Thank you so much for the extense answer!
It happens that the person who analysed the first measure (of this longitudinal dataset) used two different OS (he did it in a Virtual machine on windows and a mac OS X) to speed up this step. That data has already been used in some studies.
¿So my question now would be how much does that really affect to the results?
Hi
I am using recon-all in some longitudinal data on a ubuntu 16.04, and using the development version of freesurfer as I was told here, because the other current version doesnt work on ubuntu 16.04.
The problem is that the first scans were probablly processed using virtualbox in a windows and an older version of freesurfer.
i think I already know the answer, that I can't do this if I dont want to bias my results. Can I?
Yup. You know the answer ;) However, I did discover a way to make freesurfer 5.3 work on Ubuntu 16 by doing the following steps to install an older version of perl :
1) install perlbrew:
$> sudo apt-get install perlbrew
2) use perlbrew to install perl version 5.20.3:
$> perlbrew install perl-5.20.3
3) You must change the '/bin/perl' file to a link which points to the version of perl installed by perlbrew. (This technique is not the standard practice but it is the only way to get the mni tools to use the older version of perl):
$> cd /bin $> sudo mv perl perl.orig (can always set back later) $> ln -s ~/perl5/perlbrew/perls/perl-5.20.3/bin/perl
But I'm using a better computer now and I use the one with windows for other analyses, so I wanted to know if I could install Virtual box on my ubuntu 16.04 computer, and install the same version of linux used to process the first measure. Should I also need to use the same version of freesurfer?
You could do that, but The Virtual Box version is not really good for processing a large number of subjects because of the limited amount of disk space on the virtual image. Its really meant more as a teaching tool for users with Windows machines.
f the computer's specs are dfferent, does it matter? And if I assign different cpu and ram on my virtual machine?
Different specs are fine. Its the OS, OS version, and version of freesurfer that should stay the same.
Thank you! _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
On 11/15/2016 07:21 AM, anonymous anonymous wrote:
Thank you so much for the extense answer!
It happens that the person who analysed the first measure (of this longitudinal dataset) used two different OS (he did it in a Virtual machine on windows and a mac OS X) to speed up this step. That data has already been used in some studies.
¿So my question now would be how much does that really affect to the results?
It is impossible to know for sure without processing the results on the same OS. My guess is quite small, generally we see results differ on the order of 5% or less, but still potentially enough to introduce bias.
Hi
I am using recon-all in some longitudinal data on a ubuntu 16.04, and using the development version of freesurfer as I was told here, because the other current version doesnt work on ubuntu 16.04.
The problem is that the first scans were probablly processed using virtualbox in a windows and an older version of freesurfer.
i think I already know the answer, that I can't do this if I dont want to bias my results. Can I?
Yup. You know the answer ;) However, I did discover a way to make freesurfer 5.3 work on Ubuntu 16 by doing the following steps to install an older version of perl :
- install perlbrew:
$> sudo apt-get install perlbrew
- use perlbrew to install perl version 5.20.3:
$> perlbrew install perl-5.20.3
- You must change the '/bin/perl' file to a link which points to the
version of perl installed by perlbrew. (This technique is not the standard practice but it is the only way to get the mni tools to use the older version of perl):
$> cd /bin $> sudo mv perl perl.orig (can always set back later) $> ln -s ~/perl5/perlbrew/perls/perl-5.20.3/bin/perlBut I'm using a better computer now and I use the one with windows for other analyses, so I wanted to know if I could install Virtual box on my ubuntu 16.04 computer, and install the same version of linux used to process the first measure. Should I also need to use the same version of freesurfer?
You could do that, but The Virtual Box version is not really good for processing a large number of subjects because of the limited amount of disk space on the virtual image. Its really meant more as a teaching tool for users with Windows machines.
f the computer's specs are dfferent, does it matter? And if I assign different cpu and ram on my virtual machine?
Different specs are fine. Its the OS, OS version, and version of freesurfer that should stay the same.
Thank you! _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Oks, thank you!
Last question! And using CUDA this time... would the results also differ?
2016-11-15 16:18 GMT+01:00 Z K zkaufman@nmr.mgh.harvard.edu:
On 11/15/2016 07:21 AM, anonymous anonymous wrote:
Thank you so much for the extense answer!
It happens that the person who analysed the first measure (of this longitudinal dataset) used two different OS (he did it in a Virtual machine on windows and a mac OS X) to speed up this step. That data has already been used in some studies.
¿So my question now would be how much does that really affect to the results?
It is impossible to know for sure without processing the results on the same OS. My guess is quite small, generally we see results differ on the order of 5% or less, but still potentially enough to introduce bias.
Hi
I am using recon-all in some longitudinal data on a ubuntu 16.04, and using the development version of freesurfer as I was told here, because the other current version doesnt work on ubuntu 16.04.
The problem is that the first scans were probablly processed using virtualbox in a windows and an older version of freesurfer.
i think I already know the answer, that I can't do this if I dont want to bias my results. Can I?
Yup. You know the answer ;) However, I did discover a way to make freesurfer 5.3 work on Ubuntu 16 by doing the following steps to install an older version of perl :
- install perlbrew:
$> sudo apt-get install perlbrew
- use perlbrew to install perl version 5.20.3:
$> perlbrew install perl-5.20.3
- You must change the '/bin/perl' file to a link which points to the
version of perl installed by perlbrew. (This technique is not the standard practice but it is the only way to get the mni tools to use the older version of perl):
$> cd /bin $> sudo mv perl perl.orig (can always set back later) $> ln -s ~/perl5/perlbrew/perls/perl-5.20.3/bin/perlBut I'm using a better computer now and I use the one with windows for other analyses, so I wanted to know if I could install Virtual box on my ubuntu 16.04 computer, and install the same version of linux used to process the first measure. Should I also need to use the same version of freesurfer?
You could do that, but The Virtual Box version is not really good for processing a large number of subjects because of the limited amount of disk space on the virtual image. Its really meant more as a teaching tool for users with Windows machines.
f the computer's specs are dfferent, does it matter? And if I assign different cpu and ram on my virtual machine?
Different specs are fine. Its the OS, OS version, and version of freesurfer that should stay the same.
Thank you! _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu mailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
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.
On 15 November 2016 at 11:01, anonymous anonymous phoolandevi1989@gmail.com wrote:
Oks, thank you!
Last question! And using CUDA this time... would the results also differ?
Most certainly. For example the GPU version does quite a few reduction sums, where the regular CPU code does accumulation sums.
Richard
freesurfer@nmr.mgh.harvard.edu
-
anonymous anonymous -
R Edgar -
Z K