Hello freesurfer experts,

I am having difficulties using mris_ca_train and mris_ca_label.  I am trying to get third party-software to accept a user-defined parcellation in the form of a .gcs file, which I created from sample data using mris_ca_train.  I have reproduced the portion of the workflow that is buggy.  When I run the software, it calls mris_ca_train, followed by mris_ca_label and mri_annotation2label.  On mri_annotation2label, it produces the following output and writes no label files:

[truffles:mridat] (nmr-stable5.1-env) mri_annotation2label --subject nmr00535_FS --hemi lh --annotation icotest --outdir nmr00535_FS/label/regen
subject = nmr00535_FS
annotation = icotest
hemi = lh
outdir = nmr00535_FS/label/regen
surface   = white

Reading surface
 /autofs/cluster/neuromind/rlaplant/mridat/nmr00535_FS/surf/lh.white
Loading annotations from /autofs/cluster/neuromind/rlaplant/mridat/nmr00535_FS/label/lh.icotest.annot
reading colortable from annotation file...
colortable with 643 entries read (originally /cluster/neuromind/rlaplant/mridat/nmr00535_FS/label/ico3_regenerated_clut_lh.txt)
Seg base 0
max index = -1

The annotation at this point is corrupted and cannot be opened in matlab and shows up empty in tksurfer.  The annotation I started with as input to mris_ca_train shows up just fine.  The commands i used for mris_ca_train and mris_ca_label are as follows:

mris_ca_train -t $TEST_DIR/ico3_clut.txt lh sphere.reg ico3_morph $SUBJECTS $TEST_DIR/lh_test.gcs                                                                  

mris_ca_label nmr00535_FS lh sphere.reg $TEST_DIR/lh_test.gcs $TEST_DIR/nmr00535_FS/label/lh.icotest.annot

Can anyone point out where my labels are being clobbered?

thanks,
R

--
Roan LaPlante
Athinoula A. Martinos Center for Biomedical Imaging