for datas of the same
| patient acquired at different times.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|