: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] Clusterwise Correction for Multiple Comparisons
Can you send your full mri_glmfit-sim command line? It looks like you used a
cluster forming threshold (CFT) of 1.3. If so, when you look at the uncorrected
sig.mgh files and threshold at
Can you send your full mri_glmfit-sim command line? It looks like you
used a cluster forming threshold (CFT) of 1.3. If so, when you look at
the uncorrected sig.mgh files and threshold at 1.3, do you see the same
thing? Also, what is the threshold for the image below
On 3/1/2021 1:35 PM, Weera
Hi,
After running mri_glmfit-sim (two classes) and
freeview -f
$SUBJECTS_DIR/fsaverage/surf/rh.inflated:overlay=cache.th13.pos.sig.cluster.mgh
I get this:
I would like to know if this due to an error (do I need to change threshold) or
no significant clusters.
[cid:fdfe3765-85b2-46f6-a3b5-40fa
If it is only one space, then you can just leave off --3spaces
On 9/3/19 1:48 PM, Gwang-Won Kim wrote:
>
> External Email - Use Caution
>
> Hi there,
> To perform a cluster-wise correction for multiple comparisons, I ran
> monte carlo simulation as follows: mri_glmfit-sim --glmdir g1v2.wl
External Email - Use CautionHi there, To perform a cluster-wise correction for multiple comparisons, I ran monte carlo simulation as follows: mri_glmfit-sim --glmdir g1v2.wls --3spaces --cache 2 abs I know that "- -3spaces" means adjusting p-values for both hemispheres and subcortic
I've put a new version of mri_glmfit-sim here
https://gate.nmr.mgh.harvard.edu/safelinks/greve/mri_glmfit-sim
This should run, though I'm not 100% sure this is the right way to do it.
Let me know how it goes
On 3/19/19 10:58 AM, Victor Montal wrote:
> External Email - Use Caution
>
> Dea
External Email - Use Caution
Dear Freesurfer experts,
I am very interested in using your permutation approach in order to
correct our uncorr analyses using the newly proposed approach
(https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/MultipleComparisonsV6.0Perm).
However, I
Hello Douglas,
What do you mean by not doing the adjustement for 1-tailed test ? Does this
mean specify the sign as "abs" ?
I don't understand well how -log10(p) could be negative with 0 a écrit :
> I think it should work with a t as input. When you set the threshold, put
> in the t-value you w
I think it should work with a t as input. When you set the threshold,
put in the t-value you want, not a -log10(p). Also, don't do the
adjustment for 1-tailed test because then it will interpret the input as
-log10(p).
On 5/29/15 10:46 AM, Matthieu Vanhoutte wrote:
Dear experts,
I would like
Dear experts,
I would like to use the clusterwise correction specified in the group
analysis FreeSurfer wiki (
https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/GroupAnalysis). In
this way, I would use mri_surfcluster with the known FWHM of my GLM
analysis and my input is an *uncorrected Tmap*.
Thank you Douglas for this detailed answer.
Matthieu
2015-04-15 15:19 GMT+02:00 Douglas Greve :
> Sorry, I misread your previous email as "Indeed BPM does do Clusterwise
> correction." instead of "Indeed BPM doesn't do Clusterwise correction." The
> FWHM is the full-width-half-max and is a meas
Sorry, I misread your previous email as "Indeed BPM does do Clusterwise
correction." instead of "Indeed BPM doesn't do Clusterwise correction."
The FWHM is the full-width-half-max and is a measure of the spatial
smoothness. This is always needed for cluster-wise correction. This is
usually obta
Hello Douglas,
Sorry if I haven't been clear. I'm trying to explain my needs :
1) I use the BPM toolbox to compute the GLM analysis as mri_glmfit would
do. However, outputs of this toolbox only give me beta maps and uncorrected
Tmaps.
2) With only these uncorrected Tmaps outputs, I'd like to appl
I guess I don't understand what you are asking for. If BPM outputs a
cluster-corrected volume, then why can't you just map that to the
surface and be done with it?
