[Freesurfer] design matrix palm

[maaike rive]

Hi all,

I was adviced to use PALM for permutation of non-orthogonal surface data. 
However, I have some questions  regarding the design matrix en contrast files 
to be used in PALM, but somehow I cannot subscribe to the FSL mailing list, and 
I was hoping one of the freesurfer users could help me out.

I have a complex design and I would like to test for interactions as well for 
subgroups; in addition, I have several covariates. If I use the following 
collumns:

A1 B1 A2 B2 cov1 cov2 cov3
I assumed I can at least test for the effects of A vs B (1 - 1 1 -1 0 0 0) and 
1 vs 2 (1 1 -1 -1 0 0 0) and interactions (1 -1 -1 1 0 0 0).

Q1:
Is it true that I cannot compare subgroups, because columns denoted zero will 
be handeled as being a covariate? i.e. the contrast 1 -1 0 0 0 0 0 is 
inappropriate and I would have to build a new design matrix (A1 B1 cov1 cov2 
cov3)? This is what I was told previously when using randomise for DTI data, 
but I'm not sure PALM works the same way.

Q2:
What I also understood from randomise is that you have to define both 1 -1 and 
-1 1; there is no option for a 2tailed t-test or F test. Is that correct and 
does it also apply to PALM?

Thanks,

Maaike

___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Re: [Freesurfer] mri_glmfit-sim permutation testing running after 3 days!

[maaike rive]

Hi all,


Is using forced permutation for non-orthogonal design matrices wrong or is it 
allowed to do this instead of using tools like palm (what happens eg with the 
covariates when using forced permutation)? I used forced permutation  and it 
seemed to work, results were (partly) comparable to what I found with monte 
carlo simulations.


Thanks, Maaike


Van: freesurfer-boun...@nmr.mgh.harvard.edu 
 namens Harms, Michael 

Verzonden: vrijdag 26 augustus 2016 01:00:13
Aan: Freesurfer support list
Onderwerp: Re: [Freesurfer] mri_glmfit-sim permutation testing running after 3 
days!


Hi,
You might want to check out FSL’s PALM tool, which has a bit more sophisticated 
permutation framework, and allows for permutation in the context of 
non-orthogonal covariates.

cheers,
-MH

--
Michael Harms, Ph.D.
---
Conte Center for the Neuroscience of Mental Disorders
Washington University School of Medicine
Department of Psychiatry, Box 8134
660 South Euclid Ave. Tel: 314-747-6173
St. Louis, MO  63110 Email: mha...@wustl.edu

From: 
>
 on behalf of Ajay Kurani 
>
Reply-To: Freesurfer support list 
>
Date: Thursday, August 25, 2016 at 4:13 PM
To: Freesurfer support list 
>
Subject: Re: [Freesurfer] mri_glmfit-sim permutation testing running after 3 
days!

Hi Doug,
   Thanks for the help!  I think I figured out the issue based on your response.
1) I created a template to use for this group and named it fsaverage (including 
creating monte carlo simulations) for simplicity of integrating with freesurfer 
as I am newer to it.  This is why the sizes didn't match up as you expected
but the mri_glmfit still ran.

2) I deleted the folder and restarted without background processes.  The error 
became apparent.  Of my covariates (2 fix factors and 3 quantitative), not all 
were orthogonal.  In looking at the error more, it seems that i need to add the
--perm-force if I wanted the simulation to run, however the background 
processes were not aware of this error and kept polling as you mentioned.

This brings me to a new but related issue.  From what I have read in other 
freesurfer posts, it is statistically incorrect to use --perm-force for 
non-orthogonal covariates (or continuous covariates).  I am unsure how to 
proceed.
a) If I ran permutation testing (to overcome the issue of incorrect smoothness 
estimations from the gaussian distribution assumption), then I run into the 
issue of non-orthogonal covariates.  Is there a way to orthogonalize the data in
freesurfer, or a solution to this issue?

b) If orthogonalizing is difficult to implement, another option is running Qdec 
with the montecarlo simulation at a more conservative p value (p< 0.001).  From 
your previous posts, the testing at this p value for 10mm seems to meet the 5% 
FPR.  One question is if the non-orthogonal data affects this analysis as well 
for this model?

Thanks,
Ajay

On Thu, Aug 25, 2016 at 12:18 PM, Ajay Kurani 
> wrote:
Hi Freesurfer Experts,
   I am trying to use freesurfer's mri_glmfit-sim tool to run permutation 
testing on cortical thickness data (as recommended by Doug in my previous post: 
http://www.mail-archive.com/freesurfer%40nmr.mgh.harvard.edu/msg48653.html )

Most of the tutorials I found were not related to permutation testing so the 
subsequent steps may be incorrect.  Please let me know where I go wrong...

1) I first ran QDec to generate a folder for the analysis which would create 
the subsequent fsgd and y files needed my mri_glmfit-sim.  I am running both 
left and right hemisphere cortical thickness analysis with 10mm smoothing.  The 
following is for just the left hemisphere.   Note I am doing a 3 group 
comparison, but for this 2 group ttest I manually centered the data based on 
the 3 group mean for age and education.

2) I ran the following command:
/mri_glmfit-sim --glmdir ./HCvsPAT_lh_thickness_10mm/ --sim perm 1 2 
perm.abs.2 --sim-sign abs --bg 16

Prior to running the command above, from the y.fsdg file I deleted the fwhm 
estimate of 13mm since this was not correctly estimated (ACF with long tails).  
I assumed that by removing this estimate, it would force the permutation test 
to calculate based on the data but when looking at the log output I see the 
following which says fwhm 0:

cmdline mri_glmfit.bin --C 
./HCvsPAT_lh_thickness_10mm//tmp.mri_glmfit-sim-19468/lh-Avg-Intercept-thickness.mtx
 --C 
./HCvsPAT_lh_thickness_10mm//tmp.mri_glmfit-sim-19468/lh-Diff-Male-Female-Intercept-thickness.mtx
 --C 

[Freesurfer] permutation/monte carlo

[maaike rive]

Dear Douglas,


Sorry to bother you with this,  but after reading the thread about the Eklund 
papers  I'm getting confused about what I actually did when correcting for 
multiple comparisons and I'd also like to know whether I'm on the safe side.


I used pre-cashed Monte Carlo simulation with a cluster finding threshold of 
0.05 (at least so I thought):
mri_glmfit -sim - -glmdir- -cache 1.3 abs  - - cache-dir   - - cwpvalthresh 
0.01

Am I correct? And am I also correct that the FDR may be about 13% instead of 5% 
when doing this? (I smoothed thickness and surface area data with FWMH 10mm, 
but not the lGI data, because they were already very smooth and pre-cashed 
simulation didn't work after additional smoothing).

Thanks, Maaike
___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Re: [Freesurfer] average values per cluster

[maaike rive]

Hi Douglas,
 
Can I ask another question relating to an issue I ran into already a year ago?
My lGI data were very smooth at the time (after smoothing for 10mm). I decided 
to use the unsmoothed data, since the fwmh is in the order of 10 then 
(according to the fwhm.dat file). My results are valid then, right? Just to 
check. For area and thickness I did smooth, resulting in comparable fwhm values.
 
Thanks,
 
Maaike
 
Date: Mon, 9 Feb 2015 10:46:40 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per cluster


  

  
  


On 2/8/15 1:48 PM, maaike rive wrote:



  

But how can i run the appropriate preprocessing
  and smoothing? The --X flag s not recognised using
  mris_preproc. 
  

mri_preproc only uses the fsgd file to get the list of subjects to
assure that they are ordered in the same way as when mri_glmfit
builds the X matrix. 


  
I used the same matrix as X in an FSGD file to do
  this now, it seems to work, but I do not no if it is correct.
  Furthermore, If I try to run mri_glmfit --sim for the lGI
  data, I get an error because fwmh66 is not available. I did
  not smooth the lGI data and with the DODS/DOSS FSGD models
  there was no problem, so does this mean my new models are not
  correct?


  

Wow, it is 66 without smoothing? I'm not sure what to tell you. The
lGI is often very smooth, but that seems excessive. You can look for
outliers by loading the y files as a "time coures" in tksurfer (-t
flag). If it says at 66 then I don't think you can do the voxel-wise
analysis. Was it that way when you used the FSGD file? It could be
that your design matrix does not remove the mean offset, so check
that too.


  


  > From: r_maa...@hotmail.com

> Date: Fri, 6 Feb 2015 21:41:53 +0100

> To: freesurfer@nmr.mgh.harvard.edu

> Subject: Re: [Freesurfer] average values per cluster

> 

> Ok, thanks!

> Maaike

> 

> > Op 6 feb. 2015 om 17:49 heeft "Douglas N Greve"
<gr...@nmr.mgh.harvard.edu> het volgende geschreven:

> > 

> > 

> > There is not a way to build more flexible models
in the FSGD structure. 

> > However, you can create your own design matrix and
include anything you 

> > want in it and pass it to mri_glmfit with --X
instead of --fsgd.

> > 

> > doug

> > 

> >> On 02/05/2015 06:13 AM, maaike rive wrote:

> >> 

> >> Hi Doug,

> >> 

> >> I hope I'm not driving you crazy, but I have
some additional questions 

> >> regardig the DOSS/DODS models. Also because of
the SPSS discrepancies.

