Freesurfer

freesurfer@nmr.mgh.harvard.edu

16 participants
30084 discussions

The uncertainty of final stats from repeated runs on the same subject
by leemon albert 20 Oct '09

20 Oct '09

Hi, We run the freesurfer stable v4.4.0 on our Linux computer (Ubuntu 9.04). The program finishes without any error. However, we find the stats files (in the /subject/stats) from repeated runs on the same subject vary widely in final outcome. Meanwhile, we notice the item of the stable v4.5.0 release notes, which's about the flag of -norandomness. It's said that the binaries in the recon-all stream are subject to variability due to usage of a random number generator (mris_smooth, mris_sphere, mris_topology_fixer, mris_topo_fixer and mris_ca_label). Is this realtive with our problem? Have anyone met this problem? Do the data need to rerun using Stable v4.5.0? To take into account the number of the subjects, is there any other method to fix it? For example: tail -2 'thickness.lh.MT.songjunyi.txt' # ColHeaders StructName NumVert SurfArea GrayVol ThickAvg ThickStd MeanCurv GausCurv FoldInd CurvInd lh.MT.songjunyi.label 2330 1564 4584 2.548 0.684 0.146 0.062 52 6.5 tail -2 thickness.lh.MT.songjunyi_re1.txt # ColHeaders StructName NumVert SurfArea GrayVol ThickAvg ThickStd MeanCurv GausCurv FoldInd CurvInd lh.MT.songjunyi_re1.label 2409 1631 4786 2.539 0.698 0.146 0.075 78 7.7 tail -2 thickness.lh.MT.songjunyi_re2.txt # ColHeaders StructName NumVert SurfArea GrayVol ThickAvg ThickStd MeanCurv GausCurv FoldInd CurvInd lh.MT.songjunyi_re2.label 2341 1576 4659 2.558 0.694 0.143 0.058 39 6.0 Any help would be appreciated. Thanks.