On 4/14/15 6:48 PM, Matthieu Vanhoutte wrote:
Hello Douglas,
Indeed BPM doesn't do Clusterwise correction.
What does the FWHM o
Hello Douglas,
Indeed BPM doesn't do Clusterwise correction.
What does the FWHM of the BPM analysis correspond to ? And how to find it ?
Manu thanks !
Best regards,
Matthieu
Le 15 avr. 2015 00:23, "Douglas Greve" a écrit :
> So BPM does not do clusterwise correction? You could use mri_volcl
So BPM does not do clusterwise correction? You could use mri_volcluster,
but you'll need to know the FWHM of your BPM analysis
doug
On 4/14/15 12:17 PM, Matthieu Vanhoutte wrote:
Dear FreeSurfer's experts,
For personnal need I use the BPM toolbox (Biological Parametric
Mapping) to compute GLM
Dear FreeSurfer's experts,
For personnal need I use the BPM toolbox (Biological Parametric Mapping) to
compute GLM analysis. The outputs of this Toolbox gave me beta maps and
Tmaps.
However, precedently I used the whole FreeSurfer group analysis tutorial on
other datas, including the Clusterwise
Thank you Douglas !
Matthieu
2015-04-13 22:13 GMT+02:00 Douglas N Greve :
> clusterwise correction is a type of FWE correction. If the FWE you are
> using is good enough for you and you want to visualize the results on a
> surface, then use mri_vol2surf. If this was done on the SPM template
> sp
clusterwise correction is a type of FWE correction. If the FWE you are
using is good enough for you and you want to visualize the results on a
surface, then use mri_vol2surf. If this was done on the SPM template
space, then run the SPM T1 template through recon-all to get the surfaces
On 0
By the mean of SPM, the Biological Parmatring toolbox can do the FWE
correction and give me results on a glass brain. But I don't have any
cluster wise corrected P-maps on output.
Or did I miss something ?
Best regards,
Matthieu
2015-04-13 16:51 GMT+02:00 Douglas Greve :
> Does the BPM toolbo
Does the BPM toolbox (what's that?) not give you cluster correction? You
can try using mri_volcluster with the --grf option, but you'll need to
know the FWHM. Afterwards you can map the result to the surface
On 4/13/15 10:46 AM, Matthieu Vanhoutte wrote:
Hello Douglas,
Thanks but even if I do
Hello Douglas,
Thanks but even if I do the clusterwise correction in the volume, I can't
get an clusterwise corrected output volume image. The only things I have
are the Tmaps that I can threshold if wanted.
And I would like to project on fsaverage surface the results of the
clusterwise correctio
If you did the analysis in the volume, then you have to do the
clusterwise correction in the volume.
On 4/13/15 6:05 AM, Matthieu Vanhoutte wrote:
Dear experts,
I had to make volumic statistical analysis for the need to use BPM
toolbox. So, I got in the output of this statistical analysis som
Dear experts,
I had to make volumic statistical analysis for the need to use BPM toolbox.
So, I got in the output of this statistical analysis some volumic Tmaps.
As I did before some group analysis according to FStutorial, I would like
to apply the same method of Clusterwise Correction on my vol
Hi, Doug and all,
I have two quesiotns about clusterwise correction for multiple comparsion.
I run 2 corrections,and commands are as follows:
a. mri_glmfit-sim --glmdir g2v0.weight --cache 1.3 abs --cwpvalthresh .05
--2spaces
b. mri_glmfit-sim --glmdir g2v0.weight --cache 1.3 pos --cwpvalthresh .
It does not seem like it should take that much time. You might try this
program
ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/mri_mcsim.linux
This is what I use to create the cached results that we distribute. You
can use the cached results unless you have a different mask. If y
Hi,
I'm performing a clusterwise correction for multiple comparisons
with the command
mri_glmfit-sim --glmdir my_glmdir --sim mc-z 5000 1.3 basename \
--sim-sign abs
For version v4.5.0 this took about 40 hours to complete.