> >> 

> >> First, the model I discussed with you I used
to assess diagn x age 

> >> interactions, regressing out the effects of
state and gender. However, 

> >> in this model all interaction terms are
incorporated. Is there a way 

> >> to build the model only incorporating main
effects of diagn, state, 

> >> gender, age and diagn*age and diagn*state
interactions? Since I do not 

> >> expect any higher order interactions or state*
age interactions, for 

> >> instance (this is biologically not very
plausible, although I did not 

> >> formally test it).

> >> 

> >> Second, for areas where there is no diagn*age
interaction I want to 

> >> use a model without any interaction term with
age, so I was thinking 

> >> to use DOSS (I still want to regress out the
effects of age, so it was 

> >> added as a regressor):

> >> 

> >> diagn1*state1*gender1

> >> 

> >> diagn1*state1*gender2

> >> 

> >> diagn1*state2*gender1

> >> 

> >> diagn1*state2*gender2

> >> 

> >> diagn2*state1*gender1

> >> 

> >> diagn2*state1*gender2

> >> 

> >> diagn2*

[Freesurfer] visualization tk surfer

[maaike rive]

Dear Freesurfer experts,
I have a question regarding tk surfer. If I load the 
cache.h13.abs.sig.cluster.mgh file, I'll only see the clusters surviving the 
statistical threshold I set, don't I?  (e.g. I set the cluster finding 
threshold at 0.05 (Monte carlo simulation), keeping all clusters with p<0.001, 
which gives me for example one cluster; it's only this cluster that will be 
visualized, right?) I ask this because sometimes only a few clusters are kept, 
but I see a lot of the brain coloured in tk surfer, presumably because the 
clusters are very large.
Thanks, Maaike___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Re: [Freesurfer] does lGI needs correction for any global measure?

[maaike rive]

Hi Doug, Thanks for the reply.I'm still struggling with this issue though.  1. 
For instance, if one corrects surface area for ICV in the GLM, shouldn't the 
fact that this relationship is non-linear be taken into account? Since volume 
increases with r^3 and surface with r^2... Same goes for lGI, how is the 
relationship between lGI en ICV? 2. If I choose to correct for TSA, it seems 
natural to correct thickness for mean (or total) thickness and lGI for some 
global measure (is there such thing as mean or total lGI?) as well, or am I 
mistaken? I don't understand why one measure should be corrected for a global 
measure with the same units, whereas another should not. 3.  ICV is not 
influenced by the volumes of the different brain structures (it is not a 
summation of these volumes) so volume correction for ICV is unbiased. However 
TSA (or mean thickness, or "global lGI measure") is dependent of the area (or 
thickness or lGI) of each discrete brain structure, right? So for instance if 
one group has a smaller superior frontal area, total area will be smaller as 
well. Wouldn't correction remove part of the effect then?  Maaike> To: 
freesurfer@nmr.mgh.harvard.edu; marie.sch...@unige.ch> From: 
gr...@nmr.mgh.harvard.edu> Date: Thu, 5 May 2016 12:32:11 -0400> Subject: Re: 
[Freesurfer] does lGI needs correction for any global measure?> > > > On 
05/02/2016 07:39 AM, maaike rive wrote:> > Dear freesurfer experts,> >  
> >> >> >  > > I have a question regarding correction of 
cortical measures or global > > measures (ICV, total surface area). As I 
understand from literature > > and the freesurfer wiki, a model for thickness 
should not be corrected > > for ICV (since thickness does not scale with total 
brain volume), but > > surface area should be corrected since is does scale 
with total brain > > volume. Three questions:> >> >  > >> >> >  
> > 1. should I use ICV or should I use total surface area (TSA) as > > 
covariate in the model to correct surface area (by the way, average > > ICV and 
TSA do not differ between the groups to compare )?> >> >> Not sure. 
They will be different models. ICV should not change over the > lifetime 
whereas total surface area will.> >  > > 2. does lGI also need 
correction? does it scale with volume? If so, do > > I need to correct for ICV 
or for TSA? Usually I see no correction but > > some do use ICV...> >> >
> I don't know, perhaps Marie will weigh in> >  > > 3. even if 
the average of certain measures (like ICV/TSA, or gender, > > or age) does not 
differ between groups, should they be added to the > > model? If I do not 
correct for gender, for example, my results change > > and most papers do use 
age and gender as covariates if they compare > > groups despite a lack of 
differences between groups...> >> >> They can still reduce 
intersubject variance.> doug> >  > >> >> >  > > 
Thanks,> >> >  > >> >> >  > > Maaike> >> >> 
>> >> > ___> > Freesurfer mailing 
list> > Freesurfer@nmr.mgh.harvard.edu> > 
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer> > -- > Douglas N. 
Greve, Ph.D.> MGH-NMR Center> gr...@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/> > 
___> Freesurfer mailing list> 
Freesurfer@nmr.mgh.harvard.edu> 
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer> > > 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.>   ___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

[Freesurfer] does lGI needs correction for any global measure?

[maaike rive]

Dear freesurfer experts,I have a question regarding correction of cortical 
measures or global measures (ICV, total surface area). As I understand from 
literature and the freesurfer wiki, a model for thickness should not be 
corrected for ICV (since thickness does not scale with total brain volume), but 
surface area should be corrected since is does scale with total brain volume. 
Three questions:1. should I use ICV or should I use total surface area (TSA) as 
covariate in the model to correct surface area (by the way, average ICV and TSA 
do not differ between the groups to compare )?2. does lGI also need correction? 
does it scale with volume? If so, do I need to correct for ICV or for TSA? 
Usually I see no correction but some do use ICV...3. even if the average of 
certain measures (like ICV/TSA, or gender, or age) does not differ between 
groups, should they be added to the model? If I do not correct for gender, for 
example, my results change and most papers do use age and gender as covariates 
if they compare groups despite a lack of differences between groups... 
Thanks,Maaike
  ___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Re: [Freesurfer] average values per cluster

[maaike rive]

But how can i run the appropriate preprocessing and smoothing? The --X flag s 
not recognised using mris_preproc. I used the same matrix as X in an FSGD file 
to do this now, it seems to work, but I do not no if it is correct. 
Furthermore, If I try to run mri_glmfit --sim for the lGI data, I get an error 
because fwmh66 is not available. I did not smooth the lGI data and with the 
DODS/DOSS FSGD models there was no problem, so does this mean my new models are 
not correct?

 From: r_maa...@hotmail.com
 Date: Fri, 6 Feb 2015 21:41:53 +0100
 To: freesurfer@nmr.mgh.harvard.edu
 Subject: Re: [Freesurfer] average values per cluster
 
 Ok, thanks!
 Maaike
 
  Op 6 feb. 2015 om 17:49 heeft Douglas N Greve gr...@nmr.mgh.harvard.edu 
  het volgende geschreven:
  
  
  There is not a way to build more flexible models in the FSGD structure. 
  However, you can create your own design matrix and include anything you 
  want in it and pass it to mri_glmfit with --X instead of --fsgd.
  
  doug
  
  On 02/05/2015 06:13 AM, maaike rive wrote:
  
  Hi Doug,
  
  I hope I'm not driving you crazy, but I have some additional questions 
  regardig the DOSS/DODS models. Also because of the SPSS discrepancies.
  
  First, the model I discussed with you I used to assess diagn x age 
  interactions, regressing out the effects of state and gender. However, 
  in this model all interaction terms are incorporated. Is there a way 
  to build the model only incorporating main effects of diagn, state, 
  gender, age and diagn*age and diagn*state interactions? Since I do not 
  expect any higher order interactions or state* age interactions, for 
  instance (this is biologically not very plausible, although I did not 
  formally test it).
  
  Second, for areas where there is no diagn*age interaction I want to 
  use a model without any interaction term with age, so I was thinking 
  to use DOSS (I still want to regress out the effects of age, so it was 
  added as a regressor):
  
  diagn1*state1*gender1
  
  diagn1*state1*gender2
  
  diagn1*state2*gender1
  
  diagn1*state2*gender2
  
  diagn2*state1*gender1
  
  diagn2*state1*gender2
  
  diagn2*state2*gender1
  
  diagn2*state2*gender2
  
  age
  
  However, here there are still interaction terms with gender in the 
  model. I do want to regress out the effects of gender (since I expect 
  this to be a confounder), without incorporating the interaction term. 
  How should I do this? The only solution I can think of is building a 
  model with gender as additional regressor (containing 0 en 1) and 
  using DOSS. So:
  
  diagn1*state1
  
  diagn1*state2
  
  diagn2*state1
  
  diagn2*state2
  
  age
  
  gender
  
  Does this make sense?
  
  Third, if so, it implies that for areas where there is no diagn*state 
  interaction, and where I want to test diagn1 vs diagn2 (regressing out 
  the effects of state, since I expect this to be a confounder), I 
  should again build a new model:
  
  diagn1
  
  diagn2
  
  age
  
  gender
  
  state
  
  I realize this is a lot of work, hence I hope you could give me 
  any advice about this.
  
  Thanks,
  
  Maaike
  
  
  From: r_maa...@hotmail.com
  To: freesurfer@nmr.mgh.harvard.edu
  Date: Tue, 3 Feb 2015 13:18:55 +0100
  Subject: Re: [Freesurfer] average values per cluster
  
  Hi Doug,
  
  Sorry, the statistician doesn't understand it yet; we're currently 
  building freesurfer and SPSS models step by step to find out what's 
  going on, but it's a litte time consuming... so, to be continued!
  
  Maaike
  
  
  From: r_maa...@hotmail.com
  To: freesurfer@nmr.mgh.harvard.edu
  Date: Fri, 30 Jan 2015 20:55:27 +0100
  Subject: Re: [Freesurfer] average values per cluster
  
  On second thought, I think the reason for the discrepancy is that I 
  included state as a factor instead of covariate in the SPSS model. I 
  ran the model again after having adjusted this and included every 
  possible interaction; now I reach a p value of 0.009 (which is still 
  different but at least significant). But I'll check again Monday!
  
  
  From: r_maa...@hotmail.com
  To: freesurfer@nmr.mgh.harvard.edu
  Date: Fri, 30 Jan 2015 20:15:07 +0100
  Subject: Re: [Freesurfer] average values per cluster
  
  I thought so, but I will check with the statistician next Monday, I'll 
  get back to you as soon as I'm sure.
  
  
  Date: Fri, 30 Jan 2015 14:12:32 -0500
  From: gr...@nmr.mgh.harvard.edu
  To: freesurfer@nmr.mgh.harvard.edu
  Subject: Re: [Freesurfer] average values per cluster
  
  
  Actually, that looks correct. Is this what you fed to SPSS?
  
  On 1/30/15 2:07 PM, maaike rive wrote:
  
 Ok, I'm so sorry, I'm afraid I mixed things up. I indeed used

Re: [Freesurfer] average values per cluster

[maaike rive]

Ok, thanks!
Maaike

 Op 6 feb. 2015 om 17:49 heeft Douglas N Greve gr...@nmr.mgh.harvard.edu 
 het volgende geschreven:
 
 
 There is not a way to build more flexible models in the FSGD structure. 
 However, you can create your own design matrix and include anything you 
 want in it and pass it to mri_glmfit with --X instead of --fsgd.
 
 doug
 
 On 02/05/2015 06:13 AM, maaike rive wrote:
 
 Hi Doug,
 
 I hope I'm not driving you crazy, but I have some additional questions 
 regardig the DOSS/DODS models. Also because of the SPSS discrepancies.
 
 First, the model I discussed with you I used to assess diagn x age 
 interactions, regressing out the effects of state and gender. However, 
 in this model all interaction terms are incorporated. Is there a way 
 to build the model only incorporating main effects of diagn, state, 
 gender, age and diagn*age and diagn*state interactions? Since I do not 
 expect any higher order interactions or state* age interactions, for 
 instance (this is biologically not very plausible, although I did not 
 formally test it).
 
 Second, for areas where there is no diagn*age interaction I want to 
 use a model without any interaction term with age, so I was thinking 
 to use DOSS (I still want to regress out the effects of age, so it was 
 added as a regressor):
 
 diagn1*state1*gender1
 
 diagn1*state1*gender2
 
 diagn1*state2*gender1
 
 diagn1*state2*gender2
 
 diagn2*state1*gender1
 
 diagn2*state1*gender2
 
 diagn2*state2*gender1
 
 diagn2*state2*gender2
 
 age
 
 However, here there are still interaction terms with gender in the 
 model. I do want to regress out the effects of gender (since I expect 
 this to be a confounder), without incorporating the interaction term. 
 How should I do this? The only solution I can think of is building a 
 model with gender as additional regressor (containing 0 en 1) and 
 using DOSS. So:
 
 diagn1*state1
 
 diagn1*state2
 
 diagn2*state1
 
 diagn2*state2
 
 age
 
 gender
 
 Does this make sense?
 
 Third, if so, it implies that for areas where there is no diagn*state 
 interaction, and where I want to test diagn1 vs diagn2 (regressing out 
 the effects of state, since I expect this to be a confounder), I 
 should again build a new model:
 
 diagn1
 
 diagn2
 
 age
 
 gender
 
 state
 
 I realize this is a lot of work, hence I hope you could give me 
 any advice about this.
 
 Thanks,
 
 Maaike
 
 
 From: r_maa...@hotmail.com
 To: freesurfer@nmr.mgh.harvard.edu
 Date: Tue, 3 Feb 2015 13:18:55 +0100
 Subject: Re: [Freesurfer] average values per cluster
 
 Hi Doug,
 
 Sorry, the statistician doesn't understand it yet; we're currently 
 building freesurfer and SPSS models step by step to find out what's 
 going on, but it's a litte time consuming... so, to be continued!
 
 Maaike
 
 
 From: r_maa...@hotmail.com
 To: freesurfer@nmr.mgh.harvard.edu
 Date: Fri, 30 Jan 2015 20:55:27 +0100
 Subject: Re: [Freesurfer] average values per cluster
 
 On second thought, I think the reason for the discrepancy is that I 
 included state as a factor instead of covariate in the SPSS model. I 
 ran the model again after having adjusted this and included every 
 possible interaction; now I reach a p value of 0.009 (which is still 
 different but at least significant). But I'll check again Monday!
 
 
 From: r_maa...@hotmail.com
 To: freesurfer@nmr.mgh.harvard.edu
 Date: Fri, 30 Jan 2015 20:15:07 +0100
 Subject: Re: [Freesurfer] average values per cluster
 
 I thought so, but I will check with the statistician next Monday, I'll 
 get back to you as soon as I'm sure.
 
 
 Date: Fri, 30 Jan 2015 14:12:32 -0500
 From: gr...@nmr.mgh.harvard.edu
 To: freesurfer@nmr.mgh.harvard.edu
 Subject: Re: [Freesurfer] average values per cluster
 
 
 Actually, that looks correct. Is this what you fed to SPSS?
 
 On 1/30/15 2:07 PM, maaike rive wrote:
 
Ok, I'm so sorry, I'm afraid I mixed things up. I indeed used
DODS, and assumed these were may regressors:
 
diagn1*state1*gender1
 
diagn1*state1*gender2
 
diagn1*state2*gender1
 
diagn1*state2*gender2
 
diagn2*state1*gender1
 
diagn2*state1*gender2
 
diagn2*state2*gender1
 
diagn2*state2*gender2
 
diagn1*state1*gender1*age
 
diagn1*state1*gender2*age
 
diagn1*state2*gender1*age
 
diagn1*state2*gender2*age
 
diagn2*state1*gender1*age
 
diagn2*state1*gender2*age
 
diagn2*state2*gender1*age
 
diagn2*state2*gender2*age
 
 
However what I actually meant to do was testing for the effects of
diagnosis and interactions (e.g. diagnosis*age) by using the
appropriate contrasts, so I thought that if I used the contrast
for the diagnosis*age interaction, I was testing

Re: [Freesurfer] defining ROIs/clusters

[maaike rive]

Won't I get the data for the complete eg insula then, instead of the cluster 
within the insula resulting from the comparison?
 
Related to this: I noticed that in the aparc.stats files the magnitude of 
surface area is in the order of 100-1000 mm2, whereas in the ocn.dat files 
resulting from the clusterwise corrected analyses the average values of the 
resulting clusters are in the order of 0.something till 1 or 2, so a factor 
1000 smaller. How can this be explained?
 
Thanks again,
 
Maaike   


Date: Tue, 3 Feb 2015 10:30:10 -0500From: gr...@nmr.mgh.harvard.eduTo: 
freesur...@nmr.mgh.harvard.eduSubject: Re: [Freesurfer] defining 
ROIs/clustersJust run mris_preproc on your healthy group, then run mri_segstats 
using the cluster annotation from your first analysis and the HC stack  as the 
input. Does that make sense?doug
On 2/3/15 9:18 AM, maaike rive wrote:



Hi all, Does anyone know how to extract average (thickness/area/GI) values of a 
cluster A for subjects in a group which was NOT included in the model where 
this cluster A emerged? For example: in a t-test for diagnosis1dagnosis2 there 
is an effect in an insulacluster (found with mri_glmfit -sim); post hoc I want 
to test whether diagnosis1 and/or diagnosis2 differ from healthy controls in 
exactly the same cluster. I understand I can run the analyses in seperate 
models (one for diagnosis1 vs HC and one for diagnosis2 vs HC; if this cluster 
does not emerge in either model, there is no difference) but if I want to plot 
all groups for visualization purposes, I have no values for the HC.  Thanks in 
advance, Maaike  
 ___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer___
 Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu 
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer 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.   ___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Re: [Freesurfer] average values per cluster

[maaike rive]

Hi Doug,
 
I hope I'm not driving you crazy, but I have some additional questions regardig 
the DOSS/DODS models. Also because of the SPSS discrepancies.
 
First, the model I discussed with you I used to assess diagn x age 
interactions, regressing out the effects of state and gender. However, in this 
model all interaction terms are incorporated. Is there a way to build the model 
only incorporating main effects of diagn, state, gender, age and diagn*age and 
diagn*state interactions? Since I do not expect any higher order interactions 
or state* age interactions, for instance (this is biologically not very 
plausible, although I did not formally test it). 
 
Second, for areas where there is no diagn*age interaction I want to use a model 
without any interaction term with age, so I was thinking to use DOSS (I still 
want to regress out the effects of age, so it was added as a regressor):
diagn1*state1*gender1 
diagn1*state1*gender2 
diagn1*state2*gender1 
diagn1*state2*gender2 
diagn2*state1*gender1 
diagn2*state1*gender2 
diagn2*state2*gender1 
diagn2*state2*gender2 
age
 
However, here there are still interaction terms with gender in the model. I do 
want to regress out the effects of gender (since I expect this to be a 
confounder), without incorporating the interaction term. How should I do this? 
The only solution I can think of is building a model with gender as additional 
regressor (containing 0 en 1) and using DOSS. So: 
diagn1*state1
diagn1*state2 
diagn2*state1
diagn2*state2
age
gender
 
Does this make sense?
 
Third, if so, it implies that for areas where there is no diagn*state 
interaction, and where I want to test diagn1 vs diagn2 (regressing out the 
effects of state, since I expect this to be a confounder), I should again build 
a new model:
 
diagn1
diagn2
age
gender
state
 
I realize this is a lot of work, hence I hope you could give me any advice 
about this.
 
Thanks,
 
Maaike 


From: r_maaike@hotmail.comTo: freesur...@nmr.mgh.harvard.eduDate: Tue, 3 Feb 
2015 13:18:55 +0100Subject: Re: [Freesurfer] average values per cluster



Hi Doug,
 
Sorry, the statistician doesn't understand it yet; we're currently building 
freesurfer and SPSS models step by step to find out what's going on, but it's a 
litte time consuming... so, to be continued!
 
Maaike 


From: r_maaike@hotmail.comTo: freesur...@nmr.mgh.harvard.eduDate: Fri, 30 Jan 
2015 20:55:27 +0100Subject: Re: [Freesurfer] average values per cluster


On second thought, I think the reason for the discrepancy is that I included 
state as a factor instead of covariate in the SPSS model. I ran the model again 
after having adjusted this and included every possible interaction; now I reach 
a p value of 0.009 (which is still different but at least significant). But 
I'll check again Monday!


From: r_maaike@hotmail.comTo: freesur...@nmr.mgh.harvard.eduDate: Fri, 30 Jan 
2015 20:15:07 +0100Subject: Re: [Freesurfer] average values per cluster


I thought so, but I will check with the statistician next Monday, I'll get back 
to you as soon as I'm sure. 


Date: Fri, 30 Jan 2015 14:12:32 -0500From: gr...@nmr.mgh.harvard.eduTo: 
freesur...@nmr.mgh.harvard.eduSubject: Re: [Freesurfer] average values per 
clusterActually, that looks correct. Is this what you fed to SPSS?
On 1/30/15 2:07 PM, maaike rive wrote:



Ok, I'm so sorry, I'm afraid I mixed things up. I indeed used DODS, and assumed 
these were may regressors:  


diagn1*state1*gender1 
diagn1*state1*gender2 
diagn1*state2*gender1 
diagn1*state2*gender2 
diagn2*state1*gender1 
diagn2*state1*gender2 
diagn2*state2*gender1 
diagn2*state2*gender2 
diagn1*state1*gender1*age
diagn1*state1*gender2*age 
diagn1*state2*gender1*age
diagn1*state2*gender2*age
diagn2*state1*gender1*age
diagn2*state1*gender2*age
diagn2*state2*gender1*age

diagn2*state2*gender2*age

However what I actually meant to do was testing for the effects of diagnosis 
and interactions (e.g. diagnosis*age) by using the appropriate contrasts, so I 
thought that if I used the contrast for the diagnosis*age interaction, I was 
testing whether or not this interaction was significant regressing out the 
effects of state and gender. Another example, I thought that if I used the 
contrast for diagnosis1 versus diagnosis2, I was testing wether or not there 
was a significant effect of diagnosis regressing out the effects of state, 
gender and age.

 But that is not the case than? How should I build the model to answer these 
questions? 

Maaike



Date: Fri, 30 Jan 2015 13:33:40 -0500From: gr...@nmr.mgh.harvard.eduTo: 
freesur...@nmr.mgh.harvard.eduSubject: Re: [Freesurfer] average values per 
clusterIt depends on the actual covariates and whether you used DOSS or DODS. 
The default is DODS, which is a full interaction model. Assuming that state is 
a categorical variable with 2 levels, you'd have 4 regressors (gender by state) 
+ 4 more regressors (gender by state by age). If the SPSS model did not have 8 
regressors

[Freesurfer] defining ROIs/clusters

[maaike rive]

Hi all,

 

Does anyone know how to extract average (thickness/area/GI) values of a cluster 
A for subjects in a group which was NOT included in the model where this 
cluster A emerged? For example: in a t-test for diagnosis1dagnosis2 there is 
an effect in an insulacluster (found with mri_glmfit -sim); post hoc I want to 
test whether diagnosis1 and/or diagnosis2 differ from healthy controls in 
exactly the same cluster. I understand I can run the analyses in seperate 
models (one for diagnosis1 vs HC and one for diagnosis2 vs HC; if this cluster 
does not emerge in either model, there is no difference) but if I want to plot 
all groups for visualization purposes, I have no values for the HC. 

 

Thanks in advance,

 

Maaike

 

 
  ___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Re: [Freesurfer] average values per cluster

[maaike rive]

Hi Doug,
 
Sorry, the statistician doesn't understand it yet; we're currently building 
freesurfer and SPSS models step by step to find out what's going on, but it's a 
litte time consuming... so, to be continued!
 
Maaike 


From: r_maaike@hotmail.comTo: freesur...@nmr.mgh.harvard.eduDate: Fri, 30 Jan 
2015 20:55:27 +0100Subject: Re: [Freesurfer] average values per cluster


On second thought, I think the reason for the discrepancy is that I included 
state as a factor instead of covariate in the SPSS model. I ran the model again 
after having adjusted this and included every possible interaction; now I reach 
a p value of 0.009 (which is still different but at least significant). But 
I'll check again Monday!


From: r_maaike@hotmail.comTo: freesur...@nmr.mgh.harvard.eduDate: Fri, 30 Jan 
2015 20:15:07 +0100Subject: Re: [Freesurfer] average values per cluster


I thought so, but I will check with the statistician next Monday, I'll get back 
to you as soon as I'm sure. 


Date: Fri, 30 Jan 2015 14:12:32 -0500From: gr...@nmr.mgh.harvard.eduTo: 
freesur...@nmr.mgh.harvard.eduSubject: Re: [Freesurfer] average values per 
clusterActually, that looks correct. Is this what you fed to SPSS?
On 1/30/15 2:07 PM, maaike rive wrote:



Ok, I'm so sorry, I'm afraid I mixed things up. I indeed used DODS, and assumed 
these were may regressors:  


diagn1*state1*gender1 
diagn1*state1*gender2 
diagn1*state2*gender1 
diagn1*state2*gender2 
diagn2*state1*gender1 
diagn2*state1*gender2 
diagn2*state2*gender1 
diagn2*state2*gender2 
diagn1*state1*gender1*age
diagn1*state1*gender2*age 
diagn1*state2*gender1*age
diagn1*state2*gender2*age
diagn2*state1*gender1*age
diagn2*state1*gender2*age
diagn2*state2*gender1*age

diagn2*state2*gender2*age

However what I actually meant to do was testing for the effects of diagnosis 
and interactions (e.g. diagnosis*age) by using the appropriate contrasts, so I 
thought that if I used the contrast for the diagnosis*age interaction, I was 
testing whether or not this interaction was significant regressing out the 
effects of state and gender. Another example, I thought that if I used the 
contrast for diagnosis1 versus diagnosis2, I was testing wether or not there 
was a significant effect of diagnosis regressing out the effects of state, 
gender and age.

 But that is not the case than? How should I build the model to answer these 
questions? 

Maaike



Date: Fri, 30 Jan 2015 13:33:40 -0500From: gr...@nmr.mgh.harvard.eduTo: 
freesur...@nmr.mgh.harvard.eduSubject: Re: [Freesurfer] average values per 
clusterIt depends on the actual covariates and whether you used DOSS or DODS. 
The default is DODS, which is a full interaction model. Assuming that state is 
a categorical variable with 2 levels, you'd have 4 regressors (gender by state) 
+ 4 more regressors (gender by state by age). If the SPSS model did not have 8 
regressors, then it is not the same.doug
On 1/30/15 1:26 PM, maaike rive wrote:



How should I have described the model?


Date: Fri, 30 Jan 2015 13:09:40 -0500From: gr...@nmr.mgh.harvard.eduTo: 
freesur...@nmr.mgh.harvard.eduSubject: Re: [Freesurfer] average values per 
clusterI told him it the model is a GLM with age, gender and state as 
covariatesThis is not a sufficient description. Several design matrices could 
be derived from this description. 
On 1/30/15 11:58 AM, maaike rive wrote:



The p value from SPSS is 0.156. I checked with a statistician, just to be sure, 
but he says the SPSS model is correct... (he doesn't know anything about 
freesurfer and FSGD files, but I told him it the model is a GLM with age, 
gender and state as covariates).


Date: Fri, 30 Jan 2015 11:46:37 -0500From: gr...@nmr.mgh.harvard.eduTo: 
freesur...@nmr.mgh.harvard.eduSubject: Re: [Freesurfer] average values per 
clusterYou should report the clusterwise p-value. The p-value that you have 
computed is not interpretable. If it did not come out significant, then it 
would be worrisome, but it is only a check and cannot be used for anything. 
What p-value did you get from SPSS? I'd be curious to know what the source of 
the descrpancy is since this has happened several times now.doug
On 1/30/15 3:25 AM, maaike rive wrote:





Hi Douglas,

I used the matlab code and the p value is signifiant, although less so than the 
p-value I get from the clusterwise statistics (0.0027 vs 0.00010). (Indeed I 
used an abs threshold for the clusterwise statistics). 

Which p-value should I report? 

Thanks,

Maaike



From: r_maaike@hotmail.comTo: freesur...@nmr.mgh.harvard.eduDate: Thu, 29 Jan 
2015 19:48:44 +0100Subject: Re: [Freesurfer] average values per cluster


No not yet; I will do so!
 Date: Thu, 29 Jan 2015 13:44:33 -0500 From: gr...@nmr.mgh.harvard.edu To: 
 freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] average values per 
 cluster   The t is correct. When the contrast matrix only has one row, 
 then the F  is just an unsigned t. I don't know anything about SPSS so I 
 can't tell  from

Re: [Freesurfer] average values per cluster

[maaike rive]

The p value from SPSS is 0.156. I checked with a statistician, just to be sure, 
but he says the SPSS model is correct... (he doesn't know anything about 
freesurfer and FSGD files, but I told him it the model is a GLM with age, 
gender and state as covariates).
Date: Fri, 30 Jan 2015 11:46:37 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per cluster


  

  
  


You should report the clusterwise p-value. The p-value that you have
computed is not interpretable. If it did not come out significant,
then it would be worrisome, but it is only a check and cannot be
used for anything. What p-value did you get from SPSS? I'd be
curious to know what the source of the descrpancy is since this has
happened several times now.



doug



On 1/30/15 3:25 AM, maaike rive wrote:



  


  Hi Douglas,
  

  
  I used the matlab code and the p value is signifiant,
although less so than the p-value I get from the clusterwise
statistics (0.0027 vs
  0.00010). (Indeed I used an abs
  threshold for the clusterwise statistics). 
  


  Which p-value should I report? 
  


  Thanks,
  


  Maaike
  


  From: r_maa...@hotmail.com

  To: freesurfer@nmr.mgh.harvard.edu

  Date: Thu, 29 Jan 2015 19:48:44 +0100

  Subject: Re: [Freesurfer] average values per cluster

  

  
  No not yet; I will do so!



 Date: Thu, 29 Jan 2015 13:44:33 -0500

   From: gr...@nmr.mgh.harvard.edu

   To: freesurfer@nmr.mgh.harvard.edu

   Subject: Re: [Freesurfer] average values per
  cluster

   

   

   The t is correct. When the contrast matrix only
  has one row, then the F 

   is just an unsigned t. I don't know anything
  about SPSS so I can't tell 

   from what you have sent whether it is the same
  model or not. Try to get 

   SPSS to output the design matrix. Did you try the
  matlab code below?

   

   On 01/29/2015 01:40 PM, maaike rive wrote:

Hi Douglas,

   

I checked but as far as I see it I used the
  same models. I attached 

the SPSS model and output as wel as the FSGD
  file and contrast file 

(to test a diagnosis x age interaction). Now
  that I come to think of 

it, I think the reason for the discrepant
  findings is that I used a 

t-contrast instead of an F-contrast for the
  interaction in Freesurfer. 

Apologies I did not think about this
  earlier. But maybe there is 

something else I did completely wrong.

   

What do you think?

   

If it is indeed the t vs F contrast, than
  how should I specify that 

the contrast is an F-contrast? Is it ok to
  just add another line to 

the contrastfile (so

0 0 0 0 0 0 0 0 1 1 1 1 -1 -1 -1 -1

0 0 0 0 0 0 0 0 -1 -1 -1 -1 1 1 1 1)?

   

Thanks again,

   

Maaike

   

 Date: Mon, 26 Jan 2015 11:23:59 -0500

 From: gr...@nmr.mgh.harvard.edu

 To: freesurfer@nmr.mgh.harvard.edu

 Subject: Re: [Freesurfer] average
  values per cluster





 We get these kind of reports
  occasionally. When I ask people to confirm

 that they use exactly the same design
  matrix in SPSS, I never hear 

back,

 so I assume that it gets resolved. So
  please check. The other thing you

 can do is to run in matlab, something
  like



 cd glmdir/contrast

 X = load('Xg.dat');

 y = load('ocn.dat');

 C = load('C.dat');

 [beta rvar] = fast_glmfit(y,X);

 [F p] = fast_fratio(beta,X,rvar,C);

 p will be the p-value

Re: [Freesurfer] average values per cluster

[maaike rive]

I thought so, but I will check with the statistician next Monday, I'll get back 
to you as soon as I'm sure. 

Date: Fri, 30 Jan 2015 14:12:32 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per cluster


  

  
  


Actually, that looks correct. Is this what you fed to SPSS?



On 1/30/15 2:07 PM, maaike rive wrote:



  
  Ok, I'm so sorry, I'm afraid I mixed things up. I
indeed used DODS, and assumed these were may regressors: 




  
  
  
  
  diagn1*state1*gender1 
  diagn1*state1*gender2 
  diagn1*state2*gender1 
  diagn1*state2*gender2 
  diagn2*state1*gender1 
  diagn2*state1*gender2 
  diagn2*state2*gender1 
  diagn2*state2*gender2 
  diagn1*state1*gender1*age
  diagn1*state1*gender2*age 
  diagn1*state2*gender1*age
  diagn1*state2*gender2*age
  diagn2*state1*gender1*age
  diagn2*state1*gender2*age
  diagn2*state2*gender1*age
  
  diagn2*state2*gender2*age
  


  However
  what I actually meant to do was testing for the effects of
  diagnosis and interactions (e.g. diagnosis*age) by using
  the appropriate
  contrasts, so I thought that if I used the contrast for
  the diagnosis*age
  interaction, I was testing whether or not this interaction
  was significant
  regressing out the effects of state and gender. Another
  example, I thought that if I used the contrast for
  diagnosis1 versus diagnosis2, I was testing wether or not
  there was a significant effect of diagnosis regressing out the 
effects of state, gender and age.
  


   But that
  is not the case than? How should I build the model to
  answer these questions? 
  


  Maaike
  
  
  
  
  
  

  
  

  
  

  
  Date: Fri, 30 Jan 2015 13:33:40 -0500

From: gr...@nmr.mgh.harvard.edu

To: freesurfer@nmr.mgh.harvard.edu

Subject: Re: [Freesurfer] average values per cluster





It depends on the actual covariates and whether you used
DOSS or
DODS. The default is DODS, which is a full interaction
model.
Assuming that state is a categorical variable with 2 levels,
you'd
have 4 regressors (gender by state) + 4 more regressors
(gender by
state by age). If the SPSS model did not have 8 regressors,
then it
is not the same.



doug



On 1/30/15 1:26 PM, maaike
  rive wrote:



  
  How should I have described the model?




  Date: Fri, 30 Jan 2015
  13:09:40 -0500

  From: gr...@nmr.mgh.harvard.edu

  To: freesurfer@nmr.mgh.harvard.edu

  Subject: Re: [Freesurfer] average values per cluster

  

  I told him it the model is a GLM with age,
  gender
  and state as covariates

  

  This is not a sufficient description. Several design
  matrices
  could be derived from this description. 

  

  On 1/30/15 11:58 AM,
maaike
rive wrote:

  
  

The p value from SPSS is 0.156. I
  checked
  with a statistician, just to be sure, but he says
  the SPSS
  model is correct... (he doesn't know anything
  about
  freesurfer and FSGD files, but I told him it the
  model is
  a GLM with age, gender and state as covariates).

  
Date: Fri, 30 Jan 2015
11:46:37
-0500

From: gr...@nmr.mgh.harvard.edu

To: freesurfer@nmr.mgh.harvard.edu

Subject: Re: [Freesurfer] average values per
cluster





You should report the clusterwise p-value. The
p-value
that you have computed is not interpretable

Re: [Freesurfer] average values per cluster

[maaike rive]

Ok, I'm so sorry, I'm afraid I mixed things up. I indeed used DODS, and assumed 
these were may regressors: 

















diagn1*state1*gender1 diagn1*state1*gender2 diagn1*state2*gender1 
diagn1*state2*gender2 diagn2*state1*gender1 diagn2*state1*gender2 
diagn2*state2*gender1 diagn2*state2*gender2 
diagn1*state1*gender1*agediagn1*state1*gender2*age 
diagn1*state2*gender1*agediagn1*state2*gender2*agediagn2*state1*gender1*agediagn2*state1*gender2*agediagn2*state2*gender1*agediagn2*state2*gender2*age
However what I actually meant to do was testing for the effects of
diagnosis and interactions (e.g. diagnosis*age) by using the appropriate
contrasts, so I thought that if I used the contrast for the diagnosis*age
interaction, I was testing whether or not this interaction was significant
regressing out the effects of state and gender. Another example, I thought that 
if I used the contrast for diagnosis1 versus diagnosis2, I was testing wether 
or not there was a significant effect of diagnosis regressing out the effects 
of state, gender and age.
 But that is not the case than? How should I build the model to answer these 
questions? 
Maaike





















Date: Fri, 30 Jan 2015 13:33:40 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per cluster








It depends on the actual covariates and whether you used DOSS or
DODS. The default is DODS, which is a full interaction model.
Assuming that state is a categorical variable with 2 levels, you'd
have 4 regressors (gender by state) + 4 more regressors (gender by
state by age). If the SPSS model did not have 8 regressors, then it
is not the same.



doug



On 1/30/15 1:26 PM, maaike rive wrote:




How should I have described the model?




Date: Fri, 30 Jan 2015 13:09:40 -0500

From: gr...@nmr.mgh.harvard.edu

To: freesurfer@nmr.mgh.harvard.edu

Subject: Re: [Freesurfer] average values per cluster



I told him it the model is a GLM with age, gender
and state as covariates



This is not a sufficient description. Several design matrices
could be derived from this description. 



On 1/30/15 11:58 AM, maaike
rive wrote:




The p value from SPSS is 0.156. I checked
with a statistician, just to be sure, but he says the SPSS
model is correct... (he doesn't know anything about
freesurfer and FSGD files, but I told him it the model is
a GLM with age, gender and state as covariates).


Date: Fri, 30 Jan 2015 11:46:37
-0500

From: gr...@nmr.mgh.harvard.edu

To: freesurfer@nmr.mgh.harvard.edu

Subject: Re: [Freesurfer] average values per cluster





You should report the clusterwise p-value. The p-value
that you have computed is not interpretable. If it did
not come out significant, then it would be worrisome,
but it is only a check and cannot be used for anything.
What p-value did you get from SPSS? I'd be curious to
know what the source of the descrpancy is since this has
happened several times now.



doug



On 1/30/15 3:25 AM,
maaike rive wrote:






Hi Douglas,



I used the matlab code and the p value is
signifiant, although less so than the p-value I
get from the clusterwise statistics (0.0027 vs 0.00010). (Indeed I used an abs 
threshold for
the clusterwise statistics). 



Which p-value should I
report? 



Thanks,



Maaike



From: r_maa...@hotmail.com

To: freesurfer@nmr.mgh.harvard.edu

Date: Thu, 29 Jan 2015 19:48:44 +0100

Subject: Re: [Freesurfer] average values per
cluster




No not yet; I will do so!



 Date: Thu, 29 Jan 2015 13:44:33
-0500

 From: gr...@nmr.mgh.harvard.edu

 To: freesurfer@nmr.mgh.harvard.edu

 Subject: Re: [Freesurfer] average
values per cluster

 

 

 The t is correct. When the contrast
matrix only has one row, then the F 

 is just an unsigned t. I don't know
anything about SPSS so I can't tell 

 from what you have sent whether it is
the same model or not. Try to get 

 SPSS to output the design matrix. Did
you try the matlab code below?

 

 On 01/29/2015 01:40 PM, maaike rive
wrote:

  Hi Douglas,

 

  I checked but as far as I see it
I used the same models. I attached 

  the SPSS model and output as wel
as the FSGD file and contrast file 

  (to test a diagnosis x age
interaction). Now that I come to think of


  it, I think the reason for the
discrepant findings is that I used a 

  t-contrast instead of an
F-contrast for the interaction in
Freesurfer. 

  Apologies I did not think about
this earlier. But maybe there is 

  something else I did completely
wrong.

 

  What do you think?

 

  If it is indeed the t vs F
contrast, than how should I specify that 

  the contrast is an F-contrast?
Is it ok to just add another line to 

  the contrastfile (so

  0 0 0 0 0 0 0 0 1 1 1 1 -1 -1 -1
-1

  0 0 0 0 0 0 0 0 -1 -1 -1 -1 1 1
1 1)?

 

  Thanks again,

 

  Maaike

 

   Date: Mon, 26 Jan 2015
11:23:59 -0500

   From: gr...@nmr.mgh.harvard.edu

   To: freesurfer@nmr.mgh.harvard.edu

   Subject: Re: [Freesurfer]
average values per cluster

Re: [Freesurfer] average values per cluster

[maaike rive]

How should I have described the model?

Date: Fri, 30 Jan 2015 13:09:40 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per cluster


  

  
  
I told him it the model is a GLM with age, gender and
state as covariates



This is not a sufficient description. Several design matrices could
be derived from this description. 



On 1/30/15 11:58 AM, maaike rive wrote:



  
  The p value from SPSS is 0.156. I checked with a
statistician, just to be sure, but he says the SPSS model is
correct... (he doesn't know anything about freesurfer and FSGD
files, but I told him it the model is a GLM with age, gender and
state as covariates).


  Date: Fri, 30 Jan 2015 11:46:37 -0500

  From: gr...@nmr.mgh.harvard.edu

  To: freesurfer@nmr.mgh.harvard.edu

  Subject: Re: [Freesurfer] average values per cluster

  

  

  You should report the clusterwise p-value. The p-value that
  you have computed is not interpretable. If it did not come out
  significant, then it would be worrisome, but it is only a
  check and cannot be used for anything. What p-value did you
  get from SPSS? I'd be curious to know what the source of the
  descrpancy is since this has happened several times now.

  

  doug

  

  On 1/30/15 3:25 AM, maaike
rive wrote:

  
  

  
  
Hi Douglas,



I used the matlab code and the p value is
  signifiant, although less so than the p-value I get
  from the clusterwise statistics (0.0027 vs 0.00010). (Indeed 
I used an abs threshold for the clusterwise statistics). 


  
Which p-value should I report? 


  
Thanks,


  
Maaike


  
From: r_maa...@hotmail.com

To: freesurfer@nmr.mgh.harvard.edu

Date: Thu, 29 Jan 2015 19:48:44 +0100

Subject: Re: [Freesurfer] average values per cluster




No not yet; I will do so!

  

   Date: Thu, 29 Jan 2015 13:44:33 -0500

 From: gr...@nmr.mgh.harvard.edu

 To: freesurfer@nmr.mgh.harvard.edu

 Subject: Re: [Freesurfer] average values
per cluster

 

 

 The t is correct. When the contrast matrix
only has one row, then the F 

 is just an unsigned t. I don't know
anything about SPSS so I can't tell 

 from what you have sent whether it is the
same model or not. Try to get 

 SPSS to output the design matrix. Did you
try the matlab code below?

 

 On 01/29/2015 01:40 PM, maaike rive wrote:

  Hi Douglas,

 

  I checked but as far as I see it I
used the same models. I attached 

  the SPSS model and output as wel as
the FSGD file and contrast file 

  (to test a diagnosis x age
interaction). Now that I come to think of 

  it, I think the reason for the
discrepant findings is that I used a 

  t-contrast instead of an F-contrast
for the interaction in Freesurfer. 

  Apologies I did not think about this
earlier. But maybe there is 

  something else I did completely wrong.

 

  What do you think?

 

  If it is indeed the t vs F contrast,
than how should I specify that 

  the contrast is an F-contrast? Is it
ok to just add another line to 

  the contrastfile (so

  0 0 0 0 0 0 0 0 1 1 1 1 -1 -1 -1 -1

  0 0 0 0 0 0 0 0 -1 -1 -1 -1 1 1 1 1)?

 

  Thanks again,

 

  Maaike

Re: [Freesurfer] average values per cluster

[maaike rive]

On second thought, I think the reason for the discrepancy is that I included 
state as a factor instead of covariate in the SPSS model. I ran the model again 
after having adjusted this and included every possible interaction; now I reach 
a p value of 0.009 (which is still different but at least significant). But 
I'll check again Monday!

From: r_maa...@hotmail.com
To: freesurfer@nmr.mgh.harvard.edu
Date: Fri, 30 Jan 2015 20:15:07 +0100
Subject: Re: [Freesurfer] average values per cluster




I thought so, but I will check with the statistician next Monday, I'll get back 
to you as soon as I'm sure. 

Date: Fri, 30 Jan 2015 14:12:32 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per cluster


  

  
  


Actually, that looks correct. Is this what you fed to SPSS?



On 1/30/15 2:07 PM, maaike rive wrote:



  
  Ok, I'm so sorry, I'm afraid I mixed things up. I
indeed used DODS, and assumed these were may regressors: 




  
  
  
  
  diagn1*state1*gender1 
  diagn1*state1*gender2 
  diagn1*state2*gender1 
  diagn1*state2*gender2 
  diagn2*state1*gender1 
  diagn2*state1*gender2 
  diagn2*state2*gender1 
  diagn2*state2*gender2 
  diagn1*state1*gender1*age
  diagn1*state1*gender2*age 
  diagn1*state2*gender1*age
  diagn1*state2*gender2*age
  diagn2*state1*gender1*age
  diagn2*state1*gender2*age
  diagn2*state2*gender1*age
  
  diagn2*state2*gender2*age
  


  However
  what I actually meant to do was testing for the effects of
  diagnosis and interactions (e.g. diagnosis*age) by using
  the appropriate
  contrasts, so I thought that if I used the contrast for
  the diagnosis*age
  interaction, I was testing whether or not this interaction
  was significant
  regressing out the effects of state and gender. Another
  example, I thought that if I used the contrast for
  diagnosis1 versus diagnosis2, I was testing wether or not
  there was a significant effect of diagnosis regressing out the 
effects of state, gender and age.
  


   But that
  is not the case than? How should I build the model to
  answer these questions? 
  


  Maaike
  
  
  
  
  
  

  
  

  
  

  
  Date: Fri, 30 Jan 2015 13:33:40 -0500

From: gr...@nmr.mgh.harvard.edu

To: freesurfer@nmr.mgh.harvard.edu

Subject: Re: [Freesurfer] average values per cluster





It depends on the actual covariates and whether you used
DOSS or
DODS. The default is DODS, which is a full interaction
model.
Assuming that state is a categorical variable with 2 levels,
you'd
have 4 regressors (gender by state) + 4 more regressors
(gender by
state by age). If the SPSS model did not have 8 regressors,
then it
is not the same.



doug



On 1/30/15 1:26 PM, maaike
  rive wrote:



  
  How should I have described the model?




  Date: Fri, 30 Jan 2015
  13:09:40 -0500

  From: gr...@nmr.mgh.harvard.edu

  To: freesurfer@nmr.mgh.harvard.edu

  Subject: Re: [Freesurfer] average values per cluster

  

  I told him it the model is a GLM with age,
  gender
  and state as covariates

  

  This is not a sufficient description. Several design
  matrices
  could be derived from this description. 

  

  On 1/30/15 11:58 AM,
maaike
rive wrote:

  
  

The p value from SPSS is 0.156. I
  checked
  with a statistician, just to be sure, but he says
  the SPSS
  model is correct... (he doesn't know anything
  about
  freesurfer and FSGD files, but I told him it the
  model is
  a GLM with age, gender and state as covariates).

