Freesurfer

freesurfer@nmr.mgh.harvard.edu

17 participants
30130 discussions

RE: [Freesurfer] recon-all library dependencies
by James Voyvodic 31 Oct '05

31 Oct '05

Is there a version of Linux that we can get that will be fairly sure to be able to run FreeSurfer's RECON-ALL sequence? We have been using many different versions of FreeSurfer through the old CSURF interface on several different Linux installations (RH 7-9, Fedora Core 1). Now we want to upgrade to a recent recon-all command line version. So far, I've never been able to get recon-all to run successfully (always crashes part way through) on either RedHat 9 or Fedora Core 1. The details of my most recent attempt with the newest RH9 release are shown below. What Linux should I choose to have the best chance of running FreeSurfer? Ideally I would like to install the same version being used at MGH. Is that RH9, CentOS, Fedora? I'm running on a Dell Intel with plenty of power. Alternatively I would be happy to upgrade my RedHat 9 system to the libraries FreeSurfer wants if I knew which versions I should upgrade to (and where to find them). Do other successful recon-all users have any recommendations? thanks, Jim =============== Redhat 9 attempt to run "recon-all" using the recent Dev version: freesurfer-Linux-rh9-dev20051003-full (I had successfully reconstructed this data set using an earlier csurf version). recon-all -subjid k1140 autorecon1 ... (many steps apparently worked, and then we got the following) [jim@hawking:/data2/surfdata/subjects/k1140/mri/] [2005-10-28 14:05:25] running: /usr/local/mni/bin/nu_estimate_np_and_em -parzen -log -sharpen 0.15 0.01 -iterations 50 -stop 0.001 -shrink 4 -auto_mask -nonotify -b_spline 1 -distance 200 -quiet -execute -clobber -nokeeptmp -tmpdir /usr/tmp/nu_correct_14439/ nu0.mnc nu1.imp /usr/local/mni/bin/volume_stats: /lib/libgcc_s.so.1: version `GCC_3.3' not found (required by /usr/lib/libstdc++.so.6) nu_estimate_np_and_em: crashed while running volume_stats (termination status=256) nu_correct: crashed while running nu_estimate_np_and_em (termination status=256) ERROR: nu_correct Linux hawking 2.4.20-31.9 #1 Tue Apr 13 18:04:23 EDT 2004 i686 i686 i386 GNU/Linux recon-all exited with errors at Fri Oct 28 14:05:25 EDT 2005 (The non-development release, freesurfer-Linux-rh9-v2.1, stopped earlier than this with probelms finding tixwish libraries.) -----Original Message----- From: Nick Schmansky [mailto:nicks@nmr.mgh.harvard.edu] Sent: Wednesday, October 26, 2005 5:35 PM To: James Voyvodic Cc: Freesurfer Mailing List Subject: Re: [Freesurfer] recon-all library dependencies Jim, The short answer is that I think you may need to have your Linux installation upgraded with newer versions of the 'base' libraries. Freesurfer requires, but does not include, a number of libraries that are normally included in most Linux installations. The newer 'dev' version of Freesurfer includes the following libraries in its 'lib' directory (which themselves are built against the 'base' libraries, as you have discovered): GSL, Tcl/Tk/Tix/BLT, the MINC tools, and Qt To see which libraries a freesurfer binary depends upon, use the 'ldd' command. On a rh7.3 system of ours (the oldest Linux platform we support), here is the output for one of the binaries, where the 'base' libraries are the ones in the /usr/lib and /usr/X11 directories: [martinos01:~] (nmr-dev-env) ldd `which tkmedit` libtix.so.0 => /usr/local/freesurfer/dev/lib/tcltktixblt/lib/libtix.so.0 (0x40015000) libtk8.4.so => /usr/local/freesurfer/dev/lib/tcltktixblt/lib/libtk8.4.so (0x4004f000) libtcl8.4.so => /usr/local/freesurfer/dev/lib/tcltktixblt/lib/libtcl8.4.so (0x40112000) libglut.so.3 => /usr/lib/libglut.so.3 (0x401c8000) libSM.so.6 => /usr/X11R6/lib/libSM.so.6 (0x40204000) libICE.so.6 => /usr/X11R6/lib/libICE.so.6 (0x4020c000) libXmu.so.6 => /usr/X11R6/lib/libXmu.so.6 (0x40223000) libXi.so.6 => /usr/X11R6/lib/libXi.so.6 (0x40239000) libGLU.so.1 => /usr/lib/libGLU.so.1 (0x40241000) libGL.so.1 => /usr/lib/libGL.so.1 (0x402bc000) libXext.so.6 => /usr/X11R6/lib/libXext.so.6 (0x40318000) libdl.so.2 => /lib/libdl.so.2 (0x40325000) libX11.so.6 => /usr/X11R6/lib/libX11.so.6 (0x40328000) libgsl.so.0 => /usr/pubsw/packages/gsl/current/lib/libgsl.so.0 (0x403fe000) libgslcblas.so.0 => /usr/pubsw/packages/gsl/current/lib/libgslcblas.so.0 (0x4055a000) libjpeg.so.62 => /usr/lib/libjpeg.so.62 (0x40587000) libtiff.so.3 => /usr/lib/libtiff.so.3 (0x405a5000) libz.so.1 => /usr/lib/libz.so.1 (0x405e9000) libm.so.6 => /lib/i686/libm.so.6 (0x405f7000) libcrypt.so.1 => /lib/libcrypt.so.1 (0x4061a000) libc.so.6 => /lib/i686/libc.so.6 (0x42000000) libXt.so.6 => /usr/X11R6/lib/libXt.so.6 (0x40647000) libstdc++-libc6.2-2.so.3 => /usr/lib/libstdc++-libc6.2-2.so.3 (0x40693000) libGLcore.so.1 => /usr/lib/libGLcore.so.1 (0x406d6000) /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000) So you can see that most of the 'base' libraries are version 6. The same is true for rh9. You can compare this output to your own output from the same command to discover which libraries on your system are old. Also, I noticed that your mni tools seem to be in the /usr/local/mni directory, rather than in the $FREESURFER_HOME/lib/mni directory. You may have a lingering MINC_BIN_DIR statement in your .cshrc file, pointing to a prior installation of the MINC tools, but I don't think this is the heart of your problem (old libraries). Unfortunately, there is no 'universal' freesurfer distribution for this very reason that Linux platforms have varying versions of the base libraries. If you have a sysadmin, perhaps he can upgrade the libs on your rh9 system. The newer Linux installations, like CentOS 4, have automatic upgrade tools, which alleviate this problem, and is why we will soon be abandoning our own rh7.3 and rh9 Linux systems at the MGH NMR center (which require manual upgrade). Lastly, you can restart recon-all from different stages. Run 'recon-all -help' to see the different stages (basically autorecon1, autorecon2 and autorecon3). Nick On Wed, 2005-10-26 at 16:49 -0400, James Voyvodic wrote: > I've upgraded to the latest development release > (freesurfer-Linux-rh9-dev20051003-full.tar.gz) running > on Linux Redhat 9 on an x86 platform. > Recon-all crashes, apparently because of failed library dependencies, with > the following message: > > /usr/local/mni/bin/volume_stats: /usr/lib/libstdc++.so.6: version > `CXXABI_1.3' not found (required by /usr/local/mni/bin/volume_stats) > /usr/local/mni/bin/volume_stats: /usr/lib/libstdc++.so.6: version > `GLIBCXX_3.4' not found (required by /usr/local/mni/bin/volume_stats) > nu_estimate_np_and_em: crashed while running volume_stats (termination > status=256) > nu_correct: crashed while running nu_estimate_np_and_em (termination > status=256) > ERROR: nu_correct > Linux hawking 2.4.20-31.9 #1 Tue Apr 13 18:04:23 EDT 2004 i686 i686 i386 > GNU/Linux > recon-all exited with errors at Wed Oct 26 15:17:17 EDT 2005 > > It had previously crashed because it couldn't find libstdc++.so.6 at all, so > I tried to appease it by using the old trick of linking libstdc++.so.5 to > libstdc++.so.6, but obviously that wasn't enough. > > So the main question is: what libraries are needed to run the new version(s) > of FreeSurfer? > Is there a way to download the missing libraries from somewhere? > > When I installed and tried to run the most recent stable release > (freesurfer-Linux-rh9-v2.1.tar.gz) I was foiled because I didn't have > libtixwish installed, which is why I went to the development release that > had more libraries. Yum says I have libstdc++ and doesn't offer a .so.6 > variant. Do I need a different version of Linux? > > A side question: If recon-all crashes part way through, is there a way to > skip the successful steps when restarting or does it have to redo all the > steps in its autorecon (e.g. autorecon1) script? > > thanks, > Jim > > > ======================== > James T. Voyvodic, Ph.D. > Brain Imaging and Analysis Center > Duke University Medical Center > e-mail: jim.voyvodic(a)duke.edu > > _______________________________________________ > Freesurfer mailing list > Freesurfer(a)nmr.mgh.harvard.edu > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > >

2 1

Re: [Freesurfer] good news and bad news
by Bruce Fischl 28 Oct '05

28 Oct '05

Hi Greg, it's: mris_ca_train lh sphere.reg my_annot_file <subject 1> ... lh.my_annot_atlas.gcs. -autorecon3 won't pick it up, you'll have to specify it by hand I think, but Doug would know. cheers, Bruce On Fri, 28 Oct 2005, Greg Harris wrote: > Hi Bruce. > > mris_sample_parc seems straightforward enough, but I'll need a few > hints about mris_ca_train. I get this usage line: > > usage: mris_ca_train [options] <hemi> <canon surf> <annot file> <subject 1> > <subject 2> ... <output file> > > Apparently, none of the options is '--help' or '-h', so I have no list of > options. > Is <canon surf> a standard component of the freesurfer distribution? > Is <annot file> a table of names, labels and colors? If so, where is the > file that > drives the generation of the file label/aparc_annot.ctab in each subject? > What do I want for <output file>? Presumably this is a special file name > that > will be picked up in -autorecon3. > > Greg > > > Bruce Fischl wrote: > >> Hi Greg, >> >> I'll let someone else answer the download question, as I don't offhand >> remember which site it is. >> >> As for your own atlas, I assume you mean cortical? This shouldn't be a >> problem. We distribute a binary called mris_sample_aparc that will sample >> them onto a surface. You can then use mris_ca_train to build your own >> atlas. >> >> As for the subcortical, this is a tougher question. Note that we are still >> working on it, and it has some sequence dependence. That said though, our >> manually labeled thalamus does extend to caudate, and it also goes pretty >> far lateral where there are some very low contrast thalamic nuclei. I'm not >> really qualified to comment on this - we'll need to get the neuroanatomists >> together. >> >> cheers, >> Bruce >> >> >> On Fri, 28 Oct 2005, Greg Harris wrote: >> >>> Hi Bruce. >>> >>> I need to watch for the source tarball on the website when it appears. >>> Is the correct url for the website something like >>> https://surfer.nmr.mgh.harvard.edu/fswiki/Download >>> i.e., the FreeSurfer Wiki? Or will I need to search for a tarball at >>> ftp://surfer.nmr.mgh.harvard.edu/pub/dist/ ? >>> >>> We will need instructions on how to define our own atlas, or how to >>> vary the existing atlas that is used to produce -autorecon3. We have >>> extensive traces of 50 or more brains to base this on. We want to >>> subdivide the STG-planum area. We also want the Insula, which I can >>> sort of get by clipping the 'unknown' ribbon with a probabilistic >>> talairach mask after mapping the 'aparc' information back into brains2. >>> >>> We have concerns about the 'aseg' information; for example, your >>> Thalamus-Proper is much larger than our gross thalamus, and our >>> thalamus roi does not touch our caudate roi. Do the control points >>> have to make the subcortical neighborhood pretty much of uniform >>> intensity of 110 for the white matter spine to develop into a sphere? >>> >>> Greg Harris >>> University of Iowa Psychiatry Brain Imaging Lab >>> >>> Bruce Fischl wrote: >>> >>>> We will be posting a "read-only" tarball on the website in the next >>>> couple of weeks, and anyone is free to download it and look at it. We >>>> will *not* be supporting a make/configure type env for people to download >>>> and build, as we simply aren't ready to do so. We plan to implement such >>>> a procedure, but it will be a while. Instead of waiting until that was >>>> ready, I thought people would like to have access to the source to at >>>> least see what things are doing, and be able to see file formats and >>>> such. >>> >>> >>>> cheers, >>>> Bruce >>>> >>>> >>>> _______________________________________________ >>>> Freesurfer mailing list >>>> Freesurfer(a)nmr.mgh.harvard.edu >>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >>>> >>> >>> >>> >> > > >

