Sorry, I made a typo. Here's how it should have read:
I changed the list from
SUBJECTS='a36 A59 f10 f18 f20'; ONE_SUBJECT='a36'
to
SUBJECTS='A59 f10 f18 f20 a36'; ONE_SUBJECT='A59';
GCAupdateNodeCovariance(15, 35, 23, 0): could not find label.
(Does this mean that mri_ca_train is looking at the voxel centered at (x=15, y=35, z=23) for each subject, and can't compute a variance because there's no label to be found there in at least one subject?)
Sorry, but I'm not sure how to determine which subject is causing the failure....
Jason
On Wed, 10 Dec 2008, Bruce Fischl wrote:
sorry, not sure I understand. If you put the subject that causes the failure first in the list does it still cause the failure? On Wed, 10 Dec 2008, Jason S. Minamora wrote:
Hi Bruce,
I changed the list from
SUBJECTS='a36 A59 f10 f18 f20'; ONE_SUBJECT='a36'
to
SUBJECTS='a36 A59 f10 f18 f20'; ONE_SUBJECT='a36';
and got the same output, except for the line
GCAupdateNodeCovariance(15, 35, 23, 0): could not find label.
Using
SUBJECTS='A59 f10 f18 f20 a36'; ONE_SUBJECT='A59';
didn't solve the problem (output same as previous).
Here are the data on my version of FS:
FREESURFER_HOME: /Applications/freesurfer
Build stamp: freesurfer-Darwin-leopard-i686-stable-pub-v4.1.0
Kernel info: Darwin 9.5.0 i386
Cheers, Jason
On Wed, 10 Dec 2008, Bruce Fischl wrote:
if you change the order of subjects in your list to mri_ca_train does it change anyything? What version are you using? On Wed, 10 Dec 2008, Jason S. Minamora wrote:
Hello,
I asked about this issue about a month ago, but haven't gotten a response. I'm trying to build a new GCA using the rebuild_gca_atlas.csh script, with checking disabled (no `-check' flag). My subjects are completely acallosal, and each of the edited segmentation volumes that the script wants (seg_edited.mgz) were created by converting the corresponding aseg.mgz to one-file gzipped NIFTI (aseg.nii.gz), "stamping" manually-labeled ventricle masks on it, with the command
fslmaths aseg.nii.gz -mul left-lat-inverse-mask.nii.gz -mul right-lat-inverse-mask.nii.gz -add left-lat-mask-4.nii.gz -add right-lat-mask-43.nii.gz seg_edited.nii.gz
(here "4" and "43" mean that each of the *-lat-mask-* files contain voxels with the respective value, based on the entries in FreeSurferColorLUT.txt)
and finally converting the edited segmentation volume back to MNI format with `mri_convert -odt uchar -ns 1 seg_edited.nii.gz seg_edited.mgz'. I then copied the seg_edited.mgz into $SUBJECTS_DIR/<subj-id>/mri, and embedded the talairach transformation matrix in it using `mri_add_xform_to_header transform/talairach.xfm seg_edited.mgz'.
The first stage of the script worked OK, but the second stage (which includes a call to mri_ca_train), didn't finish:
============
mri_ca_train using all subjects, using talairach_one.m3z, producing /Users/j sm/work/subjects/average/RB_all_2008-12-10.gca... reading input 0: /Users/jsm/work/subjects/a36/mri/norm.mgz spacing priors every 2.0 mm spacing nodes every 4.0 mm using MR volume brain.mgz to mask input volume... reading segmentation from subject's mri/seg_edited.mgz directory reading xform from talairach_one.m3z reading T1 data from subject's mri/norm.mgz directory training on 5 subject and writing results to /Users/jsm/work/subjects/averag e/RB_all_2008-12-10.gca
processing subject a36, 1 of 5... reading volume /Users/jsm/work/subjects/a36/mri/brain.mgz for masking... gunzip -c /Users/jsm/work/subjects/a36/mri/transforms/talairach_one.m3z reading input 0: /Users/jsm/work/subjects/A59/mri/norm.mgz 15129 prior holes filled
processing subject A59, 2 of 5... reading volume /Users/jsm/work/subjects/A59/mri/brain.mgz for masking... gunzip -c /Users/jsm/work/subjects/A59/mri/transforms/talairach_one.m3z reading input 0: /Users/jsm/work/subjects/f10/mri/norm.mgz 15304 prior holes filled
processing subject f10, 3 of 5... reading volume /Users/jsm/work/subjects/f10/mri/brain.mgz for masking... gunzip -c /Users/jsm/work/subjects/f10/mri/transforms/talairach_one.m3z reading input 0: /Users/jsm/work/subjects/f18/mri/norm.mgz 851 prior holes filled
processing subject f18, 4 of 5... reading volume /Users/jsm/work/subjects/f18/mri/brain.mgz for masking... gunzip -c /Users/jsm/work/subjects/f18/mri/transforms/talairach_one.m3z reading input 0: /Users/jsm/work/subjects/f20/mri/norm.mgz 36063 prior holes filled
processing subject f20, 5 of 5... reading volume /Users/jsm/work/subjects/f20/mri/brain.mgz for masking... gunzip -c /Users/jsm/work/subjects/f20/mri/transforms/talairach_one.m3z GCAupdateNodeCovariance(21, 33, 6, 42): could not find label
=================
This problem _doesn't_ occur when I only use one subject. Any advice?
Thanks, Jason
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer