Hi Nick,
I keep getting the same error with mri_ca_register. I have now tried this with two different subjects and they both fail at about the same stage. I have checked both brainmask.mgz files in tkmedit for any skull stripping or talairach registration errors and they look fine. Like you had recommended, I have also checked the messages in the /var/log directory but found nothing to suggest that the system was running out of memory. I'm fresh out of ideas. Please let me know if there's any other thing that comes to mind. I've attached the logs again for reference.
Binyam
Binyam Nardos Cognitive Rehabilitation Research Group Washington University School of Medicine Box # 8505 4444 Forest Park Blvd St. Louis, MO 63108 Office: (314) 454-7795 Fax: (314) 286-1601
The contents of this e-mail message and any attachments are intended solely for the addressee(s) named in this message. This communication is intended to be and to remain confidential and may be subject to applicable physician/patient and/or work product privileges. If you are not the intended recipient of this message, or if this message has been addressed to you in error, please immediately alert the sender by reply e-mail and then delete this message and its attachments.
-----Original Message----- From: Nick Schmansky [mailto:nicks@nmr.mgh.harvard.edu] Sent: Thu 2/8/2007 3:31 PM To: Binyam Nardos Subject: RE: [Freesurfer] error with mri_ca_register
Binyam,
My run of mri_ca_register using the stable rh9 build of freesurfer also completed successfully, so I was not able to recreate the segfault problem you are seeing.
One thing I just learned is that you can view the file:
/var/log/messages
and check it for messages that say "Out of Memory". This sort of message would appear if a freesurfer binary (or any other) is unable to allocate memory. You will need root or sudo access to read the file on your system.
Nick
On Wed, 2007-02-07 at 16:46 -0600, Binyam Nardos wrote:
Nick, upgrading my RAM is definitely on top of my list. As far as the documentation that I mentioned, I don't think it's something you can update. I just remembered that I had seen that on the Dale/Fischl/Sereno 1999 paper (Cortical Surface-Based Analysis I), which, I know is pretty old, vis-a-vis all the time-consuming changes that I'm sure have been implemented since then. In any case, I think the paper reads "The entire procedure, from intensity normalization, through segmentation, tessellation, and inflation, runs completely automatically in about 1.5h on current PC architectures ..."
Binyam
Binyam Nardos Cognitive Rehabilitation Research Group Washington University School of Medicine Box # 8505 4444 Forest Park Blvd St. Louis, MO 63108 Office: (314) 454-7795 Fax: (314) 286-1601
The contents of this e-mail message and any attachments are intended solely for the addressee(s) named in this message. This communication is intended to be and to remain confidential and may be subject to applicable physician/patient and/or work product privileges. If you are not the intended recipient of this message, or if this message has been addressed to you in error, please immediately alert the sender by reply e-mail and then delete this message and its attachments.
From: Nick Schmansky [mailto:nicks@nmr.mgh.harvard.edu] Sent: Wed 2/7/2007 4:35 PM To: Binyam Nardos Subject: RE: [Freesurfer] error with mri_ca_register
The mri_ca_register step typically takes about 10 hours to complete. It took 9 hours 37 minutes on the 2GHz machine I ran it on yesterday.
Also, 1GB of RAM is pretty low. All of our machines have at least 2GB (our minimum recommended), most 4GB. This might actually be the source of your problem, as our code does not check when it runs out of memory (very bad, I know). I'll know more when this second run completes. But if it is at all possible to upgrade to 2GB or 4GB of RAM, I *highly* recommend it. I know with certainty that when someone on the mailing list upgraded from 512MB to 2GB, these sorts of problems disappeared.
Can you point me to the documentation that says the segmentation takes 2-3 hours to complete? I'd like to update that.
On Wed, 2007-02-07 at 16:20 -0600, Binyam Nardos wrote:
thanks Nick, I tried processing another subject yesterday and again got stuck somewhere around mri_ca_register, I believe it said (segmentation error ... sorry I dont' have the details because I'm in windows
world
right now.) In any case, I had another quick question for you: I am running Freesurfer on ubuntu, which is installed on a separate
HD
on my computer. Some of my specs are 1GB RAM, 3.2GHz dual processor. In some of the documetation that I've read, a typical pc takes about 2-3 hours for complete segmentation. It's been taking me about 3
times
that (i'm actually yet to fully segment a subject successfuly ). But is this normal?
thanks again, Binyam
Binyam Nardos Cognitive Rehabilitation Research Group Washington University School of Medicine Box # 8505 4444 Forest Park Blvd St. Louis, MO 63108 Office: (314) 454-7795 Fax: (314) 286-1601
The contents of this e-mail message and any attachments are intended
solely for the addressee(s) named in this message. This communication is intended to be and to remain confidential and may be subject to applicable physician/patient and/or work product privileges. If you are not the intended recipient of this message, or if this message has been addressed to you in error, please immediately alert the sender by reply e-mail and then delete this message and its attachments.
