On Fri, 26 Nov 2004, Marco Migani wrote:
> we have to read an analyze image of unsigned char (DT_UNSIGNED_CHAR on
> analyze header) obtained segmenting functional EPI coregistered with T1
> we notice that opening the volume with spm_read_vols, values stored by V
> lies between 0 and 1 while we read values between 0 and 255.
> 1)does make spm some normalization to data ?
Yes, spm uses a scaling factor to spread the intensity range of your image
voxel data [0...1] to the intensity range allowed by the datatype
[0...255]. If you display an image within SPM, then in the lower right
information box you will find something like "Intensity: Y = sf*X" - this
is your scaling factor. It is also returned by spm_vol in the field pinfo,
see the help for spm_vol for details.
> 2) our code reads so less > 0 values than spm.
> Is thas due to some kind of interpolation or in V are stored raw data?
If you used spm_read_vols to get the voxel data, then no interpolation
takes place within SPM.
> 3) we would use this segmented (white matter) image as a mask to take out
> voxels from correlation analisys.
> Non-WMatter voxel are set all to 0 or there some threshold to control?
The segmented images have a range between 0 and 1 corresponding to the
probability of a given voxel to belong to either gray, white, or csf
compartments.
--
Volkmar Glauche
-
Department of Neurology E-Mail [log in to unmask]
UKE Hamburg Phone 49(0)40-42803-5781
Martinistr. 52 Fax 49(0)40-42803-9955
20246 Hamburg
|