| I am using spm 99 and have a few worries about the spatial normalization of
| my PET data. When I looked at the defaults bounding box and the PET template
| I noticed that the value were different (91 109 91 for the template, smaller
| for the bounding box). In SPM 96 there was no difference. Is it a problem?
I'm afraid that I don't quite follow. The PET template in SPM99 is the same
size as that used by SPM96, and the default bounding box used to write the
spatially normalised images should also be the same.
| Similarly, I noticed that the defaults non linear basis function parameters
| are 7x8x7. It was set at 4x5x4 in SPM 96. I've read in the spm archives that
| 7x8x7 is suitable for T1 MRI images but not for PET images. Shall I go back
| to the 4x5x4 value ?
More basis functions generally works better than having fewer when the images
can be easily matched to the template. It is sometimes better to use fewer
basis functions if the brain images contain lesions, or if the image contrast
differs slightly from that of the templates. Alternatively, the amount of
regularisation can be varied in order to modify the amount of allowable warping.
The main reason for the extra basis functions used in SPM99, is that spatial
normalisation in SPM99 tends to be much more stable than the version in SPM96.
All the best,
-John
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|