Freesurfer

freesurfer@nmr.mgh.harvard.edu

16 participants
30098 discussions

improving white/grey segmentation in recon -all for better volume estmation
by Ramtilak Gattu 24 Apr '16

24 Apr '16

Hello Bruce, Thanks a lot for your earlier expert option suggestion in improving the segmentation during the recon process. I am attaching a snap shot where yellow is with expert options and red is with default options. My concern is where I have indicated with blue arrows . Could you please suggest why is it still failing to pick up those areas. Is it because of a) misregistration with the atlas (i) if so where can I find the results after the non linear registration and (ii) would it be possible to improve with any more expert options during the registration and the segmentation subsequently. b) or is it should be just changing the lower limit values for the WM and higher limit values for the GM. I am not sure which way to tune the values should I go low on WM values and go high on GM values. These are the actual values mri_segment -wlo 87.7 -ghi 103.3 mris_make_surfaces -min_gray_at_white_border 79.4 -max_border_white 109.3 -min_border_white 90 -max_csf 68.7 -max_gray 98.7 -max_gray_at_csf_border 84.7 -min_gray_at_csf_border 58.1 For example Should these be changed to mri_segment -wlo 80 -ghi 110 Thanks for your help in advance. Regards Tilak Thanks Regards Tilak ________________________________________ From: Ramtilak Gattu Sent: Wednesday, March 30, 2016 4:26 PM To: freesurfer(a)nmr.mgh.harvard.edu Subject: Re: [Freesurfer] improving white/grey segmentation in recon -all for better volume estmation Thank you for recommending the expert options. I have run this on two subjects (rows) and you can see the post analysis results on the right (column) in the attached image . As you have mentioned it didnt work well in the cerebellum any other ways to improve in the inferior regions? Thanks Regards Tilak ________________________________________ From: Ramtilak Gattu Sent: Monday, March 28, 2016 5:12 PM To: freesurfer(a)nmr.mgh.harvard.edu Subject: Re: [Freesurfer] improving white/grey segmentation in recon -all for better volume estmation Thank you, Attached is the scripts folder for one subject . Thanks Regards Tilak ________________________________________ From: Ramtilak Gattu Sent: Monday, March 28, 2016 4:11 PM To: freesurfer(a)nmr.mgh.harvard.edu Subject: Re: [Freesurfer] improving white/grey segmentation in recon -all for better volume estmation Sorry, Actually just found out that the receiving coil is the 12 channel head coil and the transmitting coil is the body coil. I am attaching a snapshot showing the sag orientation of the three cases that I previously mentioned and uploaded for your review. >From these figures do you really think there is not enough contrast, just wondering 1) why it worked in the bottom image and it failed in the upper two (T1_sag_comparison.jpg) 2) it failed in the left image and did good in the right image (T1_sag_comparison_2.jpg). We have similar data in 25 of the subjects. Is there a way to salvage the existing data and improve our results we will certainly follow your input in changing the existing protocol and collect some data and run the analysis to see if it improves the results. One more thought is if is there any other way around to increase the contrast by taking help from T2 image and run recon-all. Do you think it will help? Thanks for you help . Regards Tilak ________________________________________ From: Bruce Fischl <fischl(a)nmr.mgh.harvard.edu> Sent: Monday, March 28, 2016 11:22 AM To: Ramtilak Gattu Subject: Re: [Freesurfer] improving white/grey segmentation