Freesurfer

freesurfer@nmr.mgh.harvard.edu

30132 discussions

Qdec question when examining group differences with continuous covariate and nuisance factor
by James Allebone 22 Mar '18

22 Mar '18

Hi there, I am running an analysis to examine differences in cortical thickness between two groups (clinical vs control) and have entered ICV as a nuisance variable and age as a covariate. The analysis results in qdec do not appear to give a result that answers the question: does the average cortical thickness differ between clinical and control with both the nuisance factor and the covariate accounted for. The only result comparing average thickness between the groups lists ICV as a nuisance factor, and does not include age as a covariate. Other results are examining correlation between age and thickness accounting for group, or whether the thickness-age correlation differs between groups accounting for ICV. A second question: when i just include ICV as a nuisance factor and do not enter age as a covariate (groups mean age is almost identical), there are some areas of cortical thickening in the clinical group, which become much larger after correction for multiple comparisons using Monte Carlo Null-Z simulation (threshold 0.05). This seems counterintuitave to me as I would expect a reduction in significant clusters after correction. One of the main clusters traverses the cingulate gyrus, however, when I extract cortical thickness values for each individual subject and compare group means for subregions of the cingulate in SPSS, there are no significant differences on any of the subregions. In fact, they are almost identical. Could adding ICV as a nuisance variable account for this? Mean ICV values are almost identitical between groups. Or is there some additive affect at the cluster level? Many thanks for any clarity you can provide. Kind regards, James

1 0

mri_glmfit-sim on fsaverage5
by Tuominen, Lauri Johannes 22 Mar '18

22 Mar '18

Hello freesurfers! My resting state data is in fsaverage5 space. mri_info gives dimensions of 3414 x 1 x 3 x 120 and voxelsize 1x1x1. I do the group analysis and mri_glmfit creates a sig.mgh file that is has those same dimensions and results look good. Then I say: mri_glmfit-sim --glmdir /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204 --3spaces --sim mc-z 10000 1.3 mc_1.3_3space --sim-sign abs —overwrite And I get an error message saying dimension mismatch between input volume and seg. below is the tail of the message. Thank you for your help! Lauri number of voxels in search space = 9354 Done loading source values (nvtxs = 10242) overall max = 4.53945 at vertex 1977 overall min = -3.95722 at vertex 473 surface nvertices 10242 metric props tot surface area 66661.796875 group_avg_vtxarea_loaded 1 masked surface area 79140.789062 Computing voxel-wise significance CSDpvalMaxSigMap(): found 347/10242 above 0, max=0.485948 NOT Adjusting threshold for 1-tailed test thminadj = 1.3 Searching for Clusters ... thmin=1.300000 (1.300000), thmax=-1.000000 (-1), thsignid=0, minarea=0.000000 Found 46 clusters Max cluster size 1840.913696 Pruning by CW P-Value 0.05 Saving thresholded output to /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204/intercept/mc_1.3_3space.sig.masked.mgh Saving cluster numbers to /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204/intercept/mc_1.3_3space.sig.ocn.mgh Saving cluster pval /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204/intercept/mc_1.3_3space.sig.cluster.mgh Constructing output annotation Writing annotation /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204/intercept/mc_1.3_3space.sig.ocn.annot mri_segstats --seg /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204/intercept/mc_1.3_3space.sig.ocn.mgh --exclude 0 --i /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204.gm.nii.gz --avgwf /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204/intercept/mc_1.3_3space.y.ocn.dat --sum /tmp/mri_glmfit-sim.junk.8849 $Id: mri_segstats.c,v 1.121 2016/05/31 17:27:11 greve Exp $ cwd cmdline mri_segstats --seg /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204/intercept/mc_1.3_3space.sig.ocn.mgh --exclude 0 --i /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204.gm.nii.gz --avgwf /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204/intercept/mc_1.3_3space.y.ocn.dat --sum /tmp/mri_glmfit-sim.junk.8849 sysname Linux hostname bee.nmr.mgh.harvard.edu machine x86_64 user ljt22 UseRobust 0 Loading /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204/intercept/mc_1.3_3space.sig.ocn.mgh Loading /autofs/space/bee_001/users/Lauri/test10.nosm.group/lh.surf.avg.rh.204.gm.nii.gz ERROR: dimension mismatch between input volume and seg input 3414 1 3 seg 10242 1 1

