Freesurfer

freesurfer@nmr.mgh.harvard.edu

5 participants
30076 discussions

freesufer v6 beta - T2 pial refinement performance and nsigma_below setting
by Antonin Skoch 27 Nov '16

27 Nov '16

Dear FreeSurfer experts, I have processed a group of subjects by the freesurfer v6 beta version (with recent upgrade of mris_make_surfaces from 2016/11/21). It seems to me that in contrast to v5.3 version (with the same nsigma_above 2, nsigma_below 3) the T2 pial refinement cuts out too much gray matter in regions where the gray matter has rather low intensity in T2 images. I have observed this behavior quite often in my subjects. I have uploaded the example subject cut_gm_T2pial.tar.gz to the server. The problematic area has RAS location 3.9, -70, 0. Also see the screenshot. My command line was: mris_make_surfaces -nsigma_above 2 -nsigma_below 3 -aseg aseg.hires -filled filled.hires -wm wm.hires -mgz -sdir $my_subject_dir -orig white.deformed -nowhite -orig_white white.deformed -orig_pial pial.preT2.pass2 -T2 my_subject/mri/T2w_hires.norm -T1 T1w_hires.norm -output .postT2.pass2 my_subject rh and for FS 5.3 mris_make_surfaces -nsigma_above 2 -nsigma_below 3 -aseg aseg.hires -filled filled.hires -wm wm.hires -mgz -sdir $my_subject_dir -orig white.deformed -nowhite -orig_white white.deformed -orig_pial pial.preT2.pass2 -T2 my_subject/mri/T2w_hires.norm -T1 T1w_hires.norm -output .postT2.pass2.FS_5.3 my_subject rh Could you please comment on how to proceed? It seems that in contrast to v5.3 it is necessary to increase nsigma_below value since the nsigma values seem to be more "sensitive" in v6beta version. I tried to increase nsigma_below (nsigma_below for T2 in recon-all is 5) but it seemed to me that it leads to "leak" of the pial surface out of the gray matter in other areas, mainly in regions adjacent to cerebellum. See the example leaking the pial surface to the cerebellum in left hemisphere with nsigma_below 5 in the second screenshot in attachment ( I have uploaded this subject as file cut_gm_T2pial2.tar.gz, his cut-out gray matter RAS location is -33.7, -18, 45, command line is similar as above, but for left hemisphere). Yellow surface is with nsigma_below 5, blue is nsigma_below 3. Btw, the -aseg option in mris_make_surfaces does not prevent to leak surfaces into the cerebellum? Could also be the reason of these problems, that I used HCP scripts which do not do the T2 normalization with anatomy-informed mri_normalize (using aseg and white surface), but only global normalization and bias field correction? Regards, Antonin Skoch

2 1

Re: [Freesurfer] v6.0_beta - mris_make_surfaces -T2 refinement -issue with overlapping pial surfaces
by Antonin Skoch 24 Nov '16