1 0

Re: [Freesurfer] good news and bad news
by Bruce Fischl 28 Oct '05

28 Oct '05

Hi Greg, I'll let someone else answer the download question, as I don't offhand remember which site it is. As for your own atlas, I assume you mean cortical? This shouldn't be a problem. We distribute a binary called mris_sample_aparc that will sample them onto a surface. You can then use mris_ca_train to build your own atlas. As for the subcortical, this is a tougher question. Note that we are still working on it, and it has some sequence dependence. That said though, our manually labeled thalamus does extend to caudate, and it also goes pretty far lateral where there are some very low contrast thalamic nuclei. I'm not really qualified to comment on this - we'll need to get the neuroanatomists together. cheers, Bruce On Fri, 28 Oct 2005, Greg Harris wrote: > Hi Bruce. > > I need to watch for the source tarball on the website when it appears. > Is the correct url for the website something like > https://surfer.nmr.mgh.harvard.edu/fswiki/Download > i.e., the FreeSurfer Wiki? Or will I need to search for a tarball at > ftp://surfer.nmr.mgh.harvard.edu/pub/dist/ ? > > We will need instructions on how to define our own atlas, or how to > vary the existing atlas that is used to produce -autorecon3. We have > extensive traces of 50 or more brains to base this on. We want to > subdivide the STG-planum area. We also want the Insula, which I can > sort of get by clipping the 'unknown' ribbon with a probabilistic > talairach mask after mapping the 'aparc' information back into brains2. > > We have concerns about the 'aseg' information; for example, your > Thalamus-Proper is much larger than our gross thalamus, and our > thalamus roi does not touch our caudate roi. Do the control points > have to make the subcortical neighborhood pretty much of uniform > intensity of 110 for the white matter spine to develop into a sphere? > > Greg Harris > University of Iowa Psychiatry Brain Imaging Lab > > Bruce Fischl wrote: > >> We will be posting a "read-only" tarball on the website in the next couple >> of weeks, and anyone is free to download it and look at it. We will *not* >> be supporting a make/configure type env for people to download and build, >> as we simply aren't ready to do so. We plan to implement such a procedure, >> but it will be a while. Instead of waiting until that was ready, I thought >> people would like to have access to the source to at least see what things >> are doing, and be able to see file formats and such. > >> cheers, >> Bruce >> >> >> _______________________________________________ >> Freesurfer mailing list >> Freesurfer(a)nmr.mgh.harvard.edu >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >> > > >

2 1

Intensity problems
by Martin Ystad 28 Oct '05

28 Oct '05

Hi, I'm currently running the latest developmental release for RedHat 9, but I seem to have some problems concerning the intesity-normalization. Unfortunately, my images have some intensity inhomogenities, but the first normalization-step seems to take care of most of these. However, some regions inside the white matter has a somewhat high value, for example 130 -140. These regions are not included in the white matter when I run the recon-all -segmentation program. Adding control-points to these areas does not work, obviously, since only values lower than 110 are considered when the normalization-step is re-runned. (right?) Is there any way to include these areas in my final white matter whithout manually drawing them onto the wm.mgz -volume? For instance raising the threshold for wm-segmentation? If so, would this produce any kind new artifacts which I need to be aware of? Thanks, Martin Ystad Medical Student University of Bergen Institute of Biomedicine Jonas Lies vei 91, 5009 Bergen, Norway.

1 0

Re: [Freesurfer] aseg
by Jenni Pacheco 28 Oct '05

28 Oct '05

You can run recon-all -segstats -subjid <subjid> and it will output a stats file into your subject's scripts/ directory. If you've already run recon-all you may already have this file. antao(a)itsa.ucsf.edu wrote: > Hi How to calculate the volume of segmented brain like hippocampus, > subcortical nucleus? > > Thanks, > > Antao _______________________________________________ Freesurfer mailing list Freesurfer(a)nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

3 2

aseg
by antao＠itsa.ucsf.edu 27 Oct '05

27 Oct '05

Hi How to calculate the volume of segmented brain like hippocampus, subcortical nucleus? Thanks, Antao