2 1

Re: [Freesurfer] Skull stripping error - too much brain lost - solved {Disarmed}
by Matthew Grecsek 22 Mar '18

22 Mar '18

Doug, your added text would have been enough to keep me out of the weeds. Thanks. Douglas Greve <https://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=from…> Thu, 22 Mar 2018 13:28:02 -0700 <https://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=date…> I added some text to clarify in that wiki page. See if you think it issufficient (and suggest some wording if you don't:). On 3/22/18 4:22 PM, Bruce Fischl wrote: Hi Matt it's not that trivial. There are frequently multiple T1-weighted protocols, and the naming is hardly standardized. We would need to run some type of MR simulator to decide if something was T1-weighted in the most general sense, and even then it could be mag-trans or with contrast or something. cheers Bruce On Thu, 22 Mar 2018, Matthew Grecsek wrote: Hi guys, The problem was garbage in - garbage out: 1. The issue manifested in the skull stripping step but was caused by bad input 2. I am a data science guy (not neurology), new to freesurfer, andfollowed the instructions at https://surfer.nmr.mgh.harvard.edu/fswiki/FreeSurferBeginnersGuide recon-all \ -i <one slice in the anatomical dicom series> \ 3. When I issued recon-all, I gave it a -i of an image file from my mri data. I assumed that it would find the T1- weighted data from my collection automatically as in dcmunpack. 4. It seems that the -i requires specifically the T1-weighted data, because when I looked at the tkmedit brainmask images they were inverted (maybe from T2 or something else). 5. I re-ran the recon-all on the proper T1 data and it ran to completion with no errors. I received several offline requests from other users, so it seems like this may be a common mistake. I recommend making the requirement for input of T1-weighted data moreexplicit in the wiki page. As a feature request, it seems pretty easy to use dcmunpack to find the correct data from a collection and make it part of initial recon-all. I had uploaded my original data to the ftp site per reply below, but feel free to delete and close the issue as solved. Cheers, -Matt ------------ Hi Matt no idea what that would happen. If you upload the dataset we will take a look cheers Bruce On Mon, 5 Mar 2018, Matthew Grecsek wrote: Following your advice below, it was clearly apparent that the output from the skull stripping step was missing a lot. I tried adjusting the -wsthresh flag from 50 -> 200 and it still left major parts of the brain out. I have a good quality mri and subject with no known anatomical abnormalities. The log file indicates "surface validation has detected possible Error" Does this suggest a more meaningful defect in the base image or something that can be corrected via other parameters or manual editing? Log file attached. Thanks, -Matt Bruce Fischl Fri, 02 Mar 2018 08:00:11 -0800 Hi Matt check the output of the skull stripping. This error usually means that too much (everything?) has been erased and no plausible cc could be found cheers Bruce On Thu, 1 Mar 2018, Matthew Grecsek wrote: My recon-all ended prematurely with the following message: final transformation (x=136.0, yr=-10.054, zr=-0.114): 0.98464 0.00200 -0.17457 15.51372; -0.00197 1.00000 0.00035 17.22442; 0.17457 0.00000 0.98464 -17.83761; 0.00000 0.00000 0.00000 1.00000; mri_cc: no WM voxels found with norm > 40 -- check skull stripping No such file or directory Linux centos-7-1 3.10.0-693.17.1.el7.x86_64 #1 SMP Thu Jan 25 20:13:58 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux recon-all -s matt2 exited with ERRORS at Thu Mar 1 04:35:09 UTC 2018 I am running: freesurfer-Linux-centos6_x86_64-stable-pub-v6.0.0-2beb96c My full log file is attached. Please let me know suggestions for correcting and continuing the process. Thanks, -Matt _______________________________________________ Freesurfer mailing listFreesurfer@nmr.mgh.harvard.eduhttps://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer Reply via email to [image: The Mail Archive] <https://www.mail-archive.com/> - The Mail Archive home <https://www.mail-archive.com/> - freesurfer - all messages <https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/> - freesurfer - about the list <https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/info.html> - Expand <https://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=subj…> - Previous message <https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg57403.html> - Next message <https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg57404.html#>

1 0

Skull stripping error - too much brain lost - solved
by Matthew Grecsek 22 Mar '18