2 1

Re: [Freesurfer] FDR correction (fwd)
by Doug Greve 19 Oct '09

19 Oct '09

Here's Tom Nichols explanation. He points out that you use the maximum i, not the minimum, so you don't have to get past i=1. Thanks Tom! ---------- Forwarded message ---------- Date: Fri, 16 Oct 2009 23:11:09 +0100 From: Thomas Nichols <t.e.nichols(a)warwick.ac.uk> To: Doug Greve <greve(a)nmr.mgh.harvard.edu> Cc: Donna Dierker <donna(a)brainvis.wustl.edu> Subject: Re: [Freesurfer] FDR correction (fwd) Dear Doug, Great discussion! But I think it misses the point. Can you forward the following to the list? The Benjamini-Hochberg (BH) FDR procedure uses the inequality P(i) <= i / V * q for the ordered P-values P(1)<=P(2)<=...<=P(V) and desired level-q FDR control, which suggests that one will never get significance unless P(1)<q/V, just like Bonferroni. HOWEVER, the key element of BH-FDR is that you get to search over i and find the *largest* P-value that satisfies the inequality and use that P-value as a threshold. Hence when considering changing resolutions (i.e. increasing V, but increasing resolution/smoothness and roughly keeping the information content of the images the same) the magic of BH-FDR is that if the 5-th %ile of uncorrected P-values is FDR-significant at one resolution, i.e. i' = 0.05*V satisfies P( i' / V ) <= i' / V * q then one would expect that the 5-th %ile of P-values after up-sampling would have a similar value, and thus would also satisfy the inequality even though V has changed. That is the motivation that FDR is resilient to changes in resolution. However, I should note that in my own experience, leaving V fixed but changing smoothness, often changes the distribution of P-values dramatically, and thus changes the FDR result. But that should be a different beast that up-interpolation. Hope this helps! -Tom On Fri, Oct 16, 2009 at 5:43 PM, Doug Greve <greve(a)nmr.mgh.harvard.edu>wrote: > > Tom, here's an FDR question for you. It appears that the FDR > correction is dependent on the number of voxels (need p < fdr/N just > to get past i=1). Meaning that as N grows, the min p-value must also > shrink to get past i=1. Any way to get around this? > > thanks > > doug > > > > > ---------- Forwarded message ---------- > Date: Fri, 16 Oct 2009 11:30:59 -0500 > From: Donna Dierker <donna(a)brainvis.wustl.edu> > To: Michael Harms <mharms(a)conte.wustl.edu> > Cc: freesurfer(a)nmr.mgh.harvard.edu, Yulia WORBE <yulia.worbe(a)upmc.fr> > Subject: Re: [Freesurfer] FDR correction > > Regardless: FDR's sensitivity appears resolution-dependent to me. > > On 10/16/2009 10:39 AM, Michael Harms wrote: > >> Interesting post Donna, but my understanding of FDR is that it sets the >> p-value threshold based on the LARGEST p-value that satisfies the FDR >> relationship. >> >> That is, steps 3 and 4 in Genovese et al. (2002) are: >> 3) Let r be the largest i for which p <= i/V*q (assuming c=1) >> 4) Threshold the image at the p-value p(r). >> >> So, it isn't the case that you require the most significant p-value to >> satisfy p <= 0.05/V "just to get past i=1" as you put it in your post. >> >> Rather, you pick the largest p-value that satisfies the relationship, >> meaning that lower (more-significant) p-values may not have necessarily >> satisfied p <= i/V*q for their particular position in the sorted list of >> p-values. >> >> cheers, >> Mike H. >> >> >> On Fri, 2009-10-16 at 10:13 -0500, Donna Dierker wrote: >> >> I never heard anything on my post here, but it might just be high >>> surface resolution: >>> >>> >>> http://www.mail-archive.com/neuro-mult-comp@brainvis.wustl.edu/msg00026.html >>> >>> On 10/16/2009 09:58 AM, Michael Harms wrote: >>> >>> Your FDR analysis sounds correct. You probably have a rather small >>>> number of "marginally" significant vertices, which is why none survive >>>> FDR. You could try increasing the "q" value from say 0.05 to 0.1, in >>>> which case 10% of the surviving vertices would be expected to be false >>>> positives. >>>> >>>> cheers, >>>> Mike H. >>>> >>>> On Fri, 2009-10-16 at 12:03 +0200, Yulia WORBE wrote: >>>> >>>> >>>> Dear Freesurfer team, >>>>> >>>>> We are currently doing a cortical thickness studies between a group of >>>>> psychiatric patients (n=60) and controls (n=30). We tested several >>>>> smoothing levels (15mm, 20mm, 25mm) >>>>> >>>>> When setting an uncorrected threshold (such as p<0.005), we obtained >>>>> several regions of decreased thickness, which are consistent with the >>>>> pathology. >>>>> >>>>> However, when trying to correct for multiple comparisons using FDR >>>>> ("Set Using FDR" button in qdec), the computed threshold is very high >>>>> (e.g. 4.3 for 20mm smoothing) and, obviously, no significant regions >>>>> are left. >>>>> >>>>> Did we do anything wrong in the analysis ? >>>>> >>>>> Thank you very much for your help, >>>>> Yulia >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> Freesurfer mailing list >>>>> Freesurfer(a)nmr.mgh.harvard.edu >>>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >>>>> >>>>> >>>>> _______________________________________________ >>>> Freesurfer mailing list >>>> Freesurfer(a)nmr.mgh.harvard.edu >>>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >>>> >>>> >>>> >> _______________________________________________ >> Freesurfer mailing list >> Freesurfer(a)nmr.mgh.harvard.edu >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >> >> > _______________________________________________ > Freesurfer mailing list > Freesurfer(a)nmr.mgh.harvard.edu > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer > > > > -- ____________________________________________ Thomas Nichols, PhD Head of Neuroimaging Statistics Applied Neuroimaging Group University of Warwick Senior Research Fellow Oxford University FMRIB Centre

