Re: [Freesurfer] make_average_subject with large cohorts
I am a little out of the loop. Just to clarify, Marie is trying to create an average subject out of 400+ subjects that have been run through the freesurfer pipeline? I don't see a problem with that (except possibly memory issues). If the subjects are from a homogeneous population (whatever homogeneous means), I would even say the more subjects the better. --Thomas On Jun 7, 2010, at 2:43 PM, Nick Schmansky wrote: > Marie, > > I fixed the memory leak that was causing your out-of-memory error when > running mris_make_average_surface. A new mris_make_average_surface > for > Mac Intel Leopard is posted here: > > ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/misc/macos-leopard-intel/ > > I'm still looking into the mri_average failure. I dont know what to > say > about the license error, as the code that handles the license hasnt > changed since v2.0. > > As an aside, I'm skeptical of the utility of a volume or surface > composed of 400+ subjects. The volume will be too blurry to be of > use, > and the surface will have lost any group-specific patterns that might > have existed. If retaining the unique morphometric properties of your > subject set is important, you might want to consider using just 20 > or so > randomly chosen subjects. Perhaps Bruce or others can comment. > > Nick > > > On Wed, 2010-06-02 at 18:53 +0200, Marie Schaer wrote: >> Hi Nick, >> >> Thank you for your response. I added the flag --debug, please find in >> attachment the log. I am working on a Mac Intel Quad-Core Intel Xeon, >> Processor 3 GHz, RAM 5GB. I also tried with another similar machine >> which 9GB, but it exited with the same error exactly at the same >> point. Even using the -no-aseg option it stops exactly at the same >> point. I also provide you a screenshot with the top command >> running, a >> few minutes before the error, and just after the error. >> >> As for the license error, I just had this problem using the >> Freesurfer >> version 4.5, and you are right, it may be a connection problem >> because >> our data are on a server (mounted with afp). It is however strange >> that it exits always at exactly the same point... Anyway, no license >> problem with version 4.4 or 4.3. >> >> Many thanks for your help, >> >> Marie >> >> On 1 juin 10, at 16:54, Nick Schmansky wrote: >> >>> can you try adding the flag --debug to make_average_volume? it will >>> output a lot of stuff, but will pinpoint at which stage it is >>> running >>> out of memory. also, try running 'top' in another terminal to >>> monitor >>> the amount of memory it's consuming while running. >>> >>> as for the license error, that is bizarre, i can't think of a reason >>> for >>> that, unless your subjects are located on an NFS drive (a network). >>> we've seen lots of problems with NFS and the Mac OS, where the >>> connection drops for a couple seconds. >>> >>> n. >>> >>> >>> On Mon, 2010-05-31 at 22:39 +0200, Marie Schaer wrote: Hi everyone, I did not find a solution to my problem yet: how to launch make_average_subject / make_average_volume / make_average_surface on a large number of subjects on a Mac Intel? Did someone encounter this error before? Many thanks in advance for your help, Marie On 29 mai 10, at 20:07, Marie Schaer wrote: > > Hi Nick, > > Sorry for disturbing again. As advised, I tried to launch > separately > make_average_volume and make_average_surface. Even with the --no- > aseg > option with make_average_volume I got the "cannot allocate memory" > error. And with the make_average_surface I got the "license error" > in > the middle of one subject processing... Any idea? > > Thank you very much, > > Marie > > > On 28 mai 10, at 21:37, Nick Schmansky wrote: > >> try adding --no-aseg to the command line. that out-of-memory >> error >> seems to originate with creation of the average aseg.mgz file >> (hopefully >> you wont need it for your subject set, typically you dont need >> aseg.mgz >> since aseg stats are usually gathered from a set of aseg.stats >> files, >> not from an average aseg.mgz). >> >> doug: maybe we should make --no-aseg the default? >> >> n. >> >> >> On Fri, 2010-05-28 at 21:31 +0200, Marie Schaer wrote: >>> Hello everybody, >>> >>> I have a problem running make_average_subject for a large number >>> of >>> subjects (>400) on a Mac Intel. I tried with different >>> Freesurfer >>> versions (all the log outputs are attached). Using the >>> freesurfer >>> version 4.3 and 4.4 the make_average_subject script always >>> stop at >>> the >>> 297th subject with the error "Cannot allocate memory". I already >>> tried >>> to uncomment the line "limit descriptors unlimited" or to set it >>> to >>> various values, as prop