22 Mar '18

Hi guys, The problem was garbage in - garbage out: 1. The issue manifested in the skull stripping step but was caused by bad input 2. I am a data science guy (not neurology), new to freesurfer, and followed the instructions at https://surfer.nmr.mgh.harvard.edu/fswiki/FreeSurferBeginnersGuide recon-all \ -i <one slice in the anatomical dicom series> \ 3. When I issued recon-all, I gave it a -i of an image file from my mri data. I assumed that it would find the T1- weighted data from my collection automatically as in dcmunpack. 4. It seems that the -i requires specifically the T1-weighted data, because when I looked at the tkmedit brainmask images they were inverted (maybe from T2 or something else). 5. I re-ran the recon-all on the proper T1 data and it ran to completion with no errors. I received several offline requests from other users, so it seems like this may be a common mistake. I recommend making the requirement for input of T1-weighted data more explicit in the wiki page. As a feature request, it seems pretty easy to use dcmunpack to find the correct data from a collection and make it part of initial recon-all. I had uploaded my original data to the ftp site per reply below, but feel free to delete and close the issue as solved. Cheers, -Matt ------------ Hi Matt no idea what that would happen. If you upload the dataset we will take a look cheers Bruce On Mon, 5 Mar 2018, Matthew Grecsek wrote: Following your advice below, it was clearly apparent that the output from the skull stripping step was missing a lot. I tried adjusting the -wsthresh flag from 50 -> 200 and it still left major parts of the brain out. I have a good quality mri and subject with no known anatomical abnormalities. The log file indicates "surface validation has detected possible Error" Does this suggest a more meaningful defect in the base image or something that can be corrected via other parameters or manual editing? Log file attached. Thanks, -Matt Bruce Fischl Fri, 02 Mar 2018 08:00:11 -0800 Hi Matt check the output of the skull stripping. This error usually means that too much (everything?) has been erased and no plausible cc could be found cheers Bruce On Thu, 1 Mar 2018, Matthew Grecsek wrote: My recon-all ended prematurely with the following message: final transformation (x=136.0, yr=-10.054, zr=-0.114): 0.98464 0.00200 -0.17457 15.51372; -0.00197 1.00000 0.00035 17.22442; 0.17457 0.00000 0.98464 -17.83761; 0.00000 0.00000 0.00000 1.00000; mri_cc: no WM voxels found with norm > 40 -- check skull stripping No such file or directory Linux centos-7-1 3.10.0-693.17.1.el7.x86_64 #1 SMP Thu Jan 25 20:13:58 UTC 2018 x86_64 x86_64 x86_64 GNU/Linux recon-all -s matt2 exited with ERRORS at Thu Mar 1 04:35:09 UTC 2018 I am running: freesurfer-Linux-centos6_x86_64-stable-pub-v6.0.0-2beb96c My full log file is attached. Please let me know suggestions for correcting and continuing the process. Thanks, -Matt

3 2

Problems displaying pial_lgi in dec
by Dusan Hirjak 22 Mar '18

22 Mar '18

Dear Freesurfer Experts, I would like to display my results after running the LGI command on my subjects (via Freesurfer 6.0, MacOS Sierra). Although I created the .Qdecrc (text file) file with MEASURE1 = pial_lgi and put it in my home directory as well as in the folder where my study subjects are, it is still not working. I can not see "pial_lgi" in the Qdec drop down menu. I would be very thankful for any suggestions. Dusan

2 1

PetSurfer: NaNs in MGX volumes
by Elijah Mak 22 Mar '18

22 Mar '18

Hi Doug, Reposting this on the FS mailing list: After MGX-PVC and MRTM2, I sometimes get NaNs in the voxelwise corrected PET data when I load it on Freeview. Error message looks like this WARNING: 19 NaNs found in volume Is this happening because the MGX GM threshold is too high (I am using 0.2) or could there be something else happening? Appreciate your help. Please let me know if you need more information. Thanks. Best Wishes, Elijah

