Freesurfer December 2006

freesurfer@nmr.mgh.harvard.edu

49 participants
59 discussions

Freesurfer and Grid computing
by Andries van der Leij 31 Jan '07

31 Jan '07

________________________________ From: Andries van der Leij Sent: Wednesday, December 06, 2006 5:59 PM To: 'freesurfer(a)nmr.mgh.harvard.edu' Subject: Freesurfer and Grid computing Dear Freesurfer community, I'm a PHD student at the university of Amsterdam and I'm currently investigating the possibilities to streamline our MRI data processing stream. Next summer we'll obtain a research-only scanner. I'm trying to push the group to also invest in computing power and am currently investigating the applications that researchers will most probably use. I came across a project of the group of Cohen at UCLA. They have configured a Apple (unix) grid and have proposed a more or less standard setup specially designed for MRI analyses: http://airto.bmap.ucla.edu/mt-static/NICluster/archives/2005/06/welcome. html It is my understanding that one of the members has rewritten the FSL code which allow distributed parallel processing in a Grid. See the benchmarks here: http://airto.bmap.ucla.edu/bmcweb/bmc_bios/MarkCohen/Apple/Benchmarks.ht m My question is fairly simple: Are similar steps taken in the Freesurfer community? I have no experience with this app myself, but it is my understanding that Freesurfer consumes a lot of resources. Thank you very much in advance, Andries van der Leij

7 12

Cortical thickness group analyses
by Dziobek, Isabel 04 Jan '07

04 Jan '07

Dear All, I have some questions in regard to the FreeSurfer group analyses (I am using the 3.0 Version), for which I would appreciate your help. I am trying to compare a patient to a control group with respect to cortical thickness (regressing out variables such as age and IQ) and also to see if associations exist in specific ROIs between cortical thickness and cognitive variables (for the groups separately). I started constructing a group descriptor file containing the two classes (patient and control) and the variable age. First, I wanted to compare the groups disregarding any other variable and so I constructed the following contrast vector: (1 -1 0), assuming doss and thinking that the ³0² meant that this column of the design matrix would not be taken into account in the model. However, this yielded the same result as running (1 -1 1). Do I have to construct a separate fsgd file only containing the classes and then a contrast vector (1 -1)? What about regressing out more than 1 variable (e.g. age and IQ), how does the contrast vector have to look like then? Also, how do I look at correlations between those variables and cortical thickness for the groups separately? When loading the scatter plots of the fsgd, the regression line for the separate groups seems to run always parallel, which does not seem very likely (see attached jpg). Are these regression lines at all? Where can I obtain the Rs in numbers? Can I, alternatively, stream out the numbers for the cortical thickness of a defined ROI to then run correlations in SPSS. Where would I find those numbers? I was also wondering about corrections for intracranial vault. Does the registration to the group¹s average brain takes care of differences in overall head size or is it advisable to put icv as a separate regressor (as is done in very few studies)? Thanks a lot for your help and happy holidays, Isabel ************************************** Isabel Dziobek, PhD Postdoctoral Research Fellow Independent Junior Research Group Neurocognition of Decision-Making Max-Planck Institute for Human Development Lentzeallee 94 14195 Berlin Phone: +49 30 82406 619 Fax: +49 30 82406 616 dziobek(a)mpib-berlin.mpg.de