Re: [Freesurfer] make_average_subject with large cohorts
Marie, The memory leak in mri_average has also been fixed. The new version is in the same ftp location as mris_make_average_surface. N. On Mon, 2010-06-07 at 14:43 -0400, Nick Schmansky wrote: > Marie, > > I fixed the memory leak that was causing your out-of-memory error when > running mris_make_average_surface. A new mris_make_average_surface for > Mac Intel Leopard is posted here: > > ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/misc/macos-leopard-intel/ > > I'm still looking into the mri_average failure. I dont know what to say > about the license error, as the code that handles the license hasnt > changed since v2.0. > > As an aside, I'm skeptical of the utility of a volume or surface > composed of 400+ subjects. The volume will be too blurry to be of use, > and the surface will have lost any group-specific patterns that might > have existed. If retaining the unique morphometric properties of your > subject set is important, you might want to consider using just 20 or so > randomly chosen subjects. Perhaps Bruce or others can comment. > > Nick > > > On Wed, 2010-06-02 at 18:53 +0200, Marie Schaer wrote: > > Hi Nick, > > > > Thank you for your response. I added the flag --debug, please find in > > attachment the log. I am working on a Mac Intel Quad-Core Intel Xeon, > > Processor 3 GHz, RAM 5GB. I also tried with another similar machine > > which 9GB, but it exited with the same error exactly at the same > > point. Even using the -no-aseg option it stops exactly at the same > > point. I also provide you a screenshot with the top command running, a > > few minutes before the error, and just after the error. > > > > As for the license error, I just had this problem using the Freesurfer > > version 4.5, and you are right, it may be a connection problem because > > our data are on a server (mounted with afp). It is however strange > > that it exits always at exactly the same point... Anyway, no license > > problem with version 4.4 or 4.3. > > > > Many thanks for your help, > > > > Marie > > > > On 1 juin 10, at 16:54, Nick Schmansky wrote: > > > > > can you try adding the flag --debug to make_average_volume? it will > > > output a lot of stuff, but will pinpoint at which stage it is running > > > out of memory. also, try running 'top' in another terminal to monitor > > > the amount of memory it's consuming while running. > > > > > > as for the license error, that is bizarre, i can't think of a reason > > > for > > > that, unless your subjects are located on an NFS drive (a network). > > > we've seen lots of problems with NFS and the Mac OS, where the > > > connection drops for a couple seconds. > > > > > > n. > > > > > > > > > On Mon, 2010-05-31 at 22:39 +0200, Marie Schaer wrote: > > >> Hi everyone, > > >> > > >> I did not find a solution to my problem yet: how to launch > > >> make_average_subject / make_average_volume / make_average_surface > > >> on a > > >> large number of subjects on a Mac Intel? Did someone encounter this > > >> error before? > > >> > > >> Many thanks in advance for your help, > > >> > > >> Marie > > >> > > >> On 29 mai 10, at 20:07, Marie Schaer wrote: > > >> > > >>> > > >>> Hi Nick, > > >>> > > >>> Sorry for disturbing again. As advised, I tried to launch separately > > >>> make_average_volume and make_average_surface. Even with the --no- > > >>> aseg > > >>> option with make_average_volume I got the "cannot allocate memory" > > >>> error. And with the make_average_surface I got the "license error" > > >>> in > > >>> the middle of one subject processing... Any idea? > > >>> > > >>> Thank you very much, > > >>> > > >>> Marie > > >>> > > >>> > > >>> On 28 mai 10, at 21:37, Nick Schmansky wrote: > > >>> > > try adding --no-aseg to the command line. that out-of-memory error > > seems to originate with creation of the average aseg.mgz file > > (hopefully > > you wont need it for your subject set, typically you dont need > > aseg.mgz > > since aseg stats are usually gathered from a set of aseg.stats > > files, > > not from an average aseg.mgz). > > > > doug: maybe we should make --no-aseg