Hi Freesufer folks-
I was just about to start constructing a custom aseg atlas and I wondered if there were any recommendations for any of the optional flags? Any parameters yield better results (especially for specific brain regions)? I was focused on improving amygdala segmentation for my structural images (hence my asking).
Best, jamie.
Jamie,
Included in the $FREESURFER_HOME/bin directory is a script called 'rebuild_gca_atlas.csh' that we use to rebuild our atlas. You will need to read and understand this script as some edits to it will need to be made. It contains descriptive text within the file.
The script is intended to run on our cluster (collection of processor nodes), so you would have to change lines like this:
if ($RunIt) pbsubmit ${PBCONF} -c "$cmd"
to:
if ($RunIt) $cmd
so that it just runs the command locally. note that this script submits multiple subjects to multiple nodes, so if you have 30 subjects, and you dont use a cluster, it will take much much much longer to complete.
Nick
On Tue, 2009-04-07 at 12:16 -0500, Jamie Hanson wrote:
Hi Freesufer folks-
I was just about to start constructing a custom aseg atlas and I wondered if there were any recommendations for any of the optional flags? Any parameters yield better results (especially for specific brain regions)? I was focused on improving amygdala segmentation for my structural images (hence my asking).
Best, jamie.
so that it just runs the command locally. note that this script submits multiple subjects to multiple nodes, so if you have 30 subjects, and you dont use a cluster, it will take much much much longer to complete.
Hi Nick-
I am working on that script right now. One quick question about this, is there an example of the subjects.csh file (to specify files for training?). I understand I need to set for ONE_SUBJECT variable, but I wasn't sure how the other SUBJECTS should be specified (and couldn't find any other examples on-line or in my version of freesurfer).
RE: computation load / time... It sounds like 2 days w/ multiple nodes... We don't do any job queue / cluster batching system, so for a single machine, what is a rough ball-park for completion time?
Thanks much! jamie.
Jamie,
Included in the $FREESURFER_HOME/bin directory is a script called 'rebuild_gca_atlas.csh' that we use to rebuild our atlas. You will need to read and understand this script as some edits to it will need to be made. It contains descriptive text within the file.
The script is intended to run on our cluster (collection of processor nodes), so you would have to change lines like this:
if ($RunIt) pbsubmit ${PBCONF} -c "$cmd"
to:
if ($RunIt) $cmd
so that it just runs the command locally. note that this script submits multiple subjects to multiple nodes, so if you have 30 subjects, and you dont use a cluster, it will take much much much longer to complete.
Nick
On Tue, 2009-04-07 at 12:16 -0500, Jamie Hanson wrote:
Hi Freesufer folks-
I was just about to start constructing a custom aseg atlas and I wondered if there were any recommendations for any of the optional flags? Any parameters yield better results (especially for specific brain regions)? I was focused on improving amygdala segmentation for my structural images (hence my asking).
Best, jamie.
Hi Jamie,
SUBJECTS should specify all the subjects you want to put in the atlas (all the manually labeled ones).
cheers Bruce
On Wed, 8 Apr 2009, Jamie Hanson wrote:
so that it just runs the command locally. note that this script submits multiple subjects to multiple nodes, so if you have 30 subjects, and you dont use a cluster, it will take much much much longer to complete.
Hi Nick-
I am working on that script right now. One quick question about this, is there an example of the subjects.csh file (to specify files for training?). I understand I need to set for ONE_SUBJECT variable, but I wasn't sure how the other SUBJECTS should be specified (and couldn't find any other examples on-line or in my version of freesurfer).
RE: computation load / time... It sounds like 2 days w/ multiple nodes... We don't do any job queue / cluster batching system, so for a single machine, what is a rough ball-park for completion time?
Thanks much! jamie.
Jamie,
Included in the $FREESURFER_HOME/bin directory is a script called 'rebuild_gca_atlas.csh' that we use to rebuild our atlas. You will need to read and understand this script as some edits to it will need to be made. It contains descriptive text within the file.
The script is intended to run on our cluster (collection of processor nodes), so you would have to change lines like this:
if ($RunIt) pbsubmit ${PBCONF} -c "$cmd"
to:
if ($RunIt) $cmd
so that it just runs the command locally. note that this script submits multiple subjects to multiple nodes, so if you have 30 subjects, and you dont use a cluster, it will take much much much longer to complete.
Nick
On Tue, 2009-04-07 at 12:16 -0500, Jamie Hanson wrote:
Hi Freesufer folks-
I was just about to start constructing a custom aseg atlas and I wondered if there were any recommendations for any of the optional flags? Any parameters yield better results (especially for specific brain regions)? I was focused on improving amygdala segmentation for my structural images (hence my asking).
Best, jamie.
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
Jamie Hanson -
Nick Schmansky