3 2

recalculating surfaces
by Dahlia Sharon 27 Oct '05

27 Oct '05

Hi, I overwrote several surface files (.inflated, .sulc, .curv, .smoothwm, .orig) by mistake while editing another subject. The volume in mri/wm/ is still the correct one. I should be able to run some of the recon steps again to get the correct surfaces back, but I don't want to do this without guidance so as not to make mistakes and start stepping over other files. Can you tell me which commands to run to regenerate these surfaces correctly? Thanks!!! Dahlia. PS - the subject is ~dahlia/structurals/subj21_bay2

2 5

recon-all library dependencies
by James Voyvodic 27 Oct '05

27 Oct '05

I've upgraded to the latest development release (freesurfer-Linux-rh9-dev20051003-full.tar.gz) running on Linux Redhat 9 on an x86 platform. Recon-all crashes, apparently because of failed library dependencies, with the following message: /usr/local/mni/bin/volume_stats: /usr/lib/libstdc++.so.6: version `CXXABI_1.3' not found (required by /usr/local/mni/bin/volume_stats) /usr/local/mni/bin/volume_stats: /usr/lib/libstdc++.so.6: version `GLIBCXX_3.4' not found (required by /usr/local/mni/bin/volume_stats) nu_estimate_np_and_em: crashed while running volume_stats (termination status=256) nu_correct: crashed while running nu_estimate_np_and_em (termination status=256) ERROR: nu_correct Linux hawking 2.4.20-31.9 #1 Tue Apr 13 18:04:23 EDT 2004 i686 i686 i386 GNU/Linux recon-all exited with errors at Wed Oct 26 15:17:17 EDT 2005 It had previously crashed because it couldn't find libstdc++.so.6 at all, so I tried to appease it by using the old trick of linking libstdc++.so.5 to libstdc++.so.6, but obviously that wasn't enough. So the main question is: what libraries are needed to run the new version(s) of FreeSurfer? Is there a way to download the missing libraries from somewhere? When I installed and tried to run the most recent stable release (freesurfer-Linux-rh9-v2.1.tar.gz) I was foiled because I didn't have libtixwish installed, which is why I went to the development release that had more libraries. Yum says I have libstdc++ and doesn't offer a .so.6 variant. Do I need a different version of Linux? A side question: If recon-all crashes part way through, is there a way to skip the successful steps when restarting or does it have to redo all the steps in its autorecon (e.g. autorecon1) script? thanks, Jim ======================== James T. Voyvodic, Ph.D. Brain Imaging and Analysis Center Duke University Medical Center e-mail: jim.voyvodic(a)duke.edu

3 3

converting thickess file to ascii
by Thomas, Adam (NIH/NIMH) 26 Oct '05

26 Oct '05

Can mris_convert be used to convert thickness files to ascii? My attempts so far have resulted in segfaults: [adamt@antimony surf]$ mris_convert lh.thickness lh.thickness.asc Segmentation fault [adamt@antimony surf]$ Or is there another way to convert thickness files to ascii? Thanks, -Adam --- Adam Thomas adamt(a)nih.gov Functional MRI Facility, NIMH/NIH/DHHS 10 Center Dr, Room B1D708A Bethesda MD. 20892-1148

3 2

re: good news and bad news (Bruce Fischl)
by Glenn Lawyer 26 Oct '05

26 Oct '05

Hi, Regarding Bruce's "GOOD NEWS" announcement from last friday: I wanted to thank Bruce, Doug, Nick, and everyone else on the dev. team for making FreeSurfer open source (also of course the folks Bruce mentioned: Anders, Howard, and Christian). I suspect that many of the people on this mailing list are not deeply into software issues, so perhaps you may not realize just how important opening up the source code is. I do know, however, that we are all committed to science. Science progresses much faster when ideas, results, and methods are shared in an open and unrestricted manner. So thanks again to the FreeSurfer team for sharing your hard-won knowledge with the rest of the world. +glenn -- I want to go farther, farther! A. E. L. > Glenn Lawyer < > +352 061 967 244 < > Instituttgruppe for psykiatri < > Postboks 1130 Blindern < > 0318 Oslo < < http://folk.uio.no/davidgl > <><><><><><><><><><><><><><><><><:)

1 0

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Freesurfer