However, for v5.1.0 it is now running for more than
6 days and from the c
When you get p-values (sigs) out of FreeSurfer, the p-value is computed
assuming an unsigned test. This makes the p-value twice that of a signed
test (ie, it is less significant than the signed test). When you go to
correct for multiple comparisons and you specify a sign (positive in you
case),
That worked, didn't work out that good for the significance of my data
though... :-(
Thanks, Doug & Mike!
On Wed, Mar 2, 2011 at 1:19 PM, Michael Waskom wrote:
> Hi Pieter,
>
> I think I may have run into this problem before by using mri_glmfit on
> a surface dataset and forgetting to to incl
Hi Pieter,
I think I may have run into this problem before by using mri_glmfit on
a surface dataset and forgetting to to include --surf
. The model fitting runs fine, but you end up with confusing
problems at the corrections stage because the residual smoothness
estimation doesn't work properly.
Hi Pieter, something has gone wrong, but I'm not sure what. The
smoothing level in the CSD is nan (not-a-number). Can you tar up the
glmdir and drop it to me at the file drop address at the end of this email?
doug
Pieter van de Vijver wrote:
>
> /Sorry if I'm posting double, I had some trouble
Hi all,
I used QDEC compare differences between normal subjects and patients,
the activation can be seen without correction, after correction by FDR ,all
disappeared.I read about the Clusterwise Correction for Multiple Comparisons of
the manual. I am confused about the meaning of C
Looks like it did not find any vertices above the threshold that you
used for the simulation (though it should give a better error msg).
Before you run the simulation, you can look at the map at your given
threshold. If there are no vertices above threshold, then nothing is
going to show up in
I am trying to do the clusterwise correction for multiple comparisons and came
upon a problem that maybe someone can help me with.
ri_volcluster --in brianweisinger/g1g2.intercept/sig.mgh --csd
brianweisinger/csd/mc-z.negative.j001-g1g2.intercept.csd --mask
brianweisinger/mask.mgh --cwsig
bri
What version do you have? I've put a copy of it here:
ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/stem2fname
but the fact that it does not exist in your distribution could be an
indication that something else is wrong.
doug
Sharmili 57 wrote:
> Dear all,
>
> i want to run clus
Dear all,
i want to run clusterwise correction for group comparison. I use the following
command.
mri_glmfit-sim \
--glmdir
/data/Psychiatrie/01_Projects_Folder/DA_Sritharan/Freesurfer/qdec/averagethichnessdiagnosisFWHM15rh
\
--sim mc-z 1 2 mc-z.abs.2 \
--sim-sign abs \
--overwrite
I
Yes, when you are running the simulation you should pass it the fwhm
(new versions *force* you to pass it a fwhm when running the simulation,
even if it is 0). And, yes, if you do not give it the proper fwhm (or
pass 0), then it will make it more likely that clusters will be deemed
significant.
Thanks, Doug. Just so I'm clear: passing mri_glmfit the value in
fwhm.dat is the *proper* thing to do, and not giving the simulation that
information on the residual smoothness would cause the p-value estimates
to be biased towards letting clusters in, yes?
-
Jim Porter
Graduate Student
The cluster sizes/locations won't change because those are dependent on
your actual data, not the simulation. If you change the smoothing level
in the QDEC analysis, you will find that the cluster sizes/locations
change. The simulation is only for computing the pvalues (CWP) of the
clusters, an
Hi Doug-
You can see the full logs & a screenshot of the mgh files here:
https://netfiles.umn.edu/xythoswfs/webui/_xy-10431983_1-t_4xQ4HIVC
Below are the commands (full pathways removed for brevity) that I ran.
-Jim
# mri glmfit command & fwhm listing from original qdec log
mri_glmfit --y
It should not be doing that. The color table is embedded in the annot.