24 Nov '16

Bruce, and do you require that also all volumes which are passed as arguments to mris_make_surfaces must have identical vox2ras and also the vox2ras of the volumes has to be identical to vox2ras of all surfaces? And, still it is not clear to me this: In freeview, there is perfect overlap of ?h.white.deformed and lh.white, but for the selected vertex the freeview shows different coordinates in lh.white and lh.white.defomed (see screenshot, note the perfect overlap of the meshes). My click on the surface in 3D view mode selects vertex no 66639 in both surfaces, located at the same position on the screen, however the vertex coordinates shown in freeview are different. How it is possible? Antonin From: Bruce Fischl <fischl(a)nmr.mgh.harvard.edu> To: Antonin Skoch <ansk(a)ikem.cz> Sent: 11/23/2016 2:26 AM Subject: Re: [Freesurfer] v6.0_beta - mris_make_surfaces -T2 refinement -issue with overlapping pial surfaces no, they look fine in freeview because freeview is treating them as sseparate surfaces. In mris_make_surfaces we are assuming it is the same surface (same ras2vox, same number of vertices/faces, etc...) On Wed, 23 Nov 2016, Antonin Skoch wrote: > Dear Bruce, > > I am trying to understand. The vox2ras are indeed different: > > lh.white > > Volume Geometry vox2ras > -0.70000 0.00000 0.00000 108.35732; > 0.00000 0.00000 0.70000 -131.16824; > 0.00000 -0.70000 0.00000 112.74670; > 0.00000 0.00000 0.00000 1.00000; > > lh.white.deformed > > Volume Geometry vox2ras > -0.70000 0.00000 0.00000 108.33944; > 0.00000 0.00000 0.70000 -127.46620; > 0.00000 -0.70000 0.00000 109.73940; > 0.00000 0.00000 0.00000 1.00000; > > But when I load both surfaces in freeview, they perfectly fit, optically there is perfect > vertex-by vertex match (by looking at the mesh rendering or 2D surface contour, even with > extended zoom). > However, vertices show different RAS coordinates in freeview cursor display, for example > vertex 136135 in lh.white.deformed shows coords -1.36 -6.9 21.69, in lh.white shows > coords -1.38 -3.2 18.68. > > It is some kind of issue of freeview? > > Antonin > > > From: Bruce Fischl <fischl(a)nmr.mgh.harvard.edu> > To: Antonin Skoch <ansk(a)ikem.cz> > Cc: <freesurfer(a)nmr.mgh.harvard.edu> > Sent: 11/23/2016 1:43 AM > Subject: Re: [Freesurfer] v6.0_beta - mris_make_surfaces -T2 refinement -issue with > overlapping pial surfaces > > the vox2ras. Not sure about the HCP pipeline. They use a deprecated > version of FS, but in any case what you had would not work as the ?h.pial > surface didn't match the ?h.white.deformed one. If you also deformed > the pial it would be ok > > On Wed, 23 Nov 2016, Antonin > Skoch wrote: > > > Dear Bruce, > > > > thank you very much for your time. > > > > Concerning transformations, that is weird since I adopted my code from HCP > pipelines > > > >https://github.com/Washington-University/Pipelines/blob/master/FreeSurfer/s… > > > rferHiresPial.sh > > > > and there I think they use the surfaces precisely this way. They are using > the surfaces > > generated by 1mm3 volumes as input and refine them by using co-registered > higher > > resolution volumes. I use full-hires reconstruction, but for my safety and > convenience I > > left the code with transformations, which (I thought) should do not any > harm (there is > > not any resampling in my case). > > > > I also checked the white and white.deformed processed by default HCP > pipeline and they do > > NOT have the same ras2vox. > > > > BTW, which transformation in mris_info is relevant for this case? I found > several of them > > in mris_info: > > > > talairch.xfm > > > > surfaceRAS to talaraiched surfaceRAS > > > > talairached surfaceRAS to surfaceRAS > > > > Volume Geometry vox2ras > > > > Volume Geometry vox2ras-tkr > > > > Antonin > > > > > > From: Bruce Fischl <fischl(a)nmr.mgh.harvard.edu> > > To: Antonin Skoch <ansk(a)ikem.cz> > > Sent: 11/23/2016 12:42 AM > > Subject: Re: [Freesurfer] v6.0_beta - mris_make_surfaces -T2 refinement > -issue with > > overlapping pial surfaces > > > > yes, and I tracked it down and fixed it. BTW: don't use the > > white.deformed surface - use the white one. You can't use one surface > that > > has been transformed (i.e. has a different ras2vox) and other > surfaces that > > have not (you can see this if you run mris_info on the surface files, > the > > ras2vox should match in all of them) > > > > cheers > > Bruce > > > > > > > > > > > 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. > > >

2 2

Recon-all for Allen Brain Institute T1 image
by Michael Craig 24 Nov '16

24 Nov '16

Hi All, I am trying to run recon-all on the Allen Brain Institute T1 images. It works without any issue on all but one subject (H0351.1009). I saw this post from a few years ago describing the exact problem I’m having, but don’t see a solution. Has anyone figured out how to get this subject to run in recon-all? Or is it simply a matter of there not being enough contrast between grey and white matter? Thanks for the help! https://mail.nmr.mgh.harvard.edu/pipermail/freesurfer/2013-July/031482.html <https://mail.nmr.mgh.harvard.edu/pipermail/freesurfer/2013-July/031482.html> Best wishes, Mike