in recon -all for better volume estmation what was the transmit coil? I think I would either reduce the voxel sizes a bit or not use grappa. Probably go to 1.2 mm or so, which would be a 70% increase in SNR. That won't help the transmit bias field though if you have one. The 8 channel can also sometimes be a problem. The 32 channel coils work much better On Mon, 28 Mar 2016, Ramtilak Gattu wrote: > Hey Bruce, > > The scans are done on 3T Siemens TRIO and 8 coil head channel with a GRAPPA of 2. > > Here is the screen shot of protocol used. > > https://waynestateprod-my.sharepoint.com/personal/ao3080_wayne_edu/_layouts… > > Thanks > > Regards > Tilak > > > > ________________________________________ > From: Ramtilak Gattu > Sent: Sunday, March 27, 2016 3:23 PM > To: Bruce Fischl > Subject: Re: [Freesurfer] improving white/grey segmentation in recon -all for better volume estmation > > Hey Bruce, > > I have uploaded the files in > > https://waynestateprod-my.sharepoint.com/personal/ao3080_wayne_edu/_layouts… > > I have also included the 001.mgz in there. > > Thanks > > Regards > Tilak > > ________________________________________ > From: Bruce Fischl <fischl(a)nmr.mgh.harvard.edu> > Sent: Saturday, March 26, 2016 8:35 PM > To: Ramtilak Gattu > Subject: Re: [Freesurfer] improving white/grey segmentation in recon -all for better volume estmation > > can you send me the rawavg.mgz from the subjects's mri file instead of > these dicoms (which I can't read) > On Sat, 26 Mar 2016, Ramtilak Gattu wrote: > >> >> Hey Bruce, >> >> >> Thanks for the quick reply. >> >> >> I have uploaded the cases in >> >> >> https://waynestateprod-my.sharepoint.com/personal/ao3080_wayne_edu/_layouts… >> xWlpwBQnNAhG8%3d&docid=0a72431d82b4d4b9b924cccbdfb7f34f8 >> >> >> The voxel dimensions are 1x1x1 mm. >> >> >> >> Thanks >> >> >> Regards >> >> Tilak >> >> >> >> ____________________________________________________________________________________________________________________________________________________ >> From: Ramtilak Gattu >> Sent: Friday, March 25, 2016 9:59 PM >> To: freesurfer(a)nmr.mgh.harvard.edu >> Cc: tilakgattu(a)hotmail.com >> Subject: improving white/grey segmentation in recon -all for better volume estmation >> >> Hello all, >> >> >> We have run recon -all on some healthy control subjects T1 images but we found out that the gray-white segmentation on half our subjects is >> bad and it leads to increased white matter volumes and decreased gray matter volumes. I have included a snap shot showing the difference in >> segmentation from bad to good for 3 cases >> >> https://waynestateprod-my.sharepoint.com/personal/ao3080_wayne_edu/_layouts… >> WPDE1klUw%3d&docid=0dd4546a8004f46578b7a2177b192806a >> >> . The parameter in the sequence are TE=4.92 ms, TR=19 ms, Flip angle =20 degrees. and use a snapshot one slice to show the good CNR. We are using >> Freesurfer version 5.3. >> >> >> Thanks >> >> >> Regards >> >> Tilak. >> >> >> > > > 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. > > > -- ***NOTICE*** The documents accompanying this email transmission contain confidential information belonging to the sender, which is legally privileged. The information is intended only for the use of the individual(s) or entity(ies) named above. If you are not the intended recipient, you are hereby notified that any disclosure, copying, distribution, or the taking of any action in reliance on the contents of this emailed information is strictly prohibited. If you have received this email in error, please immediately notify us by telephone at 248-277-3334 to arrange for return of the original documents to us. Thank you.

