Freesurfer January 2017

freesurfer@nmr.mgh.harvard.edu

123 participants
215 discussions

Postdoctoral Fellow in Machine Learning in Neuroimaging
by Thomas Yeo 04 Jan '17

04 Jan '17

Dear All, I am looking to hire a postdoc to develop machine learning algorithms for large-scale publicly available MRI and behavioral data. Experience with probabilistic machine learning (aka graphical models) or deep learning is a plus. The ultimate goal is automatic scientific discovery (see recent paper from our lab: http://www.pnas.org/content/113/42/E6535) Topics are flexible, e.g., multi-modal fusion, individual-subject brain parcellation, individual subject behavior prediction, dynamic functional connectivity, meta-analysis, neural mass modeling, multi-scale neuroscience, graph theory, mental disorder subtypes, etc. Other details below. Regards, Thomas Requirements: Ph.D. in computer science, electrical engineering, statistics, computational neuroscience or related fields. The successful applicant will work with an interdisciplinary team of computer scientists and neuroscientists, and must be willing to learn some neuroscience. Research Webpage: https://sites.google.com/site/yeoyeo02/home Compensation: Competitive and commensurate with experience Attraction: Perform ground-breaking research at the National University of Singapore (NUS), while enjoying the beautiful sceneries and cultures of South-East Asia. NUS is a research-intensive university consistently ranked among the top 30 universities in the world (http://en.wikipedia.org/wiki/National_University_of_Singapore#University_ra…) Contact: Email BT Thomas Yeo (ythomas(a)csail.mit.edu) with your CV. Deadline: Whenever the position is filled.

1 0

Problem with freeview
by Jennifer Szeto 04 Jan '17

04 Jan '17

Hi Freesurfer Experts, I have installed Freesurfer5.3 and it seems to be working fine. However when i tried to load up the GLM analysis (after doing mris_preproc and mri_glmfit) results on freeview, i got the following message: "OpenGL warning: No pincher, please call crStateSetCurrentPointers () in your SPU Did not find any volume geometry information in the surface" Just wondering if anyone knows the solution to this? Thanks for your help! Jen

2 3

minor edits to aseg lead to exist with errors in about 25% of my scans
by Lucette Cysique 04 Jan '17

04 Jan '17

Dear FS Wiki I run my scan in FS without any problem example: #@#%# recon-all-run-time-hours 8.342 recon-all -s 877 finished without error then I do some edits to the aseg in freeview edits include: relabelling correctly lateral ventricles voxels mostly in the posterior horn of the lateral ventricles; white matter hyperintensities around those voxels which are erroneously considered as cortex also more often in the posterior horn of the lateral ventricles; and correcting erroneous allocation of choroid plexus also typically in the posterior horn of the lateral ventricles Once done, I re-run using recon-all -s $i -autorecon2-noaseg example: /mridata/workingdata/HIV/subjects/877 /usr/local/FREESURFER/freesurfer5.3.0/bin/recon-all -s 877 -autorecon2-noaseg in 25% of cases, I get the following error message and the process exists example: #-------------------------------------------- #@# Intensity Normalization2 Wed Dec 14 12:33:54 AEDT 2016 /mridata/workingdata/HIV/subjects/011/mri mri_normalize -aseg aseg.mgz -mask brainmask.mgz norm.mgz brain.mgz mri_normalize: aseg volume aseg.mgz must be conformed using segmentation for initial intensity normalization using MR volume brainmask.mgz to mask input volume... reading from norm.mgz... Reading aseg aseg.mgz Linux katana.neura.edu.au 2.6.32-504.3.3.el6.x86_64 #1 SMP Wed Dec 17 01:55:02 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux recon-all -s 011 exited with ERRORS at Wed Dec 14 12:33:56 AEDT 2016 in the other no problem example: #@#%# recon-all-run-time-hours 1.116 recon-all -s 877 finished without error at Thu Dec 1 14:42:27 AEDT 2016

2 3

6.0.0 release
by Tobias Granberg 04 Jan '17

04 Jan '17

Dear Bruce, Doug and the rest of the team, I hope you have had a pleasant start of the new year! We have noticed that there is now a 6.0.0 RC but that the modification date changes while the file name persists. We are planning on running some processing, but it seems to be better to hold off until the stable release. Any time perspective for the release? Will the Linux version be finished before the OSX version? Thanks a lot! Best wishes, Tobias _________________________________________ Tobias Granberg, MD, PhD Karolinska Institutet | Department of Clinical Science, Intervention and Technology Karolinska University Hospital | Department of Radiology 141 86 Stockholm, Sweden Cellular: +46 700 60 60 87

2 2

V6 beta freeview - stopped loading remaining files when one is missing
by Antonin Skoch 04 Jan '17

04 Jan '17

Dear experts, with V6 beta version of freeview I encountered, that when some input file specified in command line is missing, the loading of remaining files is stopped. This was not the case with 5.3 version. I found this feature handy. Do you think that it would be good to implement this behavior also for V6 version of freeview? Regards, Antonin Skoch

2 1

mri_convert error message
by Wujek, Robert 04 Jan '17

04 Jan '17

I'm trying to convert a group of .dcm files to .nii.gz, but I've been getting the following output with an error at the end... mri_convert IM-0020-0001.dcm example.nii.gz $Id: mri_convert.c,v 1.179.2.2 2011/05/16 20:53:47 greve Exp $ reading from IM-0020-0001.dcm... Starting DICOMRead2() dcmfile = /Users/Desktop/example/IM-0020-0001.dcm dcmdir = /Users/Desktop/example Ref Series No = 23 Found 482 files, checking for dicoms Found 480 dicom files in series. First Sorting Computing Slice Direction Vs: 0 0 3 Vs: 0 0 1 Second Sorting Counting frames nframes = 6 nslices = 80 ndcmfiles = 480 PE Dir = ROW (dicom read) TransferSyntaxUID: --1.2.840.10008.1.2.1-- jpegUID: --1.2.840.10008.1.2.4-- Loading pixel data TR=7584.00, TE=11.10, TI=1500.00, flip angle=111.00 i_ras = (-1, 0, 0) j_ras = (0, -1, 0) k_ras = (-0, -0, 1) writing to example.nii.gz... niiWrite(): error opening file example.nii.gz ERROR: failure writing example.nii.gz? I'm wondering how I should interpret this error. I have disk space, and proper permissions. I also have experience using mri_convert with other projects, so generally speaking, I am capable with this script. I would appreciate any suggestions. Thank you. -Robert Wujek?

2 1

freeview V6beta - reposition surface option
by Antonin Skoch 04 Jan '17

04 Jan '17

Dear experts, I have several questions regarding "reposition surface" tool in freeview (V6beta): 1. Are here any documentation how to use "reposition surface" tool? I would especially velcome the explanation of significance of "sigma", "size" and "gradient direction" parameter. I tried to play with the parameters by trial and error, but I could not get satisfactory results. 2. It seems to me that the selection of the active volume (i.e. the volume according to that the surface is repositioned) does not reflect the currently highlighted volume. It seems to always use the volume on top of the list of the volumes, but I am not sure given the unpredictive behaviour of this tool I got so far. 3. It would be maybe handy to include special flag to recon-all to rerun remaining recon-all pipeline after the reposition of the surface in freeview have been used. What do you think? 4. Do you think there is a chance to implement recon-all to preserve part of the surface which has been repositioned by this tool in case of recon-all rerun (similarly like it is done with other edits)? Regards, Antonin Skoch

1 0

Fwd: pial surface errors - V6beta - repost
by Antonin Skoch 04 Jan '17

04 Jan '17

Dear experts, I did not receive response to this post, therefore I am reposting. Also, probably due to the attachment size, my previous post did not get archived by mail archive. Regards, Antonin Skoch From: Antonin Skoch <ansk(a)ikem.cz> To: <freesurfer(a)nmr.mgh.harvard.edu> Sent: 12/21/2016 1:41 AM Subject: pial surface errors - V6beta Dear experts, in my data I quite often encounter errors in pial surface which is probably induced by errors of white surface, in some cases caused by juxtacortical lesions (even in healthy subjects). See screenshots for examples. I have tested this on freesurfer V6 beta but it is commonly also found in older versions. I have uploaded the subject pial_errors.tar.gz to your server. The error showed pial_error1.png in probably caused by small error in white surface shape in 18,-18,63 (RAS). Another similar problem is in 15,10,53 (RAS). This is probably caused by juxtacortical lesion. Another problematic spot with juxtacortical lesion is on -7,18,47 (RAS). I have observed such kind of errors in quite a large number of subjects, these are either healthy controls or subjects with psychiatric ilnesses, therefore no particular apparent macroscopic pathology should be present in these subjects. Do you think that there could be some improvement in the mris_make_surfaces code to prevent this kind of behaviour (i.e. very small error in white has gross effect on pial), or it is necessary in all these cases to manually intervene ? Especially, I do not see any particular problem in image in case of pial_error1.png. And, in case of editing of problems errors by juxtacortical lesions I find fastest way to edit wm (due to the fact that I can constrain the brush not to draw to existing wm) in contrast to editing aseg (i.e. label the spot as a lesion). Do you think that aseg editing in contrast to wm editing has some benefit in this case? Regards, Antonin Skoch