2 1

Reconstructing cerebellum in FreeSurfer 5.3
by Alexander Zent 24 Nov '16

24 Nov '16

Hi, I have been trying to figure out how to extract and reconstruct the cerebellum and possible the pons and brain stem. I have already ran recon-all and extracted the hemispheres but cannot seem to find any documentation on how to extract the other structures. I am running version 5.3 on Windows via VirtualBox. Thanks for you help. -Alex

2 1

Univariate_Algorithm does not converge after 50 iterations!!!
by Kasper Jessen 24 Nov '16

24 Nov '16

Dear FreeSurfer group, I am running the univariate analysis to test if there is a change in thickness between two groups over time. The dataset has a baseline and a follow-up scan (i.e., two time points) and consist of about 40 patients and 40 controls. I receive an an error which states "Algorithm does not converge after 50 iterations!!!" when I am running the following commands: test_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni); test_stats = lme_fit_FS(X,[2],Y(:,1)+Y(:,2),ni); Why am I getting this error? The qdec table looks as shown below and is loaded into Matlab by the following command: Qdec = fReadQdec('.../path/qdec/qdec.table.dat'); The fsid and actual numbers are changed for this forum post (but the original is similar). The thickness scores has been extracted using the QDEC (GUI). fsid fsid-base time group age gender hand_score lh_MeanThickness_thickness rh_MeanThickness_thickness 17a 17_base 0.000000000 1.000000000 29.000000000 1.000000000 90.000000000 2.686580000 2.490300000 17c 17_base 7.428512429 1.000000000 29.000000000 1.000000000 90.000000000 2.728190000 2.507870000 13a 13_base 0.000000000 1.000000000 34.000000000 0.000000000 60.000000000 2.418060000 2.214930000 13c 13_base 6.857134857 1.000000000 34.000000000 0.000000000 60.000000000 2.321630000 2.109870000 12a 12_base 0.000000000 0.000000000 21.000000000 0.000000000 100.000000000 2.569720000 2.439500000 12c 12_base 6.857134857 0.000000000 21.000000000 0.000000000 100.000000000 2.651950000 2.444010000 53a 53_base 0.000000000 0.000000000 19.000000000 1.000000000 70.000000000 2.620390000 2.375250000 53c 53_base 13.55314571 0.000000000 19.000000000 1.000000000 70.000000000 2.606590000 2.381630000 ... I run the following commands: Qdec = rmQdecCol(Qdec,1); sID = Qdec(2:end,1); Qdec = rmQdecCol(Qdec,1); M = Qdec2num(Qdec); Y = M(:,6:7); M = M(:,1:5); [M,Y,ni] = sortData(M,1,Y,sID); X = [ones(length(M),1) M M(:,1).*M(:,2)]; When I then run the step below I get the following errors: A) test_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni); # Algorithm does not converge after 50 iterations!!! B) test_stats = lme_fit_FS(X,[1],Y(:,1)+Y(:,2),ni); # This works! C) test_stats = lme_fit_FS(X,[2],Y(:,1)+Y(:,2),ni); # Algorithm does not converge after 50 iterations!!! However, when I change the QDEC table, so time starts at 1 instead of 0, the error disappears for one of them but not the one with two random effects. A) test_stats = lme_fit_FS(X,[1 2],Y(:,1)+Y(:,2),ni); # Algorithm does not converge after 50 iterations!!! B) test_stats = lme_fit_FS(X,[1],Y(:,1)+Y(:,2),ni); # This works! C) test_stats = lme_fit_FS(X,[2],Y(:,1)+Y(:,2),ni); # This works! Why am I getting this error and how should I fix it? Best wishes Kasper ________________________________ Denne e-mail indeholder fortrolig information. Hvis du ikke er den rette modtager af denne e-mail eller hvis du modtager den ved en fejltagelse, beder vi dig venligst informere afsender om fejlen ved at bruge svarfunktionen. Samtidig bedes du slette e-mailen med det samme uden at videresende eller kopiere den.