3 4

dti and bvecs
by Alina Jurcoane 19 Oct '09

19 Oct '09

Hi, I have a Siemens DTI dataset which I am converting with mri_convert (freesurfer) from dicom to nifti and then fitting the tensors with FSL. Since mri_convert outputs my data in Neurological orientation, despite of fsl's warnings about flipping, I have to do fslswapdim and fslorient -swaporient in order to see my data properly in fsl. The problem is that tensors are fitted wrongly unless one flips the bvecs (outputted by freesurfer) as well. Is there any simple way to do the whole process dicom->nifti->fit without having to flip data and bvecs? For instance, is there some option to tell mri_convert to output my data AND gradient information in Radiological orientation? Thanks for the help! Alina

2 1

A questions on the calculation of LGI
by Zhangyuanchao 19 Oct '09

19 Oct '09

Hi,Marie In the paper "A surface based approach to quantify local gyrification" on TMI. There is a discussion the relation between lGI and curvature-based method. I noticed the following discussion: "In the current study, all the GI measurements were computed on cortical surfaces based on images in their native space. Independently of intersubject registration, the absence spatial normalization ensures that the observed differences are due to a real decrease in cortical area (related to diminished frequency and depth of sulci) rather than to an effect of scaling." Given the above discussion, I wonder whether I should follow the freesurfer processing pipeline in generating the pial surface? If not, how should I manage to obtain the pial surface? since I noticed that in the freesurfer pipeline there is a step called Talairach Transformation. What is the effect of Talairach Transformation on lGI What is the effect of intersubject registration on lGI? When I use lGI to measure cortical folding, whether I should follow the freesurfer pipeline or not since there is intersubject registration in the pipeline? By the way, in your paper, the radius is 20 or 25, what is the exact default value of the radius in freesurfer? Thanks a lot! Best regards, Yuanchao __________________________________________________ 赶快注册雅虎超大容量免费邮箱? http://cn.mail.yahoo.com

2 1

How to smooth functional retinotopic data on the gray matter surface?
by Andreas Schindler 19 Oct '09

19 Oct '09

Hello I am trying to analyse my retinotopic data with freesurfer FS-FAST. Regarding the workflow, I found very helpful information on the freesurfer WIKI, except one preprocessing issue: Instead of smoothing my data in the volume I want to to smooth the retinotopic data on the surface before starting the actual analysis as fieldsign maps are also organised within the plane rather than the 3D volume. I already tried mri_vol2surf then smoothed my functional data following a re-transformation into volume data using mri_surf2vol. However, after processing the functional data with mri_vol2surf it leaves me with an image of several dimensions rather than 2 (as I expected), which I don't know how to smooth. So my question is if my approach is the right way to do functional surface smoothing (and if yes how to interpret / smooth the output of mri_vol2surf) or if there is a better solution for my problem I haven't found, yet. Thanks already in advance! best wishes Andreas

1 0

processing time for autorecon1 in Freesurfer v4.5
by Chris Hyatt 17 Oct '09

17 Oct '09

We recently installed a Dell Blade server cluster running Freesurfer v4.5 under CentOS 64-bit Linux, with each Blade server having two Intel Xeon E5405 quad core CPUs, running at 2.0 GHz, 2x6144 KB cache and 16 GB of RAM, with a swap file of size between 8-16 GB. Using a single Blade server of this cluster, a single subject takes between 70 and 100 minutes to complete the Freesurfer autorecon1 stage, which seems overly long. We also have another computer, running an older version of Freesurfer (v4.0.5) under Fedora 64-bit Linux. This computer is similar to a single Blade server - it has two Xeon X5482 quad core CPUs, running at 3.2 GHz, 2x6144 KB cache and also with 16 GB RAM with a swap file of 16 GB. This computer can complete autorecon1 stage in about 18 minutes. The only real difference with this computer and the Blade computers that I aware of is the clock speed (3.2 vs 2.0 GHz, 1.6x faster), so the Dell Blade computers should be able to complete autorecon1 in no worse than about 30 minutes. Does anyone have an idea as to why Freesurfer v4.5 is running so slow on our Dell Blade cluster? Has anyone found the newest version of Freesurfer (v4.5) slower than v4.0.5? Thanks in advance for your help.

5 5

FDR correction
by Yulia WORBE 17 Oct '09

17 Oct '09

Dear Freesurfer team, We are currently doing a cortical thickness studies between a group of psychiatric patients (n=60) and controls (n=30). We tested several smoothing levels (15mm, 20mm, 25mm) When setting an uncorrected threshold (such as p<0.005), we obtained several regions of decreased thickness, which are consistent with the pathology. However, when trying to correct for multiple comparisons using FDR ("Set Using FDR" button in qdec), the computed threshold is very high (e.g. 4.3 for 20mm smoothing) and, obviously, no significant regions are left. Did we do anything wrong in the analysis ? Thank you very much for your help, Yulia

6 7

Different TIV/ICV estimates
by DRC SPM 16 Oct '09

16 Oct '09

Dear FreeSurfers, I wanted just eTIV from a bunch of subjects, and it looked to me from an example log file that both old (with skull) and new (without) estimates would be available after autorecon1, since registration with and without skull seems to have been performed early on. However, I've noticed that the results of both old and new differ from one run of just -autorecon1 versus one of -all: $ mri_segstats --old-etiv-only --subject ALL atlas_icv (eTIV) = 1465825 mm^3 (det: 1.466751 ) $ mri_segstats --old-etiv-only --subject RECON1 atlas_icv (eTIV) = 1453815 mm^3 (det: 1.478867 ) $ mri_segstats --etiv-only --subject ALL atlas_icv (eTIV) = 1368231 mm^3 (det: 1.423813 ) $ mri_segstats --etiv-only --subject RECON1 atlas_icv (eTIV) = 1360078 mm^3 (det: 1.432349 ) The above are using FS v.4.5. I've also noticed that none of the above exactly reproduce some results that someone else in my lab previously found with FS v.4.03, though I'm not sure whether they changed some of the expert parameters/options or not... $ mri_segstats --old-etiv-only --subject v403 atlas_icv (eTIV) = 1449214 mm^3 (det: 1.483562 ) $ mri_segstats --etiv-only --subject v403 atlas_icv (eTIV) = 1366893 mm^3 (det: 1.425207 ) If parameters were the same, should I expect old-etiv-only to be consistent before and after v4.3 when the default changed, or did some other things relevant to eTIV change too? Finally, if I just want TIV estimates, but these might be more accurate after -all than -autorecon1, is there a simple combination of -autorecon1 with some extra directives that would reproduce those from -all without doing all of the really time-consuming cortical stuff? Many thanks, Ged

1 0

generate mask
by Daniel G. Wakeman 16 Oct '09

16 Oct '09

Hi, I need to generate a mask from a 'BEM' surface (10242 vertices). I have tried doing this utililzing mri_surf2vol; however, this doesn't produce a good mask as only the vertices are utilized creating a speckled volume. Is there any way to create a proper mask utilizing all of the surfaces of a surface? Thanks Dan

2 1

nu_correct or mri_nu_correct
by McLean John (NHS Greater Glasgow & Clyde) 16 Oct '09

16 Oct '09

> Hi, > > What's the difference between applying nu_correct and mri_nu_correct? > > Many thanks > John ******************************************************************************************************************** This message may contain confidential information. If you are not the intended recipient please inform the sender that you have received the message in error before deleting it. Please do not disclose, copy or distribute information in this e-mail or take any action in reliance on its contents: to do so is strictly prohibited and may be unlawful. Thank you for your co-operation. NHSmail is the secure email and directory service available for all NHS staff in England and Scotland NHSmail is approved for exchanging patient data and other sensitive information with NHSmail and GSI recipients NHSmail provides an email address for your career in the NHS and can be accessed anywhere For more information and to find out how you can switch, visit www.connectingforhealth.nhs.uk/nhsmail ********************************************************************************************************************

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