1 0

mri_glmfit-sim cannot determine file for mask after mri_glmfit --no-prune
by Dowling, Kevin Francis 04 Jan '17

04 Jan '17

Hello Freesurfer Experts, My apologies if this is a rather simple/obvious question. I am trying to run 3 Monte Carlos using mri_glmfit-sim with the following commands: Left hemisphere surface: mri_glmfit-sim --glmdir $SUBJECTS_DIR/.../LH_7vFix/ --sim mc-z 10000 1.3 mcz10000_1.3 --sim-sign abs -cwpvalthresh .01 Right Hemisphere surface: mri_glmfit-sim --glmdir $SUBJECTS_DIR/.../RH_7vFix/ --sim mc-z 10000 1.3 mcz10000_1.3 --sim-sign abs -cwpvalthresh .01 Volume: mri_glmfit-sim --glmdir $SUBJECTS_DIR/.../MNI305_7vFix/ --sim mc-z 10000 1.3 mcz10000_1.3 --sim-sign abs -cwpvalthresh .01 I'm running this command after my mri_glmfit that includes the --no-prune flag (command below). While the Monte Carlos for the surface ran without problems, there is an error when I try to run the volume based analysis: Error: could not determine file for /cluster/roffman/.../MNI305_7vFix//mask I believe this error refers to the absence of a mask.mgh file. In my analysis the mask.mgh file is only generated for the fsaverage lh and rh surface analyses after mri_glmfit (when the --no-prune is included), while no mask file is generated for the volume. When I re-run the mri_glmfit command with --prune, the mask.mgh file is generated for the volume and the Monte Carlo appears to run. I don't know if this matters but we run our current volume analyses in the mni305 standard space (constructed via mkanalysis-sess with -mni305 2 and not the -native flag). I'm a bit confused why a mask.mgh file isn't generated for the --no-prune volume analyses but is generated for the --no-prune surface analyses. If my mri_glmfit-sim error is referencing the lack of a mask.mgh file, is there a way that I could create a substitute mask.mgh file to feed into the Monte Carlo? Alternatively might there be another better way to work around the no mask error? My mri_glmfit command for the volume is the following: mri_glmfit --y $SUBJECTS_DIR/.../SIRP_Stable5.3_121316_sm5_mni305/7vFix/ces.nii.gz --fsgd /cluster/.../152_GDD_7vFix_Covary_Continuous_Serum_Folate_Linear_Nuis_GenderClass_Age_RT.fsgd --C /cluster/.../TwoGroups1CovariateRegressOut2covariates.mtx --glmdir /cluster/.../MNI305_7vFix/ --fwhm 6 --no-prune I am running a Linux using Release 6.7 (Santiago) and Freesurfer 5.3 (nmr-stable53-env). I would be most grateful for any suggestions or troubleshooting insights you might have! Best wishes, Kevin Kevin F. Dowling Clinical Research Coordinator Brain Genomics Laboratory Division of Psychiatric Neuroimaging Massachusetts General Hospital 149 13th Street Charlestown, MA, 02129 (p) 617.643.2479

1 0

Structural analysis
by Silas 04 Jan '17

04 Jan '17

Dear FreeSurfer team, I'm currently doing a structural FS analysis with 50 Multiple Sclerosis (MS) patients and 25 Healthy Controls (HC). FLAIR-lesions were drawn onto the aseg.mgz as White Matter Hypointensities (=77). I've finished the analysis, but have some questions regarding my results. Results: HC (n=25) MS (n=50) p-value Mean TIV (SD) [ml] 1594.9 (154.7) 1559.5 (140.1) > 0.05 Mean WMV (SD) [ml] 503.6 (57.3) 485.2 (59.5) > 0.05 Mean GMV (SD) [ml] 657.2 (69.9) 635.1 (47.9) > 0.05 Mean WMHV (SD) [ml] 1.12 (0.4) 5.97 (5.7) 0.0000677 1) Does these results look correct? WMV < GMV? 2) What is the difference between WMV+GMV and TIV? CSF? Ventricles? 3) I find no atrophy in the MS patients; are there reported any problems with FS and MS patients? 4) What is White Matter Hypointensities comprised of in HC? Best, Silas

1 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 January 2017