2 2

Postdoctoral Fellow in Machine Learning in Neuroimaging
by Thomas Yeo 23 Apr '16

23 Apr '16

Dear All, I am looking to hire a postdoc to develop machine learning algorithms for fMRI. Topics are flexible. Example topics include individual-subject brain parcellations, dynamic functional connectivity, meta-analysis, neural mass modeling, monkey-human homologies, 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

Coordinates of voxels in matlab
by Mohamed Abdelhack 23 Apr '16

23 Apr '16

Dear all, So I am doing some matlab based analysis where I have to align several fmri volumes. To load the data from Freesurfer I used f = MRIread('filename.nii') where I have f.vol is the voxel intensity series. How is it possible to extract the [x,y,z] coordinates of the voxels from that structure to match the registered volume space? Thank you, Best, Mohamed =)

1 0

hippo subfield segmentation new beta version (version 6)
by Barbara Kreilkamp 23 Apr '16

23 Apr '16

Dear all, I've been reading the manuscript on version 6 freesurfer hippocampal subfield segmentation "A computational atlas of the hippocampal formation using ex vivo, ultra-high resolution MRI: application to adaptive segmentation of in vivo MRI" Can you please tell me if the type of sequence (e.g. coronal T2FLAIR vs T2STIR) and voxel sizes of .4x.4x2mm^3 vs .4x.4x4mm^3 really make a difference for this multi-modal algorithm? I previously ran recon-all on 1mm isotropic T1-w scans all with similar contrast. It seems that the mesh fitting in version 6 is driven by the relation of the subfields to each other rather than by contrast or signal intensities; and also I notice the algorithm brings everything to isotropic space anyway. Is this correct? Do you see a major methodological issue when using T2FLAIR (.4x.4x4mm^3) and T2STIR (.4x.4x2mm^3) subfield volumes together? Thank you very much, Best wishes, Barbara P.s. as a side-note, I've not noticed any major differences in subfield segmentations visually, when comparing segmentations via these two different T2 scans.

2 2

Re: [Freesurfer] preproc-sess error opening tmp brain_mask
by Douglas N Greve 22 Apr '16

22 Apr '16

Is your $FSLDIR environment variable set? On 04/22/2016 02:09 PM, Mcnorgan, Christopher wrote: > That doesn't seem to be the problem. Attached are the file listings for the /usr/bin directory for fsl* for the working and not working computers. The symlinks for fsl binaries on both computers point to the same place, and I can run something like fslview on both computers. > ________________________________________ > From: Douglas N Greve <greve(a)nmr.mgh.harvard.edu> > Sent: Friday, April 22, 2016 1:54 PM > To: Mcnorgan, Christopher > Subject: Re: [Freesurfer] preproc-sess error opening tmp brain_mask > > Do you have FSL installed on the machines that work and not on the > machine that does not work? > > On 04/22/2016 01:49 PM, Mcnorgan, Christopher wrote: >> Hi Doug, >> >> >> Thanks for your generous time donation resolving this problem. There >> was a delay in getting this diagnostic information to you (which is >> itself diagnostic) because prior to your response, I retraced my steps >> preparing the data for running FS-FAST using my wiki checklist, but on >> a different computer. After rerunning mkanalysis-sess, >> mkcontrast-sess, double-checking my par-files and any other related >> files that needed to exist, I reran preproc-sess on FS_T1_501 without >> incident ... on a different machine. >> >> >> This suggested that either I made a mistake that I fixed on my >> do-over, or else that the error is machine-specific. I'm repeating the >> procedure on data for the same participant doing the same task >> collected on the same scanner at time point 2, so the subject in >> question is FS_T2_501. Again I come to the same failure on the machine >> that seems to be giving me problems. >> >> >> I've attached the dng.log file you requested. I hope it has some >> useful clues. If you need to know anything about the particular >> computer configuration, let me know, though two other machines that >> work are similarly configured (one has identical hardware and all are >> accessing the fMRI data stored on a remote server on a cifs share >> mounted at ~/ubfs/). >> >> >> Can you run >> >> /mkbrainmask-sess//-debug -s FS_T1_501 //-fsd bold/ |& tee dng.log >> >> and send me dng.log ? >> >> doug >> >> > -- > Douglas N. Greve, Ph.D. > MGH-NMR Center > greve(a)nmr.mgh.harvard.edu > Phone Number: 617-724-2358 > Fax: 617-726-7422 > > Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting > FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 > www.nmr.mgh.harvard.edu/facility/filedrop/index.html > Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/ > > > > 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. -- Douglas N. Greve, Ph.D. MGH-NMR Center greve(a)nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422 Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/

1 0

Re: [Freesurfer] Viewing eroded structure after -seg-erode 1
by Elijah Mak 22 Apr '16

22 Apr '16

Hi Doug, I have pasted the previous emails below. I have computed mri_segstats to extract the FA and MD from resampled aparc+aseg.mgz in diffusion space. -seg-erode 1 was performed during mri_segstats. Now, I will like to visually inspect the results of that erosion for the left and right hippocampal structures. Is there anyway I can do that and compare that against the uneroded hippocampus? I don't think mri_segstats -seg-erode 1 produce any volumetric outputs so I am not sure. Thanks. Best Wishes, Elijah Douglas N Greve <http://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=from:…> Mon, 11 Apr 2016 08:11:06 -0700 <http://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=date:…> You can load it as an overlay in freeview make sure to specify the registration file as that compted for the DTI. Also make sure to set the threshold to something like 0.5 On 04/11/2016 09:41 AM, Elijah Mak wrote: > Hi Freesurfer Community, > > I have performed -seg-erode 1 during mri_segstats on resampled > subcortical strutures (T1 to FA/MD). Can I view the eroded masks in T1 > anatomical space to visually inspect whether PVE has been adequately > accounted for? > > Thank you. > > Best Wishes, > Elijah

