Dear SPMer's,
I have a question concerning image normalization into the template of MNI.
The functional images have generally a lower resolution than the template.
For an example, the voxel size for functional images that we use is
generally 4 x 4 x 5 and the voxel size in the template given in SPM is 2
x 2 x 2. Looking at the reports from literature, I realized that in most
of these reports, the normalized images are described as exhibiting a 2 x 2
x2 resolution despite the fact that original images have a lowest
resolution. I see one disadvantage of such over-sampling ( a greater size
of images during statistical treatment and storage) and I suspect that the
results of statistical analysis performed with the over-sampled images can
be questionable ( overestimated size of clusters, modification of spatial
correlations, results depending of the interpolation method). So my
questions are: what is the benefit of writing the normalized image with a
greater resolution than the original images? Is it some risk concerning the
validity of the statistical analysis performed with over-sampled images?
I thank you in advance for any comment.
Pierre
Pierre Fonlupt
INSERM - Unité 280
Processus Mentaux et Activation Cérébrale
151 Cours Albert Thomas
69424 LYON CEDEX 03
Tél : 33 (0) 6 60 54 68 29
Fax :33 (0) 4 72 68 19 02
|