  
Date: Fri, 30 Jan 2015

Re: [Freesurfer] average values per cluster

[maaike rive]

Hi Douglas,
I used the matlab code and the p value is signifiant, although less so than the 
p-value I get from the clusterwise statistics (0.0027 vs 0.00010). (Indeed I 
used an abs threshold for the clusterwise statistics). 
Which p-value should I report? 
Thanks,
Maaike
From: r_maa...@hotmail.com
To: freesurfer@nmr.mgh.harvard.edu
Date: Thu, 29 Jan 2015 19:48:44 +0100
Subject: Re: [Freesurfer] average values per cluster




No not yet; I will do so!

 Date: Thu, 29 Jan 2015 13:44:33 -0500
 From: gr...@nmr.mgh.harvard.edu
 To: freesurfer@nmr.mgh.harvard.edu
 Subject: Re: [Freesurfer] average values per cluster
 
 
 The t is correct. When the contrast matrix only has one row, then the F 
 is just an unsigned t. I don't know anything about SPSS so I can't tell 
 from what you have sent whether it is the same model or not. Try to get 
 SPSS to output the design matrix. Did you try the matlab code below?
 
 On 01/29/2015 01:40 PM, maaike rive wrote:
  Hi Douglas,
 
  I checked but as far as I see it I used the same models. I attached 
  the SPSS model and output as wel as the FSGD file and contrast file 
  (to test a diagnosis x age interaction). Now that I come to think of 
  it, I think the reason for the discrepant findings is that I used a 
  t-contrast instead of an F-contrast for the interaction in Freesurfer. 
  Apologies I did not think about this earlier. But maybe there is 
  something else I did completely wrong.
 
  What do you think?
 
  If it is indeed the t vs F contrast, than how should I specify that 
  the contrast is an F-contrast? Is it ok to just add another line to 
  the contrastfile (so
  0 0 0 0 0 0 0 0 1 1 1 1 -1 -1 -1 -1
  0 0 0 0 0 0 0 0 -1 -1 -1 -1 1 1 1 1)?
 
  Thanks again,
 
  Maaike
 
   Date: Mon, 26 Jan 2015 11:23:59 -0500
   From: gr...@nmr.mgh.harvard.edu
   To: freesurfer@nmr.mgh.harvard.edu
   Subject: Re: [Freesurfer] average values per cluster
  
  
   We get these kind of reports occasionally. When I ask people to confirm
   that they use exactly the same design matrix in SPSS, I never hear 
  back,
   so I assume that it gets resolved. So please check. The other thing you
   can do is to run in matlab, something like
  
   cd glmdir/contrast
   X = load('Xg.dat');
   y = load('ocn.dat');
   C = load('C.dat');
   [beta rvar] = fast_glmfit(y,X);
   [F p] = fast_fratio(beta,X,rvar,C);
   p will be the p-value
  
   If you used an unsigned cluster-forming threshold (ie, abs), then it is
   possible that some of the voxels are pos and some are neg so that they
   average out
  
   doug
  
  
  
  
   On 01/26/2015 09:03 AM, maaike rive wrote:
Dear Freesurfer experts,
   
Sorry to bother you again, but I have two more questions about
extracting (thickness/surface/GI) values from a certain cluster.
   
As I understood, the abs.y.ocn.dat file gives the average values 
  for a
given significant cluster (e.g. a cluster where there is a 
  significant
AxB interaction).
   
I may be completely misunderstanding things, but if I use these 
  values
in SPSS for further statistics and test the same interaction (AxB),
than according to SPSS this interaction is /not /significant
(corrected for the same covariates as in the FSGD file).
   
Could you tell me what is going wrong here? I do not trust my 
  results now.
   
Furthermore, is it possible (and if so, how?) to extract the average
values of exactly the same cluster, but in an independent group not
used in the analysis, for post-hoc comparisons in SPSS?
   
Thanks,
   
Maaike
   
   
___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
  
   --
   Douglas N. Greve, Ph.D.
   MGH-NMR Center
   gr...@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/
  
   ___
   Freesurfer mailing list
   Freesurfer@nmr.mgh.harvard.edu
   https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
  
  
   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.
  
 
 
  ___
  Freesurfer mailing list
  Freesurfer@nmr.mgh.harvard.edu
  https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
 
 -- 
 Douglas N. Greve

Re: [Freesurfer] average values per cluster

[maaike rive]

Hi Douglas,
I checked but as far as I see it I used the same models. I attached the SPSS 
model and output as wel as the FSGD file and contrast file (to test a diagnosis 
x age interaction). Now that I come to think of it, I think the reason for the 
discrepant findings is that I used a t-contrast instead of an F-contrast for 
the interaction in Freesurfer. Apologies I did not think about this earlier. 
But maybe there is something else I did completely wrong. 
What do you think?
If it is indeed the t vs F contrast, than how should I specify that the 
contrast is an F-contrast? Is it ok to just add another line to the 
contrastfile (so0 0 0 0 0 0 0 0 1 1 1 1 -1 -1 -1 -10 0 0 0 0 0 0 0 -1 -1 -1 -1 
1 1 1 1)?
Thanks again,
Maaike
 Date: Mon, 26 Jan 2015 11:23:59 -0500
 From: gr...@nmr.mgh.harvard.edu
 To: freesurfer@nmr.mgh.harvard.edu
 Subject: Re: [Freesurfer] average values per cluster
 
 
 We get these kind of reports occasionally. When I ask people to confirm 
 that they use exactly the same design matrix in SPSS, I never hear back, 
 so I assume that it gets resolved. So please check. The other thing you 
 can do is to run in matlab, something like
 
 cd glmdir/contrast
 X = load('Xg.dat');
 y = load('ocn.dat');
 C = load('C.dat');
 [beta rvar] = fast_glmfit(y,X);
 [F p] = fast_fratio(beta,X,rvar,C);
 p will be the p-value
 
 If you used an unsigned cluster-forming threshold (ie, abs), then it is 
 possible that some of the voxels are pos and some are neg so that they 
 average out
 
 doug
 
 
 
 
 On 01/26/2015 09:03 AM, maaike rive wrote:
  Dear Freesurfer experts,
 
  Sorry to bother you again, but I have two more questions about 
  extracting (thickness/surface/GI) values from a certain cluster.
 
  As I understood, the abs.y.ocn.dat file gives the average values for a 
  given significant cluster (e.g. a cluster where there is a significant 
  AxB interaction).
 
  I may be completely misunderstanding things, but if I use these values 
  in SPSS for further statistics and test the same interaction (AxB), 
  than according to SPSS this interaction is /not /significant 
  (corrected for the same covariates as in the FSGD file).
 
  Could you tell me what is going wrong here? I do not trust my results now.
 
  Furthermore, is it possible (and if so, how?) to extract the average 
  values of exactly the same cluster, but in an independent group not 
  used in the analysis, for post-hoc comparisons in SPSS?
 
  Thanks,
 
  Maaike
 
 
  ___
  Freesurfer mailing list
  Freesurfer@nmr.mgh.harvard.edu
  https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
 
 -- 
 Douglas N. Greve, Ph.D.
 MGH-NMR Center
 gr...@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/
 
 ___
 Freesurfer mailing list
 Freesurfer@nmr.mgh.harvard.edu
 https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
 
 
 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.
 

  

FSGD_diagnbystate_ex.txt~ copy
Description: Binary data
0 0 0 0 0 0 0 0 1 1 1 1 -1 -1 -1 -1


SPSS model and output.docx
Description: MS-Word 2007 document
___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Re: [Freesurfer] average values per cluster

[maaike rive]

No not yet; I will do so!

 Date: Thu, 29 Jan 2015 13:44:33 -0500
 From: gr...@nmr.mgh.harvard.edu
 To: freesurfer@nmr.mgh.harvard.edu
 Subject: Re: [Freesurfer] average values per cluster
 
 
 The t is correct. When the contrast matrix only has one row, then the F 
 is just an unsigned t. I don't know anything about SPSS so I can't tell 
 from what you have sent whether it is the same model or not. Try to get 
 SPSS to output the design matrix. Did you try the matlab code below?
 
 On 01/29/2015 01:40 PM, maaike rive wrote:
  Hi Douglas,
 
  I checked but as far as I see it I used the same models. I attached 
  the SPSS model and output as wel as the FSGD file and contrast file 
  (to test a diagnosis x age interaction). Now that I come to think of 
  it, I think the reason for the discrepant findings is that I used a 
  t-contrast instead of an F-contrast for the interaction in Freesurfer. 
  Apologies I did not think about this earlier. But maybe there is 
  something else I did completely wrong.
 
  What do you think?
 
  If it is indeed the t vs F contrast, than how should I specify that 
  the contrast is an F-contrast? Is it ok to just add another line to 
  the contrastfile (so
  0 0 0 0 0 0 0 0 1 1 1 1 -1 -1 -1 -1
  0 0 0 0 0 0 0 0 -1 -1 -1 -1 1 1 1 1)?
 