2 1

Comparison of the slope of the pvr analysis between groups
by Sophie Maingault 22 Apr '16

22 Apr '16

Dear FS experts, There is already a post about that but the concerned person never answered so there is no solution. I’m currently running per-vertex analyses with the --pvr flag of mri_glmfit command. I did correlations between 2 vertex-wise variables correcting for age. I also would like to compare the slope of the pvr between 2 groups: in my fsgd file I have the subject ID and the group. I run the following command: mri_glmfit --y lh.1.448tvs.mgh --fsgd 448_cate30.fsgd --C contrast1.mtx --surf fsaverage lh --cortex --glmdir glm.lh.12_448tvs --pvr lh.2.448tvs.mgh with contrast1.mtx = 0 0 -1 1. I get the following error: ERROR: dimension mismatch between X and contrast contrast1.mtx X has 3 cols, C has 4 cols Do you know where is my mistake and if it possible to do comparison of the slope of a pvr analysis between groups with mir_glmfit ? Thank you for your help, Best regards, Sophie

2 1

Qdec stats question
by Ajay Kurani 22 Apr '16

22 Apr '16

Hi Freesurfer Experts, I have used freesurfer smoothed images (lh.thickness.fsaverage.fwhm15.gii) to perform a cortical thickness analysis. I used qdec with a surface-based morphometric analysis to compare two groups. As my discrete fixed factor I used group membership. As Nuisance factors I used age (years), sex (0/1), site(1-10) and years of education(years). I used Threshold min=2, max=5 which I assume refers to a T-stat threshold. I see an FDR correction which and the Monte Carlo Null-Z simulation. 1) Do I run both monte carlo and fdr or just one after my figure apprears? Is the monte-carlo with threshold = 1.3(0.5) similar to a clusterwise threshold with a fixed p value? Are results valid when just setting the threshold min=2, max=5 but they are just uncorrected results? 2) When running Qdec with no nusiance factors and just a discrete fixed factor of group membership, my results look almost identical to AFNI's 3dttest++ output when the tmin=2. However, when I add the nuisance factors the results between the two softare packages are different. When nuisances covariates are added does Qdec still use a ttest or switch to some other method? How is centering assumed between groups for covariances? Is the mean from each covariate calculated and removed from each group separately based on group membership or calculated together with all subjects and removed? Does the analysis assume to have pooled or unpooled variance across the two groups? 3) For the viewer it displaces a logarithmic scale from what I have read in other papers. Is there a way to switch this to displace tvalue? 4) There was a previous post I saw yesterday saying qdec only uses 2 nuisance covariates and so I wanted to see if this is the case or not so I can try to understand why my afni and freesurfer results match with no covariates added but are very different once covariates are added. Thanks, Ajay

2 3

Label registeration
by Mohamed Abdelhack 22 Apr '16

22 Apr '16

Dear all, I have been trying so far to export Freesurfer labels into a volume created by SPM I tried to coregister the anatomical data from Freesurfer and SPM and use the registeration file to turn my label file into volume but the resulting output just goes outside the brain. Has anyone tried to do that before that can advise me? Thank you, Best, Mohamed

2 1

mri_convert doubling file size
by Afzal, Afsana 22 Apr '16

22 Apr '16

Hi everyone, I'm using mri_convert to reinforce TR information into the Nifti header in the following manner: mri_convert $FILE $FILE -tr $TR Before the conversion, my file size is usually around 550MB but after the conversion is doubles in size to ~1.1GB. In the Nifto file header I've noticed that the data type usually changes from int16 to float32. Is there a quick way to decrease the file size or revert back to int16 without affecting the data? Thanks for you help, Afsana ______________________________ Afsana Afzal Clinical Research Coordinator Massachusetts General Hospital Division of Neurotherapeutics Department of Psychiatry: Neurosciences 149 13th St, Room 2612 Charlestown, MA 02129 Phone: 617-643-5129 Fax: 617-726-4078

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