2 1

Re: [Freesurfer] Hippocampal subfields
by Iglesias Gonzalez, Eugenio 22 Mar '18

22 Mar '18

Dear Arsenije The atlas is encoded in a tetrahedral mesh, so there is no nifti file for it. Having said that, it is possible to rasterize the atlas at a target resolution. The command is the following: kvlRasterizeAtlasMesh $FREESURFER_HOME/average/HippoSF/atlas/AtlasMesh.gz $FREESURFER_HOME/average/HippoSF/atlas/AtlasDump.mgz X -1 where X is the label number (see $FREESURFER_HOME/average/HippoSF/atlas/compressionLookupTable.txt for a listing; 0 is cortex, 1 is cerebral white matter, etc). -1 indicates the reference position of the atlas, so do not modify that. This command will create a file called rasterized.mgz. You can convert it to nifty with mri_convert, eg. mri_convert rasterized.mgz probability_label_X.nii.gz You can do this with all the different hippocampal labels. And if you want a probability for the whole hippocampus, which is probably what you need for the registration with your manual mask, you can sum the probabilities of the different subfields into a single volume (e.g., with fslmaths) I hope this helps, /Eugenio -- Juan Eugenio Iglesias ERC Senior Research Fellow Translational Imaging Group University College London http://www.jeiglesias.com http://cmictig.cs.ucl.ac.uk/ From: <freesurfer-bounces(a)nmr.mgh.harvard.edu> on behalf of Arsenije Subotic <arsenije.subotic1(a)ucalgary.ca> Reply-To: Freesurfer support list <freesurfer(a)nmr.mgh.harvard.edu> Date: Thursday, 22 March 2018 at 15:15 To: "freesurfer(a)nmr.mgh.harvard.edu" <freesurfer(a)nmr.mgh.harvard.edu> Subject: [Freesurfer] Hippocampal subfields Hi all I would like to use subfield specific hippocampus maps that were produced Dr Iglesias's Lab: https://sites.google.com/site/jeiglesias/hippocampal-subfields These maps are available via FreeSurfer, however our lab is trying to segment hippocampus subregions from already manually extracted hippocampuses. In other words, we have manually traced hippocampus masks in a NIFTI format that we would like to co-register with these maps. How can I directly download subfield specific hippocampal atlases in a NIFTI format so that we can use non linear co registration after? Thank you very much. Look forward hearing from you, Arsenije

1 0

Hippocampal subfields
by Arsenije Subotic 22 Mar '18

22 Mar '18

Hi all I would like to use subfield specific hippocampus maps that were produced Dr Iglesias's Lab: https://sites.google.com/site/jeiglesias/hippocampal-subfields These maps are available via FreeSurfer, however our lab is trying to segment hippocampus subregions from already manually extracted hippocampuses. In other words, we have manually traced hippocampus masks in a NIFTI format that we would like to co-register with these maps. How can I directly download subfield specific hippocampal atlases in a NIFTI format so that we can use non linear co registration after? Thank you very much. Look forward hearing from you, Arsenije

1 0

Fwd: Correnting image bias before segmentation and parcelation - due to frontal and occipital shading.
by Gamaliz 22 Mar '18

22 Mar '18

Dear Freesurfer Team, I have about 200 T1 coronal scans that were obtained with some shading in the frontal and occipital regions. Not sure if it can be seen from the attached picture, but the image fades toward the top and bottom (on the axial view). The segmentation left out a large section of the frontal and occipital, probably because the signal loss brings those voxels below the threshold for white and gray matter. I have tried to run FSL bias correction and got a small improvement, but I wonder if there is any way these faded slides can be corrected so that the segmentation is capable of seeing these as proper white and gray matter. In the attached picture, the blue is the wm surface and the red is the gray matter (wm and pial surfaces) The image shown is actually the T1.mgz from freesufer that has already gone through the nu_correct step in recon_all, the original image looks very similar... Thanks Gabriel -- gAbE

2 4

Re: [Freesurfer] Regarding hippocampus tesellation
by Iglesias Gonzalez, Eugenio 22 Mar '18

22 Mar '18

Hi Pooja I'd suggest reading in the mgz volume with MATLAB (you can use the function MRIread in $FREESURFER_HOME/matlab) and export it to your favorite format from there Cheers Eugenio Sent from my phone, please excuse brevity and typos ________________________________ From: pooja prabhu <prabhuppooja(a)gmail.com> Sent: Thursday, March 22, 2018 9:47:52 AM To: freesurfer(a)nmr.mgh.harvard.edu Subject: Regarding hippocampus tesellation Hello, This is Pooja. I am trying to segment the hippocampus using the function introduced by Freesurfer i.e -hippocampal-subfields. Segmentation was successful but the final file format is .mgz. Could you please guide me how can i tessellate the hippocampus and cortical surface? i want this surface to place the dipole so please help me. Thank -- Thank You Pooja Prabhu 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.

1 0

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