2 1

mri_surf2surf resampling error
by Julien 04 Jan '07

04 Jan '07

Hello there, I'm trying to resample some thickness difference maps using mri_surf2surf. Unfortunately, I keep getting the following error: ------------------------------------------------------------------- [engram:Multispect] (nmr-std-env) mri_surf2surf --srcsubject 002_MEF_A \ --sval $SUBJECTS_DIR/thickness/002_MEF_A-002_MPRAGE_A/surf/lh.thickness_diff.w \ --src_type curv \ --trgsubject 001_MEF_A \ --tval $SUBJECTS_DIR/thickness/002_MEF_A-002_MPRAGE_A/surf/lh.thickness_diff_resampled_to_001_MEF_A.w \ --trg_type paint \ --hemi lh \ --surfreg sphere.reg Registration surface changed to sphere.reg srcsubject = 002_MEF_A srcval = /space/engram/1/users/Multispect/thickness/002_MEF_A-002_MPRAGE_A/surf/lh.thickness_diff.w srctype = curv trgsubject = 001_MEF_A trgval = /space/engram/1/users/Multispect/thickness/002_MEF_A-002_MPRAGE_A/surf/lh.thickness_diff_resampled_to_001_MEF_A.w trgtype = paint surfreg = sphere.reg srchemi = lh trghemi = lh frame = 0 fwhm-in = 0 fwhm-out = 0 Reading source surface reg /space/engram/1/users/Multispect/002_MEF_A/surf/lh.sphere.reg Loading source data Reading curvature file /space/engram/1/users/Multispect/thickness/002_MEF_A-002_MPRAGE_A/surf/lh.thickness_diff.ÔC MRISreadCurvature: could not open /space/engram/1/users/Multispect/thickness/002_MEF_A-002_MPRAGE_A/surf/lh.thickness_diff.ÔC No such file or directory ERROR: reading curvature file ------------------------------------------------------------------- From the initial command output, it appears as though the paths are input correctly, but upon reading the file it appears to be trying to open "lh.thickness_diff.ÔC" instead of "lh.thickness_diff.w" and subsequently failing. Am I doing anything incorrectly here? Thanks, Julien -- Julien Wonderlick Research Assistant, Behavioral Neuroscience Laboratory Department of Brain and Cognitive Sciences, MIT Building 46-5127 617.253.7437 jswonder(a)mit.edu

7 13

talairach questions
by Dominick, Kelli, Ray Carpenter 02 Jan '07

02 Jan '07

Hi All, We have had a number of brains with very bad talairach transforms however aparc+aseg looks fine. In addition, while the talairach.xfm transform is very bad talairach.Ita is not. Is the bad xfm transform still a problem even though it is not affecting the labels. And if so which transform should I attempt to correct in tkregister2 in order to fix the problem? (I ask because it looks like that some of the tutorials still refer to old file names that have changed in the stable versions). I am using v3.03. Thanks in advance for your help! Kelli

2 3

mris_preproc
by Lars M. Rimol 31 Dec '06

31 Dec '06

Hi, When I try to run mris_preproc wiht the nocleanup flag I get an error message: flag --nocleanup not recognized Same thing happens for --cleanup (i.e., not recognized), and "--tmpdir tmpdir" doesn't seem to work (it still does cleanup). But all these flags are listed under the help function for mris_preproc: --tmpdir dir : use tmpdir. Implies --nocleanup. --nocleanup : do not delete tmpdir --cleanup : delete tmpdir (default) I'm using /RH4-x86_64-R302. Thank you! -- yours, LMR

1 0

once more: Cortical thickness group analyses
by Dziobek, Isabel 29 Dec '06

29 Dec '06

Hi, I posted the message below about a week ago but didn't get a reply. Any help is much appreciated! Thanks and best, Isabel -----Original Message----- From: Dziobek, Isabel Sent: Wednesday, December 20, 2006 11:38 AM To: Freesurfer(a)nmr.mgh.harvard.edu Subject: Cortical thickness group analyses Dear All, I have some questions in regard to the FreeSurfer group analyses (I am using the 3.0 Version), for which I would appreciate your help. I am trying to compare a patient to a control group with respect to cortical thickness (regressing out variables such as age and IQ) and also to see if associations exist in specific ROIs between cortical thickness and cognitive variables (for the groups separately). I started constructing a group descriptor file containing the two classes (patient and control) and the variable age. First, I wanted to compare the groups disregarding any other variable and so I constructed the following contrast vector: (1 -1 0), assuming doss and thinking that the ³0² meant that this column of the design matrix would not be taken into account in the model. However, this yielded the same result as running (1 -1 1). Do I have to construct a separate fsgd file only containing the classes and then a contrast vector (1 -1)? What about regressing out more than 1 variable (e.g. age and IQ), how does the contrast vector have to look like then? Also, how do I look at correlations between those variables and cortical thickness for the groups separately? When loading the scatter plots of the fsgd, the regression line for the separate groups seems to run always parallel, which does not seem very likely (see attached jpg). Are these regression lines at all? Where can I obtain the Rs in numbers? Can I, alternatively, stream out the numbers for the cortical thickness of a defined ROI to then run correlations in SPSS. Where would I find those numbers? I was also wondering about corrections for intracranial vault. Does the registration to the group¹s average brain takes care of differences in overall head size or is it advisable to put icv as a separate regressor (as is done in very few studies)? Thanks a lot for your help and happy holidays, Isabel ************************************** Isabel Dziobek, PhD Postdoctoral Research Fellow Independent Junior Research Group Neurocognition of Decision-Making Max-Planck Institute for Human Development Lentzeallee 94 14195 Berlin Phone: +49 30 82406 619 Fax: +49 30 82406 616 dziobek(a)mpib-berlin.mpg.de

