Hi Martin,
Thanks for your help, I think you are right, there is no need to average so many MPRAGE images, I will simply select the best one to reconstruct.
Thank you very much for helping me.
Best
Dong
发件人: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces@nmr.mgh.harvard.edu] 代表 Martin Reuter
发送时间: 2015年11月30日 22:13
收件人: Freesurfer support list
主题: Re: [Freesurfer] 答复: recon-all with more than 10 inputs
Hi Dong,
you should see in the scripts/recon-all.log file, what the exact command is.
The list of all inputs gets created via:
set RunList = `ls $subjdir/mri/orig/[0-9][0-9][0-9].mgz`
so 4 digit file names are not supported. It should work with three, not sure what it going on. Please post the mri_robust_template command, that it produces in the motion-correction stage.
The algorithm that recon-all uses is a call to mri_robust_template, where it passes all the inputs as --mov, jointly registers them (to the first input from the list) and outputs the average after registration.
One thing though. Why in the world would you try to average so many images? We usually recommend to select the best one and run that instead of averaging. MPRAGE or multi echo MPRAGE are usually good quality and don't really need averaging at all (I could not show an improvement of averaging so far). Also the average is of course affected by all inputs, so if some have motion artifacts, that will go into the average. Much better to discard those low quality images. If you have evidence that for your data averaging is a good idea, then why not take 2 or 3 images of the best images?
Best, Martin
On 11/30/2015 05:04 AM, Dong Haoming wrote:
Dear freesurfer exports,
I still have the problem to recon the surface with more than 10 inputs from one person with following command:
“recon-all -s ${subjec} -autorecon1 –notal-check -clean-bm -gcut -no-isrunning -noappend”,
Is there a limit on the number of the inputs?
What is the algorithm applied by “recon-all” to average these images? Is it viable to average input images with other methods, then apply “recon-all” to the averaged result?
Thanks very much for your help!
Dong Haoming
发件人: freesurfer-bounces@nmr.mgh.harvard.edu [mailto:freesurfer-bounces@nmr.mgh.harvard.edu] 代表 董昊铭
发送时间: 2015年11月26日 13:11
收件人: freesurfer@nmr.mgh.harvard.edu
主题: [Freesurfer] recon-all with more than 10 inputs
Dear freesurfer exports,
I am trying to recon more than 10 T1 scans from a single subject, but something went wrong.
I have converted all T1 nifti files into mgz format into mri/orig directory with following names: 001.mgz 002.mgz …. 010.mgz 011.mgz… But when I run the command:
“recon-all –s ${subjec} –autorecon1 –notal–check –clean-bm –gcut –no-isrunning -noappend”,
it seems that recon-all only recognizes the files with names 01*.mgz(i.e. 0013.mgz), the 00*.mgz(i.e. 001.mgz) files failed to be processed. Once I change the files name to “001.mgz 002.mgz …. 0010.mgz 0011.mgz”, the 00*.mgz(i.e. 001.mgz) are able to be recognized, but the 001*.mgz(i.e. 0013.mgz) files failed.
How should I deal with this to make all files recognized?
Sincerely,
Dong Haoming
_______________________________________________Freesurfer mailing listFreesurfer@nmr.mgh.harvard.eduhttps://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Martin Reuter, PhDAssistant Professor of Radiology, Harvard Medical SchoolAssistant Professor of Neurology, Harvard Medical SchoolA.A.Martinos Center for Biomedical ImagingMassachusetts General HospitalResearch Affiliate, CSAIL, MITPhone: +1-617-724-5652Web : http://reuter.mit.edu