the default? > > > > n. > > > > > > On Fri, 2010-05-28 at 21:31 +0200, Marie Schaer wrote: > > > Hello everybody, > > > > > > I have a problem running make_average_subject for a large number > > > of > > > subjects (>400) on a Mac Intel. I tried with different Freesurfer > > > versions (all the log outputs are attached). Using the freesurfer > > > version 4.3 and 4.4 the make_average_subject script always stop at > > > the > > > 297th subject with the error "Cannot allocate memory". I already > > > tried > > > to uncomment the line "limit descriptors unlimited" or to set it > > > to > > > various values, as proposed in the mailing list, but without > > > success > > > (it always stop at the 297th subject)
Re: [Freesurfer] make_average_subject with large cohorts
Marie, I fixed the memory leak that was causing your out-of-memory error when running mris_make_average_surface. A new mris_make_average_surface for Mac Intel Leopard is posted here: ftp://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/misc/macos-leopard-intel/ I'm still looking into the mri_average failure. I dont know what to say about the license error, as the code that handles the license hasnt changed since v2.0. As an aside, I'm skeptical of the utility of a volume or surface composed of 400+ subjects. The volume will be too blurry to be of use, and the surface will have lost any group-specific patterns that might have existed. If retaining the unique morphometric properties of your subject set is important, you might want to consider using just 20 or so randomly chosen subjects. Perhaps Bruce or others can comment. Nick On Wed, 2010-06-02 at 18:53 +0200, Marie Schaer wrote: > Hi Nick, > > Thank you for your response. I added the flag --debug, please find in > attachment the log. I am working on a Mac Intel Quad-Core Intel Xeon, > Processor 3 GHz, RAM 5GB. I also tried with another similar machine > which 9GB, but it exited with the same error exactly at the same > point. Even using the -no-aseg option it stops exactly at the same > point. I also provide you a screenshot with the top command running, a > few minutes before the error, and just after the error. > > As for the license error, I just had this problem using the Freesurfer > version 4.5, and you are right, it may be a connection problem because > our data are on a server (mounted with afp). It is however strange > that it exits always at exactly the same point... Anyway, no license > problem with version 4.4 or 4.3. > > Many thanks for your help, > > Marie > > On 1 juin 10, at 16:54, Nick Schmansky wrote: > > > can you try adding the flag --debug to make_average_volume? it will > > output a lot of stuff, but will pinpoint at which stage it is running > > out of memory. also, try running 'top' in another terminal to monitor > > the amount of memory it's consuming while running. > > > > as for the license error, that is bizarre, i can't think of a reason > > for > > that, unless your subjects are located on an NFS drive (a network). > > we've seen lots of problems with NFS and the Mac OS, where the > > connection drops for a couple seconds. > > > > n. > > > > > > On Mon, 2010-05-31 at 22:39 +0200, Marie Schaer wrote: > >> Hi everyone, > >> > >> I did not find a solution to my problem yet: how to launch > >> make_average_subject / make_average_volume / make_average_surface > >> on a > >> large number of subjects on a Mac Intel? Did someone encounter this > >> error before? > >> > >> Many thanks in advance for your help, > >> > >> Marie > >> > >> On 29 mai 10, at 20:07, Marie Schaer wrote: > >> > >>> > >>> Hi Nick, > >>> > >>> Sorry for disturbing again. As advised, I tried to launch separately > >>> make_average_volume and make_average_surface. Even with the --no- > >>> aseg > >>> option with make_average_volume I got the "cannot allocate memory" > >>> error. And with the make_average_surface I got the "license error" > >>> in > >>> the middle of one subject processing... Any idea? > >>> > >>> Thank you very much, > >>> > >>> Marie > >>> > >>> > >>> On 28 mai 10, at 21:37, Nick Schmansky wrote: > >>> > try adding --no-aseg to the command line. that out-of-memory error > seems to originate with creation of the average aseg.mgz file > (hopefully > you wont need it for your subject set, typically you dont need > aseg.mgz > since aseg stats are usually gathered from a set of aseg.stats > files, > not from an average aseg.mgz). > > doug: maybe we should make --no-aseg the default? > > n. > > > On Fri, 2010-05-28 at 21:31 +0200, Marie Schaer wrote: > > Hello everybody, > > > > I have a problem running make_average_subject for a large number > > of > > subjects (>400) on a Mac Intel. I tried with different Freesurfer > > versions (all the log outputs are attached). Using the freesurfer > > version 4.3 and 4.4 the make_average_subject script always stop at > > the > > 297th subject with the error "Cannot allocate memory". I already > > tried > > to uncomment the line "limit descriptors unlimited" or to set it > > to > > various values, as proposed in the mailing list, but without > > success > > (it always stop at the 297th subject). Then I tried with the > > Freesurfer version 4.5 (therefore I had to recompute the > > lh.inflated.H > > and lh.inflated.K which are not automatically produced in previous > > versions), but with the 4.5 version I got a strange "license > > error" > > message in the middle of the script (and my licence is certainly > > fine > > as I can run any other script without problems). > > > > Does anyon
[Freesurfer] tksurfer issue
I am having a problem using tksurfer. It does not appear to be the same issue that others have had in the past with NVIDIA drivers etc. When I try to load a surface file the entire process freezes and eventually if I let it run, I will get the following error: gkl...@gklein-nspc:~$ tksurfer CL lh full.flat.patch.3d surfer: current subjects dir: /home/gklein/fssubjects surfer: not in "scripts" dir ==> using cwd for session root surfer: session root data dir ($session) set to: surfer: /home/gklein surfer: Reading header info from /home/gklein/fssubjects/CL/mri/T1.mgz /home/gklein/fssubjects/CL/surf/lh.full.flat.patch.3d: face[0].v[0] = 0, but face 0 not in vertex 0 face list Cannot allocate memory As I am new to FS, any help would be appreciated. Thanks. -Gad ___ 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] Label files for the Cortical Ribbon
Hello freesurfers! After creating some labeling files from aparc.a2005s.annot, I noticed that my sulci and gyri volumes generated from the .label files (i.e., after running mri_annotation2label and mri_label2vol) were only representing the outer layer of the white surface. Is there any way to generate .label files that include the entire cortical ribbon? Thanks, any help would be appreciated. Tina Jeon ___ 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] Label files for the Cortical Ribbon
Hello all! After creating some labeling files from aparc.a2005s.annot, I noticed that my sulci and gyri volumes generated from the .label files (i.e., after running mri_annotation2label and mri_label2vol) were only representing the outer layer of the white surface. Is there any way to generate .label files that include the entire cortical ribbon? Thanks, any help would be appreciated. Tina ___ 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] standard deviation map
Dear FS experts, I have a anatomical scan of a subject with a rare neurogenetical disorder. I'd like to compare this with a age matched control group. How can I visualize a z-score transform map of the cortical thickness, i.e. how would I compute and vizualize the number/amount of standard deviations that my subjects differs from the mean of the control group at each vertex? Thanks, Michael ___ 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] cannon open .mgh file using MRIread
When I load the file directly to load_mgh I still get the error
ERROR: could not open
/home/freesurfer4.4.0/subjects/corinna/ADNI/003_S_1057/mri/lh.white.projfrac35.mgh
for reading
??? Attempt to reference field of non-structure array.
Error in ==> gwratiotest at 5
gray2=gray.vol;
here is my matlab file:
function gwratiotest
gray=load_mgh('/home/freesurfer4.4.0/subjects/corinna/ADNI/003_S_1057/mri/lh.white.projfrac35.mgh');
white=load_mgh('/home/freesurfer4.4.0/subjects/corinna/ADNI/003_S_1057/mri/lh.white.projdist.neg1.mgh');
gray2=gray.vol;
white2=white.vol;
gwratio.vol = gray2 / white2;
MRIwrite(gwratio,
'/home/freesurfer4.4.0/subjects/corinna/ADNI/003_S_1057/mri/lh.gwratio.mgh');
end
On Fri, Jun 4, 2010 at 12:20 PM, Bruce Fischl wrote:
> Hi Corinna,
>
> sorry, I'm not sure what's going wrong. Can you try loading that volume
> calling load_mgh directly? Make sure the filename is exactly right. It's
> hard to tell with all the carriage returns in your email.
