no, they should be the same
On Tue, 24 Jul 2018, Rosemary Nicholas wrote:
External Email - Use Caution
Hi Andrew,
Thanks!
So as long as we set the seed in mris_ca_label the results should be
consistent, right? Would there be any difference to running it this way
(independently) and running it with recon-all?
Thanks,
Rosemary
On Mon, Jul 23, 2018 at 6:27 PM, Hoopes, Andrew <ahoo...@mgh.harvard.edu>
wrote:
Hi Rosemary, the seeds for those commands are set from
recon-all. The individual binaries do not seed themselves with a
default value, so it’s expected that mris_ca_label produces
inconsistent results when run independently (and without the
-seed flag). The algorithm processes the surface vertices in a
random order to help prevent any spatial ordering effects, so
the given seed will initialize this order.
best,
Andrew
From: <freesurfer-boun...@nmr.mgh.harvard.edu> on behalf of
Rosemary Nicholas <rosemary.nicho...@gmail.com>
Reply-To: FS Help <freesurfer@nmr.mgh.harvard.edu>
Date: Monday, July 23, 2018 at 1:22 PM
To: FS Help <freesurfer@nmr.mgh.harvard.edu>
Subject: [Freesurfer] mris_ca_label seed
External Email - Use Caution
Hello,
I am running a dataset through version 6.0 of Freesurfer on Linux,
extracting results for both the Desikan atlas and our own atlas. When
extracting the stats for our atlas, the results were varying
considerably (sometimes up to 10%) in test-retest runs.
The release notes for 4.5 indicate that the default setting for
mris_ca_label should use the same number as a seed, and therefore give
the same results, but it doesn't.
"The binaries in the recon-all stream which are subject to variability
due to usage of a random number generator (mris_smooth, mris_sphere,
mris_topology_fixer, mris_topo_fixer and mris_ca_label) are now seeded
with the same number, with the result being that repeated runs of the
recon-all stream on a subject on the same platform will always give
identical results."
The issue seems to be resolved by adding -seed 1234 as a flag to
mris_ca_label.
I would like to understand what this is actually doing though, before
applying it to all my data. What is the random number generator used
for? What does setting it to 1234 actually do? It isn't listed in
the optional flagged arguments. Is it just deciding which vertex is
labeled first?
Here is how we are running it (in two independent scripts):
FIRST:
# autorecon1 including data conversion
recon-all -i $mprage/001 -autorecon1 -subjid $s
# autorecon2 with threshold adjustments
recon-all -seg-wlo 60 -seg-ghi 70 -autorecon2 -subjid $s
# autorecon3
recon-all -autorecon3 -subjid $s
THEN:
mris_ca_label $s $h sphere.reg
/g5/dcn/freesurfer_testing/subjects/"$h".rantaFL.gcs rantaFL.annot
modified to:
mris_ca_label -seed 1234 $s $h sphere.reg
/g5/dcn/freesurfer_testing/subjects/"$h".rantaFL.gcs rantaFL.annot
Thank you,
Rosemary Nicholas
Center for Neurodevelopmental and Imaging Research
Kennedy Krieger Institute
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom
it is
addressed. If you believe this e-mail was sent to you in error and the
e-mail
contains patient information, please contact the Partners Compliance
HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you
in error
but does not contain patient information, please contact the sender
and properly
dispose of the e-mail.
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.