1 0

OPPORTUNITY OF FULL-TIME POSTDOCTORAL FELLOWSHIP AT THE UNIVERSITY OF SÃO PAULO, BRAZIL
by Neuroimagem IPQHCFMUSP - LIM21 24 Nov '16

24 Nov '16

One postdoctoral position is available within the Laboratory of Psychiatric Neuroimaging (LIM21) (www.neuroimagem.org.br/) at the University of São Paulo Medical School, Brazil, directed by Geraldo F. Busatto, Ph.D. Candidates should send a message to neuroimagem.ipq(a)hc.fm.usp.br with his/her CV and a letter of support. Deadline: 06 Dec 2016 *JOB DESCRIPTION* We are looking for a full-time researcher with previous experience in the use of software for the analysis of neuroimaging data. This post-doctoral fellow will be requested to work mainly with data acquired using resting-state functional magnetic resonance imaging methods (rs-fMRI) to evaluate patterns of functional connectivity in the brain. In our project, funded by the *FUNDAÇÃO DE AMPARO À PESQUISA DO ESTADO DE SÃO PAULO, BRAZIL* (FAPESP), such brain imaging data is being acquired in groups of cognitively-intact elderly individuals, patients with Alzheimer´s disease (AD), and patients with mild cognitive impairment (MCI). Pre-processing steps will be carried out using the software Analysis of Functional NeuroImages (AFNI), while the actual processing of data will be conducted using Matlab-based programs. The work will also involve use of software to evaluate the diagnostic accuracy to distinguish between AD, MCI and healthy elderlies, generating multivariate classifiers based on machine-learning methods (support vector machine). With such strategy, we aim to identify which are the functional connectivity characteristics in the brain that are most critical for the progression from normal cognition to MCI and then to AD. The post-doctoral fellow should have knowledge about the acquisition and analysis of fMRI data and other neuroimaging modalities. We aim to bring an individual with abilities to carry out, independently, analyses and organization of computational brain imaging data, interaction with other researchers involved in the same project, and participating in the writing up of scientific papers. These characteristics are expected to boost the scientific productivity from the funded project, both quantitatively and qualitatively. The professional will also have the opportunity to interact with scientists who are at the forefront of technologies for amyloid imaging in the brain, and data will be combined with fMRI analyses in the same elderly subjects evaluated with those two imaging technologies. The post-doctoral fellow will have the opportunity to practice abilities in analysing data of a challenging scientifc project, and he/she will be expected to establish and use optimized processing pipelines for the analysis of rs-fMRI data and other neuroimaging modalities. *REQUIREMENTS* - Completed PhD in areas related to computer sciences, neuroscience, neurology or psychiatry - Research interest and brain imaging applied to research studies of brain aging - Previous experience in the usage of processing and statistical analysis methods for research studies applied to magnetic resonance imaging studies of the brain. best regard *Laboratory of Psychiatric Neuroimaging - HCFMUSP*

1 0

Trouble with Two Stage Longitudinal Analysis
by Timothy Hendrickson 23 Nov '16

23 Nov '16

FreeSurfer experts, I am attempting to use the two stage model for longitudinal analysis, however I keep receiving an error message that my design matrix is poorly scaled. I belive I have cross-checked my qdec, fsgd, design matrix, and contrast files and they all match so I do not think the problem is a mismatch. I am wondering if I made some sort of mistake in my script design or design matrix creation... Anyway here is the command line input: mri_glmfit --glmdir lh.AllSite.gender.group.Age.thickness.fwhm10.twoStage.glmdir --X DOSSDS.twoStage_AllSite_Gender_Group_Age.dat --y lh.testretest.thickness-pc1.stack.fwhm10.mgh --C age_AcctGroup_AcctSite_AcctGender_twoStage.DOSSDS.mtx --C DiffDiagnosis_ControlMore_AcctSite_AcctGender_AcctAge_twoStage.DOSSDS.mtx --C DiffDiagnosis_ControlMore_AgeSlope_AcctSite_AcctGender_twoStage.DOSSDS.mtx --C DiffGender_MaleMore_AcctSite_AcctDiagnosis_AcctAge_twoStage.DOSSDS.mtx --C DiffGender_MaleMore_AgeSlope_AcctSite_AcctDiagnosis_twoStage.DOSSDS.mtx --label lh.testretest.fsaverage.cortex.label --surf fsaverage lh The error message is attached within file AllSite_twoStageLongitudinal_DsgnMtxErr.txt Additionally I've included my analysis script, all timepoint qdec file, temporal average qdec file, fsgd file, and design matrix. -Tim Timothy Hendrickson Department of Psychiatry University of Minnesota Office: 612-624-6441 Mobile: 507-259-3434 (texts okay)