>
> cheers
> Bruce
>
>
>
> On Fri, 4 Jun 2010, corinna bauer wrote:
>
> When I run mri_info I get:
>>
>> Volume information for
>>
>> /home/freesurfer4.4.0/subjects/corinna/ADNI/003_S_1057/mri/lh.white.projfrac.35.mgh
>> type: MGH
>> dimensions: 23334 x 1 x 5
>> voxel sizes: 1., 1., 1.
>> type: FLOAT (3)
>> fov: 23334.000
>> dof: 0
>> xstart: -11667.0, xend: 11667.0
>> ystart: -0.5, yend: 0.5
>> zstart: -2.5, zend: 2.5
>> TR: 3000.00 msec, TE: 3.61 msec, TI: 1000.00 msec, flip angle:
>> 8.00 degrees
>> nframes: 1
>> ras xform present
>> xform info: x_r = -1., y_r = 0., z_r = 0., c_r =
>> 5.5017
>> : x_a = 0., y_a = 0., z_a = 1., c_a =
>> 11.5254
>> : x_s = 0., y_s = -1., z_s = 0., c_s =
>> 0.
>>
>> talairach xfm :
>>
>> /home/freesurfer4.4.0/subjects/corinna/ADNI/003_S_1057/mri/transforms/talairach.xfm
>> Orientation : LIA
>> Primary Slice Direction: coronal
>>
>> voxel to ras transform:
>> -1. 0. 0. 11672.5020
>> 0. 0. 1. 9.0254
>> 0. -1. 0. 0.5000
>> 0. 0. 0. 1.
>>
>> voxel-to-ras determinant -1
>>
>> ras to voxel transform:
>> -1. 0. 0. 11672.5020
>> -0. -0. -1. 0.5000
>> -0. 1. -0.-9.0254
>> 0. 0. 0. 1.
>>
>>
>>
>> On Fri, Jun 4, 2010 at 9:58 AM, Bruce Fischl > >wrote:
>>
>> if you run mri_info on the .mgh file does it return reasonable looking
>>> values?
>>>
>>> On Fri, 4 Jun 2010, corinna bauer wrote:
>>>
>>> Hi Bruce,
>>>
I am having trouble getting my .mgh files to open in MRIread. I get the
following error message.
ERROR: could not open /home/freesurfer4.4.0/
subjects/corinna/ADNI/003_S_1057/mri/lh.white.projfrac35.mgh for reading
ERROR: loading
/home/freesurfer4.4.0/subjects/corinna/ADNI/003_S_1057/mri/lh.white.projfrac35.mgh
as MGH
??? Attempt to reference field of non-structure array.
Error in ==> gwRatio at 24
gwratio.vol = gray.vol / white.vol;
here's my .m file
function gwRatio
rootdir = '/home/freesurfer4.4.0/subjects/corinna/ADNI';
paths=dir([rootdir,'/*']);
numel(paths)
num = numel(paths);
for ii = 1:num
if(paths(ii).isdir && ~strcmp(paths(ii).name,'.') &&
~strcmp(paths(ii).name,'..'))
graypath=
fullfile(rootdir,paths(ii).name,'mri/lh.white.projfrac35.mgh');
whitepath =
fullfile(rootdir,paths(ii).name,'mri/lh.white.projdist.neg1.mgh');
outpath = fullfile(rootdir,paths(ii).name,'mri/lh.gwratio.mgh');
fprintf('- Processing %i of %i -\n', ii, num);
fprintf('gray: %s\n', graypath);
fprintf('white: %s\n', whitepath);
fprintf('out: %s\n', outpath);
gray = MRIread(graypath);
white = MRIread(whitepath);
gwratio = gray;
gwratio.vol = gray.vol / white.vol;
MRIwrite(gwratio, outpath);
end
end
corinna
>>> 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/fr