  Thanks again,
 
  Maaike
 
   Date: Mon, 26 Jan 2015 11:23:59 -0500
   From: gr...@nmr.mgh.harvard.edu
   To: freesurfer@nmr.mgh.harvard.edu
   Subject: Re: [Freesurfer] average values per cluster
  
  
   We get these kind of reports occasionally. When I ask people to confirm
   that they use exactly the same design matrix in SPSS, I never hear 
  back,
   so I assume that it gets resolved. So please check. The other thing you
   can do is to run in matlab, something like
  
   cd glmdir/contrast
   X = load('Xg.dat');
   y = load('ocn.dat');
   C = load('C.dat');
   [beta rvar] = fast_glmfit(y,X);
   [F p] = fast_fratio(beta,X,rvar,C);
   p will be the p-value
  
   If you used an unsigned cluster-forming threshold (ie, abs), then it is
   possible that some of the voxels are pos and some are neg so that they
   average out
  
   doug
  
  
  
  
   On 01/26/2015 09:03 AM, maaike rive wrote:
Dear Freesurfer experts,
   
Sorry to bother you again, but I have two more questions about
extracting (thickness/surface/GI) values from a certain cluster.
   
As I understood, the abs.y.ocn.dat file gives the average values 
  for a
given significant cluster (e.g. a cluster where there is a 
  significant
AxB interaction).
   
I may be completely misunderstanding things, but if I use these 
  values
in SPSS for further statistics and test the same interaction (AxB),
than according to SPSS this interaction is /not /significant
(corrected for the same covariates as in the FSGD file).
   
Could you tell me what is going wrong here? I do not trust my 
  results now.
   
Furthermore, is it possible (and if so, how?) to extract the average
values of exactly the same cluster, but in an independent group not
used in the analysis, for post-hoc comparisons in SPSS?
   
Thanks,
   
Maaike
   
   
___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
  
   --
   Douglas N. Greve, Ph.D.
   MGH-NMR Center
   gr...@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/
  
   ___
   Freesurfer mailing list
   Freesurfer@nmr.mgh.harvard.edu
   https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
  
  
   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.
  
 
 
  ___
  Freesurfer mailing list
  Freesurfer@nmr.mgh.harvard.edu
  https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
 
 -- 
 Douglas N. Greve, Ph.D.
 MGH-NMR Center
 gr...@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/
 
 ___
 Freesurfer mailing list
 Freesurfer

[Freesurfer] average values per cluster

[maaike rive]

Dear Freesurfer experts,
Sorry to bother you again, but I have two more questions about extracting 
(thickness/surface/GI) values from a certain cluster.
As I understood, the abs.y.ocn.dat file gives the average values for a given 
significant cluster (e.g. a cluster where there is a significant AxB 
interaction).
I may be completely misunderstanding things, but if I use these values in SPSS 
for further statistics and test the same interaction (AxB), than according to 
SPSS this interaction is not significant (corrected for the same covariates as 
in the FSGD file). 
Could you tell me what is going wrong here? I do not trust my results now.
Furthermore, is it possible (and if so, how?) to extract the average values of 
exactly the same cluster, but in an independent group not used in the analysis, 
for post-hoc comparisons in SPSS?
Thanks,
Maaike___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

[Freesurfer] Extracting vertexvalues

[maaike rive]

Dear Freesurfer experts,

 

I have a question about extracting vertex values of VtxMax for each subject 
(for surface eara, in this case).

Based on earlier postings, I used the command mri_segstats (--seg 
${x}/surf/lh/area.pial --i  ${x}/surf/lh/area.pial --vox vertexno 0 0). 

 

First, I'm not exactly sure whether this is correct and second, I get 2 columns 
whereas I expected 1 column, so I wonder, what do these two values per subject 
mean?

 

Thanking you in advance,

 

Maaike

 

 
  ___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Re: [Freesurfer] Extracting vertexvalues

[maaike rive]

Thanks!

 Date: Mon, 5 Jan 2015 14:57:12 -0500
 From: gr...@nmr.mgh.harvard.edu
 To: freesurfer@nmr.mgh.harvard.edu
 Subject: Re: [Freesurfer] Extracting vertexvalues
 
 
 It is reporting both the value at that vertex and the mean of all the 
 other vertices. To get only the value for the vertex, add --id 1
 doug
 
 
 On 01/05/2015 03:52 AM, maaike rive wrote:
  Dear Freesurfer experts,
 
  I have a question about extracting vertex values of VtxMax for each 
  subject (for surface eara, in this case).
  Based on earlier postings, I used the command mri_segstats (--seg 
  ${x}/surf/lh/area.pial --i  ${x}/surf/lh/area.pial --vox vertexno 0 0).
 
  First, I'm not exactly sure whether this is correct and second, I get 
  2 columns whereas I expected 1 column, so I wonder, what do these two 
  values per subject mean?
 
  Thanking you in advance,
 
  Maaike
 
 
 
 
  ___
  Freesurfer mailing list
  Freesurfer@nmr.mgh.harvard.edu
  https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
 
 -- 
 Douglas N. Greve, Ph.D.
 MGH-NMR Center
 gr...@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/
 
 ___
 Freesurfer mailing list
 Freesurfer@nmr.mgh.harvard.edu
 https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
 
 
 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.
 
  ___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

[Freesurfer] DODS design and demeaning

[maaike rive]

Dear Freesurfer experts,
 
I would be very greatful if you could help me with three (probably basic) 
questions regarding an mri_glmfit design.
 
For the design and contrasts I used FSGDF. I have a 2x2 factorial design with 
one covariate (age). I specified all four groups (A1B1, A1B2, A2B1, A2B2) and 
the covariate in one model, because I would like to test for the interaction 
between A and B (regressing out the effects of age). Age was demeaned using the 
mean of all subjects (i.e. of all 4 groups).
 
My first question regards the contrasts:
If, besides the interaction and main affects of A and B, I would like to assess 
the difference between individual groups, for example A1B1 - A1B2, is it okay 
to use the same model (1 -1 0 0 + further 0's for the age regressors); or 
should I make a new FSGDF containing only the subjects of group A1B1 and A1B2? 
 
My second question is associated with the first:
If it is allowed to use the same model, how does freesurfer deal withdemeaning? 
Since I used the mean of all subjects to demean and not just of the subjects of 
for example A1B1 and A1B2. 
 
My third question regards the DODS/DOSS issue.
I used DODS, because I do not believe the slopes of age by volume (or age by 
area/thickness) will be exactly parallel between groups. I tested this also be 
using contrasts of the age regressors, (for example using the contrast 0 0 0 0 
1 -1 -1 1 for a AxBxage interaction) and indeed there are interactions with age 
in some brain areas. For thickness, they were FDR corrected still significant, 
for other measeurs the were not significant after correction, but that seems 
natural given the sample sizes. No I wondered whether it would suffice to use a 
DODS design (since slopes differ between groups) or whether I should stratify 
groups further for age (for example use two models, one for young subjects, on 
for old subjects). If the latter is the case, I would end up with very small 
samples, I am afraid.
 
Thanking you in advance,
 
Maaike___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.

Re: [Freesurfer] DODS design and demeaning

[maaike rive]

Hi Doug,
Thank you for your quick response. Concerning question3, do you mean that if 
there is an interaction with age, it is all right to use DODS (without 
stratification)?
Thanks, Maaike

Date: Wed, 24 Dec 2014 15:31:53 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] DODS design and demeaning


  

  
  


On 12/24/14 5:21 AM, maaike rive wrote:



  
  
Dear Freesurfer experts,

 

I would be very greatful if you could help me with
  three (probably basic) questions regarding an mri_glmfit
  design.

 

For the design and contrasts I used FSGDF. I have a
  2x2 factorial design with one covariate (age). I specified all
  four groups (A1B1, A1B2, A2B1, A2B2) and the covariate in one
  model, because I would like to test for the interaction
  between A and B (regressing out the effects of age). Age was
  demeaned using the mean of all subjects (i.e. of all 4
  groups).

 

My first question regards the contrasts:

If, besides the interaction and main affects of A
  and B, I would like to assess the difference between
  individual groups, for example A1B1 - A1B2, is it okay to use
  the same model (1 -1 0 0 + further 0's for the age
  regressors); or should I make a new FSGDF containing only the
  subjects of group A1B1 and A1B2? 

  

It is a tricky question. You will get the same group means
regardless of whether you combine all subjects or use a 2nd FSGD.
The difference will be that you will have a much higher DOF if you
combine them all. This means that the noise in the other subjects
informs you about the noise in those two groups. If you can justify
this, then you can leave them combined. If the 2nd model gives you
the results you want, then reporting that would be a stronger
scientific result.


  
 

My second question is associated with the first:

If it is allowed to use the same model, how does
  freesurfer deal withdemeaning? Since I used the mean of all
  subjects to demean and not just of the subjects of for example
  A1B1 and A1B2. 

  

The demeaning is also tricky. I would actually test whether there is
an interaction between group and age. If there is no interaction,
then I would re-run using a DOSS model in which case the demeaning
won't make a difference.


  
 

My third question regards the DODS/DOSS issue.

I used DODS, because I do not believe the slopes of
  age by volume (or age by area/thickness) will be exactly
  parallel between groups. I tested this also be using contrasts
  of the age regressors, (for example using the contrast 0 0 0
  0 1 -1 -1 1 for a AxBxage interaction) and indeed there are
  interactions with age in some brain areas. For thickness, they
  were FDR corrected still significant, for other measeurs the
  were not significant after correction, but that seems natural
  given the sample sizes. No I wondered whether it would suffice
  to use a DODS design (since slopes differ between groups) or
  whether I should stratify groups further for age (for example
  use two models, one for young subjects, on for old subjects).
  If the latter is the case, I would end up with very small
  samples, I am afraid.

  

If  there is no interaction in the areas that show up in your
contrast of interest, then it is safe to use DOSS. If there are,
then I don't think that stratifying them solves the problem.



doug


  
 

Thanking you in advance,

 

Maaike

  
  

  
  

  ___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer



  


___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


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.___
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer


The information in this e-mail is intended only for the person to whom