Hi Kai,
Quoting Kai Li <[log in to unmask]>:
[snip]
> I asked this question because I saw a "Smooth" button and curious if
> I should use it before segmentation.
>
The "smooth" button is most important as a preprocessing step before
analysis of
functional data in a voxelwise way. It is not usually used for analysis of
structural images. Basically, a voxelwise analysis does an independent
test at
each voxel. However, the voxels are *not* independent-- they are correlated
with their close neighbors. The smoothing (or blurring) hinders a little in
that it spreads out functional activity, but helps much more in that it allows
us to estimate the degree of spatial covariance in a robust way.
[snip]
>> The values are the probability that each voxel is of a given tissue
>> class. They
>> should fall on the range [0 .. 1].
>
> I just tried SPM5 on several datasets and the values are actually in
> the range [0 .. 255] ( I selected the "Native space" option). Can I
> just rescale them to let them fall on the range [0 .. 1]?
>
Easy answer: yes you can rescale them after reading them into matlab.
Now Im not completely sure about the next part, but here goes. The data are
probably stored in the Analyze file as uint8's. The Analyze file
format allows
for a value in the header that is used to scale the images after reading them
in. It might be the case that in your header this is set to 1/255, indicating
that the data should be interpreted as falling on [0 .. 1]. This scalefactor
is not always used by all programs that read Analyze (or Nifti) data.
Ken
----------------------------------------------------
Ken Roberts
Woldorff Laboratory
Center for Cognitive Neuroscience, Duke University
(919) 668-1334
----------------------------------------------------
|