Hi Marco,
SureFit's Raw2Minc (or Analyze2Minc) always converted short/float to
unsigned byte. The quality of Caret-generated surfaces will be equal to
that of SureFit-generated surfaces (not bad, in my opinion). But in
Caret, once you've segmented, patched, etc., you can view your surface
with the *unscaled*, full-brain (*.LR.*) volume, if you so choose; just
make sure its origin matches that of the pre-cropped, but AC-set volume
used for the SureFit segmentation. Does that make sense?
Donna
On 06/09/2005 04:23 PM, marco tettamanti wrote:
Hi Donna,
yes! I have converted my original data from 16 to 8-bit format and
segmentation now seems to run fine.
Why does Caret only read unsigned byte volumes? SureFit read 16-bit volumes
without any problems (or are they converted to 8-bit in backgound?).
Isn't the loss of intensity associated with converting to unsigned format
going to be a problem for surface image quality?
Thanks a lot,
Marco
Hi Marco,
Very nice structural volume. My Caret gives a very nice segmentation
with no complaints, but I had to scale it from float to byte (not sure
what your original datum type was). We ported the SureFit algorithm
almost completely "as is," and its "SureFitOps" utilities work only on
unsigned byte (0-255) volumes. Shortly after we released Caret 5.2,
John added some features to rescale the volume in the Edit Volume
Attributes menu. If your Caret doesn't have a Data tab on the Volume
Attributes Editor with a button that says "Rescale Voxels...," then you
can get an updated version here:
http://brainmap.wustl.edu/pub/john/mcw/
login pub
password download
Alternatively, you can write your original full brain volume in unsigned
byte form and input it to Caret. There should be no need to rescale if
the input is byte.
I think this probably is the problem (relatively easy fix).
Donna
On 06/09/2005 02:14 PM, marco tettamanti wrote:
Hi Donna,
I remember you too! You have provided me a valuable help quite many times.
Thank you for your quick reply.
I have uploaded the following files:
Human.lvs7b.L.params_file_19.2005-06-09.params
wmean_anat3D_AC_L+orig.BRIK
wmean_anat3D_AC_L+orig.HEAD
Processing the same images with SureFit 5.2 is not a problem, so I do not
think there is a problem with the data; but I hope it turns out I a wrong!
All the best,
Marco
Hi Marco,
I remember you! You helped educate me about the differences between
talairach and MNI space a couple of years ago.
Is there any chance you're trying to segment both hemispheres at once?
If not, then can you upload your cropped .HEAD and .BRIK structural
volumes along with your params file (or just tell me your GM/WM peaks)
here:
http://pulvinar.wustl.edu/cgi-bin/upload.cgi
Other potential problems that can cause such problems:
* AC not set correctly
* Bad peaks
The best way for me to help is to look at your volume.
Donna
_______________________________________________
caret-users mailing list
caret-users@brainvis.wustl.edu
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