1 0

mri_convert does not recognize dicom files
by Julia Hamstra 22 Dec '06

22 Dec '06

Hello, I am unable to use mri_convert to import dicom files. Below are examples of the errors. For a GE scan: mri_convert -it ge -ot mgz -i 01.v2 -o orig.mgz reading from 01.v2... genesisRead(): can't determine file name format for ./01.v2 For a Siemens scan: mri_convert -it siemens_dicom -ot mgz -i 0001.dcm -o orig.mgz reading from 0001.dcm... ERROR (ScanSiemensSeries): ./0001.dcm is not a Siemens DICOM file ERROR: could not find any files (SeriesList==NULL) Similar errors are given when the generic dicom is used. Other programs, such as Osirix and Analyze are able to process the DICOM files. Is there an obvious error in the command? What else should I consider? Thank you ~julia -- UC Davis Neuroscience Graduate Group The M.I.N.D. Institute - Amaral Lab (916) 703-0376

2 1

aseg, aparc regions reference table (aka "data dictionary")?
by Graham Wideman 22 Dec '06

22 Dec '06

Folks: It would be *really* great if there was a page somewhere (or just a file) giving some info on the various measures that are captured in the aseg and aparc files. For each measure (in the table or the "comments" above the table): -- Description of what it includes -- some are obvious from the name, others not. -- Whether it's deprecated. -- Whether FS V-whatever actually produces this value, or just zero. (Very useful, and probably not too much work? Plus I'd like to adapt this info so as to pull only the non-inert rows from the data file). (Or is there some other way to determine this?) -- Some suggestions on what numbers need to be added to get popular subtotals. I've just read through the last six months of freesurfer list traffic, and I see numerous questions along these lines... and replies with some clues here and there. Also references to papers, but these don't detail the nitty gritty of what FS V-something actually does. Probably numerous users would get a lot of use out of having this info in one place. Thanks, Graham

2 1

maybe a problem of tcl
by lv bin 21 Dec '06

21 Dec '06

hi, I am a newcomer of freesurfer.Sereral days ago,I installed the freesurfer(its version is freesurfer-Linux-centos4_x86_64-stable-pub-v3.0.4),and practiced the data "bert"with tkmedit,tksurfer,it works ok. These days ,I meet a problem about tksurfer.At first,I run the command %tksurfer bert lh white it's good to show the surface ,and then when I try to load the Curvature " lh.curv",the tksurfer interface is closed suddenly,and the wrong information that appears on the termial is that %alloc:invalid block: 0xb1ea60: 0 0 75 Abort I googled this with"alloc:invalid block",maybe it is a problem of "tcl",but I don't know how to slove it,any suggestion would be appreciated! Best wishes! lvbin

2 3

GLM - Multiple Comparisons
by David Zai 21 Dec '06

21 Dec '06

In running a glm analysis with multiple classes (A, B, C, D, E, F) and no variables, is it better to: (1) have one fsgd with a contrast vector of (1 -1 0 0 0 0) to compare Group A thickness to Group B thickness, or (2) should I make an fsgd with just my Group A and B subjects and have my contrast vector (1 -1)? We ran both options to compare Group A and B as an example and the results of each are extremely similar but there are some very minor differences in the significance maps. We're wondering about the two methods because we eventually want to compare every two-group pairing (A vs C, A vs D, E vs F, etc.), and was wondering whether one method is more accurate or preferred over the other when we try multiple group comparisons. Also, could there be a randomizing effect that causes the slightly different maps for our example? Thanks. David

1 0

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Freesurfer December 2006