Can you describe it further? There's not a good way to load only the
significant clusters, but you can load the cluster significance file as
an overlay and set the threshold to whatever you want.
doug
Dankner, Nathan (NIH/
Another question: I've noticed that when loading the cluster .annot file,
tksurfer also loads the desikan parcellation color table. How can I load the
clusters by themselves? And also, I know that loading the cluster overlay
allows you to see which clusters are significant by looking at thei
If you are running an MC simulation (not perm) directly from mri_glmfit
(and not mri_glmfit-sim) then you must pass it a fwhm or the results
will be wrong. The smoothing level is measured by QDEC (mri_glmfit) as
part of the analysis. You can't run the simulation from QDEC, so
specifying the sm
Can you send your mri_glmfit and mri_surfcluster command-lines?
doug
James Porter wrote:
> Sorry, I mistyped. To clarify, the original qdec analyses were run on
> the 10mm smoothed data from qcache, and the sims were then run with
> and without passing the value from fwhm.dat into the FWHM flag
Doug,
Thanks for the info. One quick follow-up question: Let's say for example the
smoothing before I run the simulation is 2.75. When you say I need to match
the smoothing, do I achieve that by adding 2.75mm of smoothness to the data
before running the cluster simulation? In the qdec GUI i
Sorry, I mistyped. To clarify, the original qdec analyses were run on
the 10mm smoothed data from qcache, and the sims were then run with and
without passing the value from fwhm.dat into the FWHM flag in mri_glmfit.
-
Jim Porter
Graduate Student
Clinical Science & Psychopathology Researc
I have recently run simulations with and without the FWHM flag, and the
results were 100% identical, in both the mgh output and cluster summary
files. From what is below, I would think that the results should have
changed in at least some way. The original qdec analysis was run with
10mm smooth
1) The smoother the data, the more likely a cluster will be found by
chance. When the data are created, they start with some smoothness
level. When you smooth them you add more. So you need to match the total
level of smoothing when you do the simulations, otherwise your clusters
will be way to
Hello all,
I have a couple of questions regarding the way the cluster correction
simulation in freesurfer works. I've read the wiki pages on the subject, but
if I've missed something and any of this is answered elsewhere please let me
know. My technical knowledge of these things is not great
The input is already specified as the --glmdir
doug
Rosa Steimke wrote:
> HI!
> i have a question regarding multiple comparion correction in qdec. can i use
> a clusterwise correction?
>
> i found the following script for glm:
>
> mri_glmfit-sim \
> --glmdir lh.gender_age.glmdir \
> --sim mc
Rosa,
If you have the newest version of qdec, it has a button to generate the
mri_glmfit-sim script for you, with the proper directory input and
output generated for you.
You'll still have to load the results on your own in tksurfer by
following these instructions:
http://surfer.nmr.mgh.harvard.
HI!
i have a question regarding multiple comparion correction in qdec. can i use a
clusterwise correction?
i found the following script for glm:
mri_glmfit-sim \
--glmdir lh.gender_age.glmdir \
--sim mc-z 5 4 mc-z.negative \
--sim-sign neg \
--overwrite
but i dont understand how i shoul
Thank you for your help. As regards to the display, I am using the command
that is found in the mri_glmfit-sim documentation. Even with that command I
don't see the display as it is shown in the documentation. In addition to
showing the clusters (all of them), the display also shows what looks l
It is for the whole brain. Not sure what you display command is, but
mri_surfcluster will save all clusters in the cwp map. You can then set
the tksurfer threshold to exclude the ones below your target sig level.
doug
Devdutta W wrote:
Hi,
In the mri_glmfit-sim command, when it does a "cluste
Hi,
In the mri_glmfit-sim command, when it does a "clusterwise correction" is it
correcting across the whole brain? In other words, does the term
"clusterwise" mean that it is only correcting within that cluster or is it
correcting across the whole brain, cluster-by-cluster (instead of, say,
voxel
55 matches
Mail list logo