From: Nick Schmansky [mailto:nicks@nmr.mgh.harvard.edu] Sent: Wed 2/7/2007 4:10 PM To: Binyam Nardos Subject: RE: [Freesurfer] error with mri_ca_register
Binyam,
I was able to run mri_ca_register successfully with this subject,
but
I was using a 'development' build (not the one used publicly), and I
was
running on a 64bit system. So this problem is not a blatant bug in the code.
I am going to re-run mri_ca_register again, this time using the rh9 stable build (which is what you used) to see if I can get it to segfault (the problem that you are encountering).
Nick
On Tue, 2007-02-06 at 12:01 -0600, Binyam Nardos wrote:
thanks Nick
Binyam
Binyam Nardos Cognitive Rehabilitation Research Group Washington University School of Medicine Box # 8505 4444 Forest Park Blvd St. Louis, MO 63108 Office: (314) 454-7795 Fax: (314) 286-1601
The contents of this e-mail message and any attachments are
intended
solely for the addressee(s) named in this message. This
communication
is intended to be and to remain confidential and may be subject to applicable physician/patient and/or work product privileges. If
you
are not the intended recipient of this message, or if this message
has
been addressed to you in error, please immediately alert the
sender
by
reply e-mail and then delete this message and its attachments.
-----Original Message----- From: Nick Schmansky [mailto:nicks@nmr.mgh.harvard.edu] Sent: Tue 2/6/2007 11:41 AM To: Binyam Nardos Subject: RE: [Freesurfer] error with mri_ca_register
Binyam
I've received the files and will have a look at them today.
Nick
On Tue, 2007-02-06 at 11:32 -0600, Binyam Nardos wrote:
Hi Nick, I have just uploaded the 4 mgz files for the subject. I hope
that's
what you were looking for. I checked the skull-strip and
talairach
registration and it looks fine.
thanks, Binyam
Binyam Nardos Cognitive Rehabilitation Research Group Washington University School of Medicine Box # 8505 4444 Forest Park Blvd St. Louis, MO 63108 Office: (314) 454-7795 Fax: (314) 286-1601
The contents of this e-mail message and any attachments are
intended
solely for the addressee(s) named in this message. This
communication
is intended to be and to remain confidential and may be subject
to
applicable physician/patient and/or work product privileges. If
you
are not the intended recipient of this message, or if this
message
has
been addressed to you in error, please immediately alert the
sender
by
reply e-mail and then delete this message and its attachments.
-----Original Message----- From: Nick Schmansky [mailto:nicks@nmr.mgh.harvard.edu] Sent: Tue 2/6/2007 11:18 AM To: Binyam Nardos Cc: Freesurfer Mailing List Subject: Re: [Freesurfer] error with mri_ca_register
Binyam,
The recon-all.log indicates that mri_ca_register hit a
segmentation
fault, which is bad. Can you send me the subject data so that I
can
try to recreate this segfault in a debugger? You can upload to me
via
the
filedrop:
https://www.nmr.mgh.harvard.edu/facility/filedrop/index.html
In the meantime, also look at the brainmask.mgz file with
tkmedit
and
confirm that it is skull-stripped and that the cerebellum is
attached.
Also check the talairach registration with tkregister2.
Nick
On Tue, 2007-02-06 at 10:39 -0600, Binyam Nardos wrote:
hi all, I just ran autorecon2 on one of our subjects (for the second
time,
after getting a similar error, i.e. ERROR: mri_ca_register
with
non-
zero status. I just rerun recon-all with -autorecon2 without removing/changing anything produced by the previous
autorecon2).
Is
there any logical place to start to fix the problem or do I
have
to
rerun mri_ca_register as shown below:
mri_ca_register -align -nobigventricles -mask brainmask.mgz -T transforms/talairach.lta norm.mgz /usr/local/freesurfer//average/RB_all_2006-02-15.gca transforms/talairach.m3z
I have attached the recon-all log for reference. Please advise,
thanks, Binyam
Binyam Nardos Cognitive Rehabilitation Research Group Washington University School of Medicine Box # 8505 4444 Forest Park Blvd St. Louis, MO 63108 Office: (314) 454-7795 Fax: (314) 286-1601
The contents of this e-mail message and any attachments are
intended
solely for the addressee(s) named in this message. This
communication
is intended to be and to remain confidential and may be
subject
to
applicable physician/patient and/or work product privileges.
If
you
are not the intended recipient of this message, or if this
message
has
been addressed to you in error, please immediately alert the
sender
by
reply e-mail and then delete this message and its attachments.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer