Freesurfer

freesurfer@nmr.mgh.harvard.edu

10 participants
30093 discussions

mrisComputeCorrelationTerm: delta is not finite at vno 0
by Juhyoung Ryu 07 Feb '17

07 Feb '17

Hi FS developers, I received errors in one of steps in recon-all. I analyzed 111 individual subjects' T1 data using same method, but there are errors in only 14 subjects. Below is screenshot image in the end of recon-all.log file. I attached log files from two exemplar subjects below, also. I’d appreciate if you might help me. Juhyoung.

3 7

QDEC table .dat
by Leblanc Élizabel 07 Feb '17

07 Feb '17

Hi Freesurfer's expert, I have some trouble with the dec.table.dat file, as many others as I can see. I am using a mac OS x, I took an excel file with my 4 variables and open it in a .txt format. then I tried to convert it to a .dat file. I think it worked well (I wrote mv qdec.table.txt qdec.table.dat in the command terminal). But when I try to load the table using QDEC it still give me this error message. Loading data table /Applications/freesurfer/subjects/qdec.table.dat... ERROR: QdecDataTable::Load could not find column named 'fsid' or 'ID' in /Applications/freesurfer/subjects/qdec.table.dat! Error loading the data table. I have attached to this email my qdec table. Can you please give me advice on how to create a qdec table (as simple as possible because I am not pretty familiar with it to this point). Thank you very much

2 1

Re: [Freesurfer] -FLAIRpial in v6.0 positions pial surfaces much more externally in comparison to v5.3
by Antonin Skoch 07 Feb '17

07 Feb '17

Dear Matt, I do not use HCP pipeline scripts for this project. This is standard 1x1x1 mm3 data so I thought standard freeSurfer pipeline could work well here. I also thought that standard recon-all version 6.0 with this data should work better than v5.3 which, I expect, solved some issues with v5.3 (some of them I was discussing in the forum previously): http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg41880.html http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg42312.html http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg41375.html But anyway, thank you for your feedback. Antonin Since I believe you said you are using my code, I would recommend to stick with 5.3 at this time.Peace, Matt. From: <freesurfer-boun...(a)nmr.mgh.harvard.edu> on behalf of Antonin Skoch <a...(a)ikem.cz> Reply-To: Freesurfer support list <freesurfer(a)nmr.mgh.harvard.edu> Date: Tuesday, February 7, 2017 at 2:59 PM To: <freesurfer(a)nmr.mgh.harvard.edu> Subject: Re: [Freesurfer] -FLAIRpial in v6.0 positions pial surfaces much more externally in comparison to v5.3 Dear Bruce, thank you for the feedback. I would definitely be interested in 6.1 beta. What is the approximate timeline of release? We are working on quite large project (involving -FLAIRpial) and I have to decide whether to go to v6.0 or to stick with v5.3. Do you think that so far the 5.3 version is better for T2/FLAIR? Why this is the case? Antonin Hi Antonin yes, V6 was not meant for T2/FLAIR and we didn't test it extensively. We have been optimizing things in the past week for T2/FLAIR and I expect 6.1 will be better for them. We can give you a beta soon if you like. cheers Bruce On Tue, 7 Feb 2017, Antonin Skoch wrote: > Dear experts, > > I have reprocessed my groups of subjects (originally processed by v5.3) by > v6.0. I am using -FLAIRpial for pial surface refinement. > > I have noticed that the pial surfaces in v6.0 after -FLAIRpial are positioned > much more externally in comparison to -FLAIRpial in v5.3. The v6.0 version > includes to pial sheet much more partial volume voxels in > comparison to v5.3. > > In contrast to v5.3, where the surfaces after -FLAIRpial were mostly > internally with respect to pial surfaces determined by T1, in v6.0 there are > many regions where the pial surfaces after -FLAIRpial are placed > much externally with respect to pial surfaces determined by T1. > > See the screenshot (red surfaces are pial surfaces after -FLAIRpial with > v6.0, yellow are surfaces after -FLAIRpial with v.5.3). > > I have used -nsigma_above 8 in mris_make_surfaces to prevent excluding some > cortical regions with high signal intensity but this I used consistently in > both versions so it should not be the reason of difference. > > I have uploaded the subject processed both by 5.3 and 6.0 version to your > server, files FLAIRpial_v5.3.tar.gz and FLAIRpial_v6.0.tar.gz. > > Could you please look at my data and check whether the pial surfaces after > -FLAIRpial in v6.0 look OK from your perspective? > > Regards, > > Antonin Skoch

1 0

Re: [Freesurfer] -FLAIRpial in v6.0 positions pial surfaces much more externally in comparison to v5.3
by Antonin Skoch 07 Feb '17

07 Feb '17

Dear Bruce, thank you for the feedback. I would definitely be interested in 6.1 beta. What is the approximate timeline of release? We are working on quite large project (involving -FLAIRpial) and I have to decide whether to go to v6.0 or to stick with v5.3. Do you think that so far the 5.3 version is better for T2/FLAIR? Why this is the case? Antonin Hi Antonin yes, V6 was not meant for T2/FLAIR and we didn't test it extensively. We have been optimizing things in the past week for T2/FLAIR and I expect 6.1 will be better for them. We can give you a beta soon if you like.cheers Bruce On Tue, 7 Feb 2017, Antonin Skoch wrote: > Dear experts, > > I have reprocessed my groups of subjects (originally processed by v5.3) by > v6.0. I am using -FLAIRpial for pial surface refinement. > > I have noticed that the pial surfaces in v6.0 after -FLAIRpial are positioned > much more externally in comparison to -FLAIRpial in v5.3. The v6.0 version > includes to pial sheet much more partial volume voxels in > comparison to v5.3. > > In contrast to v5.3, where the surfaces after -FLAIRpial were mostly > internally with respect to pial surfaces determined by T1, in v6.0 there are > many regions where the pial surfaces after -FLAIRpial are placed > much externally with respect to pial surfaces determined by T1. > > See the screenshot (red surfaces are pial surfaces after -FLAIRpial with > v6.0, yellow are surfaces after -FLAIRpial with v.5.3). > > I have used -nsigma_above 8 in mris_make_surfaces to prevent excluding some > cortical regions with high signal intensity but this I used consistently in > both versions so it should not be the reason of difference. > > I have uploaded the subject processed both by 5.3 and 6.0 version to your > server, files FLAIRpial_v5.3.tar.gz and FLAIRpial_v6.0.tar.gz. > > Could you please look at my data and check whether the pial surfaces after > -FLAIRpial in v6.0 look OK from your perspective? > > Regards, > > Antonin Skoch

2 1

-first_wm_peak with -hires in v6.0 (search for peak in second derivative)
by Antonin Skoch 07 Feb '17

07 Feb '17

Dear Bruce, Would you recommend to use -first_wm_peak with -hires option? My tests with -first_wm_peak on v6.0 ( with search for peak in second derivative) indicate that the white surface is placed too much internally with respect to actual gm/wm interface. See the screenshots (with -first_wm_peak and without -first_wm_peak). I think that before update to second derivative (december 2016) the placement of white surface with -first_wm_peak was more anatomically relevant. Do you plan also to optimize this for v6.1? Regards, Antonin Dear experts, I am reposting here my mail to FreeSurfer experts (so far without response) concerning my results of testing of mris_make_surfaces in V6 freeSurfer version with modified HCP pipeline (tested on dev version from 3rd January, which is practically the same as now released 6.0.0 version, since no additional commits regarding mris_make_surfaces have been done). My inquiry was concerning the change of behavior of -first_wm_peak option of mris_make_surfaces in 6.0.0 version. Apart from previous version the -first_wm_peak option now tries to search also for peak in second derivative. It seems to me that this modification of -first_wm_peak behavior leads now to the white surface placement too internally in gray matter which is not very anatomically relevant. See the screenshots where I compared the results: 1. standard recon-all -hires version 6.0.0 adding -first_wm_peak to mris_make_surfaces (trying to search for peak in second derivative) 2. FreeSurferHiResWhite.sh with 6.0.0 version and -first_wm_peak (trying to search for peak in second derivative) 3. FreeSurferHiResWhite.sh with V6 beta version (from december 2016, before change) and -first_wm_peak (without trying to search for peak in second derivative) 4. standard recon-all -hires version 6.0.0 default setting, i.e. without adding -first_wm_peak to mris_make_surfaces - quite similar but not identical results to the 3 - some regions are more internally, some more externally wrt to 3 - the difference can be also caused by better gm/wm contrast in brain.finalsurfs wrt T1w_hires_norm used in FreeSurferHiResWhite.sh. Could you please get time to look at the results and comment on which option you assess as most anatomically relevant? Regards, Antonin Skoch Subject: Re: posts about recent updates of freesurfer V6beta Dear experts, I also investigated further the second issue - change of the behavior of -first_wm_peak of mris_make_surfaces. I am using modified HCP pipeline where the mris_make_surfaces -first_wm_peak is used. I worked to adapt this pipeline to work with V6beta version of FreeSurfer with using -cm flag to impose high-resolution reconstruction. With the change of behavior of -first_wm_peak option (to look for first peak in 2nd derivative) by commit in 20th December 2016 the results with up-to-date versions are no longer compatible with previous versions (where first peak in 1st derivative was used I suppose). My testing using -first_wm_peak with the version after 20th December 2016 shows that the white surface is placed much more internally towards white matter. I am not sure the placement of white surface so internally is anatomically relevant. See the screenshots of examples where 1. -first_wm_peak with version from 3rd January 2017 using brain.finalsurfs 2. -first_wm_peak with version from 3rd January 2017 using normalized T1w image (i.e. image without which did not underwent full previous steps of recon-all - this is default implementation of FreeSurferHiResWhite.sh in HCP pipeline). 3. -first_wm_peak with version from November 2016 using normalized T1w image All versions use -orig -orig_white options with white surface generated by recon-all -white (i.e. mris_make_surfaces run without -first_wm_peak using brain.finalsurfs). This white surface generated by recon-all -white in 3rd January version is shown in screenshot 4. Here are my specific questions / suggestions: Do you think the -first_wm_peak results with the 3rd January 2017 version are anatomically relevant? To assure backward compatibility, I would vote for keeping option to look for first peak in 1st derivative in mris_make_surfaces (aside from the new option with looking to the first peak in the 2nd derivative). Could you please comment on? Regards, Antonin Skoch

2 1

(no subject)
by Smagula, Stephen Fearns 07 Feb '17

07 Feb '17

Dear Dr. Iglesias, Using the originally released version 6 (the one that was "unreleased") I ran the recon -all -all and the subfield segmentations on a sample. I noticed that the program was taken down then put back online after some updates. Do you recommend we re-run these segmentations with the newly released version? That is, if the segmentation ran with the "old FreeSurfer 6", would you expect there to be any difference with the new release? Thank you, Stephen Smagula Post-doc University of Pittsburgh

3 2

Re: [Freesurfer] T2 and control points
by Antonin Skoch 07 Feb '17

07 Feb '17

Dear Paul, I don't know it it solves your problem, but if you are using bash shell, you have to source SetUpFreeSurfer.sh, not SetUpFreeSurfer.csh. And, do you have FREESURFER_HOME variable properly set before running the script? To check that, put echo $FREESURFER_HOME to the beginning of the script. Antonin Thanks Antonin and Bruce. Also, I wrote a recon-all script and when I tried to run it, it gave me an error (command not found). Please can anyone help me diagnosis what's wrong with my scripts. I did set my subject directory to the current folder (setenv SUBJECTS_DIR /........../folder/; cd /........../folder) before running the script. Thanks. The scripts is below#!bin/bash # FreeSurfer Pipeline (stages 1-31) # Written By Paul set -e # Source FreeSurfer source $FREESURFER_HOME/SetUpFreeSurfer.csh # Start Scripts echo -e "\n START: FreeSurferEntirePipeline" # Input Variables SubjectID="$1" SubjectDIR="$2" T1wImage="$3" #T1 FreeSurfer Input T2wImage="$4" #T2 FreeSurfer Input recon-all -s $SubjectID -autorecon-all -i "$T1wImage"_T1.nii -T2 "$T2wImage"_T2.nii -T2pial -parallel -hires echo -e "done" Best, Paul On Tue, Feb 7, 2017 at 4:40 AM, Antonin Skoch <a...(a)ikem.cz> wrote: > Dear Paul, > > maybe this is implicit, but to assure: > > you have also to include -autorecon3. > > Therefore the full command is: > > recon-all -s subject_id -autorecon2-cp -autorecon3 -T2pial > > Antonin > > Yes, I think so > > > On Feb 5, 2017, at 2:47 PM, miracoo...(a)gmail.com wrote: > > > > Hello, > > I ran the entire recon pipeline with T2 option and then added control > > points. Do I need to run recon-all -autorecon2-cp with T2 flag to redo the > > pial surface or not? Thanks. > > Paul > > >

2 1

-FLAIRpial in v6.0 positions pial surfaces much more externally in comparison to v5.3
by Antonin Skoch 07 Feb '17

07 Feb '17

Dear experts, I have reprocessed my groups of subjects (originally processed by v5.3) by v6.0. I am using -FLAIRpial for pial surface refinement. I have noticed that the pial surfaces in v6.0 after -FLAIRpial are positioned much more externally in comparison to -FLAIRpial in v5.3. The v6.0 version includes to pial sheet much more partial volume voxels in comparison to v5.3. In contrast to v5.3, where the surfaces after -FLAIRpial were mostly internally with respect to pial surfaces determined by T1, in v6.0 there are many regions where the pial surfaces after -FLAIRpial are placed much externally with respect to pial surfaces determined by T1. See the screenshot (red surfaces are pial surfaces after -FLAIRpial with v6.0, yellow are surfaces after -FLAIRpial with v.5.3). I have used -nsigma_above 8 in mris_make_surfaces to prevent excluding some cortical regions with high signal intensity but this I used consistently in both versions so it should not be the reason of difference. I have uploaded the subject processed both by 5.3 and 6.0 version to your server, files FLAIRpial_v5.3.tar.gz and FLAIRpial_v6.0.tar.gz. Could you please look at my data and check whether the pial surfaces after -FLAIRpial in v6.0 look OK from your perspective? Regards, Antonin Skoch

2 1

Re: [Freesurfer] T2 and control points
by Antonin Skoch 07 Feb '17

07 Feb '17

Dear Paul, maybe this is implicit, but to assure: you have also to include -autorecon3. Therefore the full command is: recon-all -s subject_id -autorecon2-cp -autorecon3 -T2pial Antonin Yes, I think so > On Feb 5, 2017, at 2:47 PM, miracoo...(a)gmail.com wrote: > > Hello, > I ran the entire recon pipeline with T2 option and then added control > points. Do I need to run recon-all -autorecon2-cp with T2 flag to redo the > pial surface or not? Thanks. > Paul >

2 1

Percentage of cortex showing significant differences
by Bronwyn Overs 07 Feb '17

07 Feb '17

Dear mailing list, I have completed a matlab based lme analysis using cortical thickness, and now have a list of significant clusters for each coefficient. I now want to know what proportion of the cortex differs significantly for each coefficient. I can add up the total size of the significant clusters for each coefficient, but I don’t know what value to use in the denominator (the total size of the cortex). Would it be the total area of the fsaverage inflated surface that I used in my analysis? Kind regards, Bronwyn Overs Research Assistant Neuroscience Research Australia Margarete Ainsworth Building Barker Street Randwick Sydney NSW 2031 Australia M 0411 308 769 T +61 2 9399 1883 F +61 2 9399 1265 neura.edu.au <http://neura.edu.au/> <https://twitter.com/neuraustralia> <https://www.facebook.com/NeuroscienceResearchAustralia> <http://www.neura.edu.au/help-research/subscribe>

2 1

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