Hi Tim,
When I saw your cropped structural MRI volume, I thought perhaps you had
run our colin brain through FSL's brain extraction tool (BET) before
segmenting it.
Something in your preprocessing is pumping the intensities in the white
matter behind the eye up 9% above the white matter peak, which is
causing SureFit to mistake it for eye fat. The fact that the real eye
fat already has been removed by pre-processing probably contributes to
this mistake. If you have a version with the skull/eye in tact, it
might do better in SureFit. But I'm guessing some non-uniformity
correction (NUC) might help with this volume.
Currently, SureFit doesn't have a built-in NUC tool, but we hope to
change that someday. Since I'm guessing you work with Dr. David
Rottenberg, who co-authored a paper comparing NUC tools, you probably
have your pick of tools. I'll try running your volume through FSL's
fast and see how the bias-corrected volume does in SureFit.
Donna
On 06/23/2004 02:20 PM, jarv0075 wrote:
On 23 Jun 2004, Donna Hanlon wrote:
Hi Timothy,
Hmmm. I just to recreate your problem using the output of 3dAFNItoMINC
Human.colin.LR.TLRC-711-2B.111+tlrc.HEAD, and my initial segmentation
after filling ventricles, but before automatic error correction, had no
missing chunks like yours. I'd like to see your cropped structural
*.L.full*sMRI.mnc file and .params file. Could you upload them here:
http://pulvinar.wustl.edu/cgi-bin/upload.cgi
How did you convert the dataset to minc?
There are some significant non-uniformity issues that relate to what I
call the "stripey minc" problem
(http://brainmap.wustl.edu/SureFit/RELEASE_NOTES.html#mincscaling),
which has since been fixed in SureFit, but which remain in our atlas
dataset. But based on my trial, this is not causing the problem you
report.
Thanks,
Donna
Donna,
Thanks for the quick response!
I just uploaded "input_tal.L.full.sMRI.params" and
"input_tal.L.full.sMRI.mnc"
The original dataset was a local analyze format version of colin, which was
then converted to minc using our own idl routines, and subsequently
converted to the standard taliarach space used by Insect/Classify (from
MNI).
-Tim
---------------------------------------
Timothy R. Jarvis
---------------------------------------
Graduate Research Assistant
International Neuroimaging Consortium
VA Medical Center
612-467-2619
http://www.neurovia.umn.edu
---------------------------------------
_______________________________________________
caret-users mailing list
[EMAIL PROTECTED]
http://pulvinar.wustl.edu/mailman/listinfo/caret-users