1 0

Request for patch for mri_convert, nifti1Read(): unsupported slice timing pattern 5 error
by Jennifer Barredo 23 Nov '16

23 Nov '16

Hello Freesurfer experts, I'm trying to run the TRACULA preprocessing pipeline using data in NIFTI-1 format, but am receiving the following error: "nifti1Read(): unsupported slice timing pattern 5" Several other users have run into this problem. In the archives, there are a few links to dead file transfer protocols for a patch for mri_convert that might solve my problem. May I please have a patch for mri_convert? I'm using freesurfer-Darwin-lion-stable-pub-v5.3.0 and OSX Yosemite. Thank you, Jennifer -- Jennifer Barredo, PhD Staff Scientist Providence VA Medical Center Research Affiliate Brown Institute for Brain Sciences jlbarredo(a)gmail.com Jennifer.Barredo(a)va.gov

1 0

Freeview -dti Display as Vectors Discrepancy
by Douglas Merkitch 23 Nov '16

23 Nov '16

Hello experts, I have come across what I think could be an inconsistency with how freeview interprets and/or displays dti data as vectors. Specifically, the x component seems to be inverted in freeview from freesurfer-Darwin-lion-stable-pub-v5.3.0 as compared to fslview (version 509) AND freeview from freesurfer-Darwin-OSX-stable-v6-beta-20161116-5037eae. Is this a known issue with a patch available? Any insight would be greatly appreciated. Thanks, Doug Douglas Merkitch Neurological Sciences Rush University Medical Center Phone: (312) 563-3853<tel:%28312%29%20563-3853> Fax: (312) 563-4660<tel:%28312%29%20563-4660> Email: Douglas_Merkitch(a)rush.edu<mailto:Douglas_Merkitch@rush.edu>

2 1

Re: [Freesurfer] mris_make_surfaces in longitudinal stream - explanation of parameters
by Antonin Skoch 23 Nov '16

23 Nov '16

Dear Martin, thank you for the explanations, and sorry, I missed your previous response. Just I am wondering, what is the reason of the shrinkage? Why not to start directly from pial surface of the base? And, I also found a binary mris_longitudinal_surfaces, which is, however, not used in recon-all. What is the significance of this binary? Antonin ------------------------------- Hi Antonin, max does not constrain thickness. The location of the max is constrained to be within 2*max for every iteration, so it will constrain how far outwards it searches. But it will be done at each scale, so it can still deform a long way. We initialize the surface processing in long with the surface from the base. But we don’t use the pial from base directly, instead we shrink it a little first and start from there. Not sure about the last question (difference between orig and prig-white parameters). Best, Martin > On Nov 14, 2016, at 10:17 PM, Antonin Skoch <ansk at ikem.cz> wrote: > > Dear experts, > > I am trying to understand how the pial and white surfaces are generated in longitudinal stream using information from the reconstructed base template (freeSurfer development version). > > In longitudinal stream, apart from the -orig, -orig_pial and -orig_white there are also parameters -long and -max 3.5. > > May I ask what is the significance of these parameters? > > -max 3.5 means constraining cortical thickness to 3.5 mm? Why this constraint is applied specifically in longitudinal stream? > > In mris_make_surfaces.c I found that -long means that initial pial vertex positions are set between final white and orig pial, slightly inside orig_pial. What is the significance of this setting in longitudinal stream? > > Last question, not related specifically to longitudinal stream: What is the difference between -orig and -orig_white parameters of mris_make_surfaces? From the reading of source codes (quite tough to me) I got an impression that from -orig the gray and white matter histogram values are computed whereas -orig_white defines initial position of the vertices in the first iteration (which, for some reasons, does not necessary has to be identical to -orig). Am I correct? > > Regards, > > Antonin Skoch > _______________________________________________ > Freesurfer mailing list > Freesurfer at nmr.mgh.harvard.edu > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

3 2

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