External Email - Use Caution
Hello,
and thanks for the explanation!
Here's my first impression:
I do not think that there is a really satisfying solution given these
particular stats maps and the effect size measure you describe.
I may be missing something, but I also have to admit that it is not immediately
clear to me what motivates the choice of cohen's d (and the calculation you
describe) in the current scenario - I've always thought of it as a measure of
'difference in means'.
In a regression context, I would personally look for a measure of the
"variance-explained" type. There seem to be at least two papers (Xu, 2003,
Statistics in Medicine, and Selya, 2012, Frontiers) that cover these for mixed
effects models. The calculation/implementation of these measures seems to be,
however, not as straightforward as for classical multiple regression models;
and as far as I can see, it would also involve fitting Null or reduced models,
and thus running analyses again.
By far the easiest and most straightforward way in my eyes is to report
unstandardized effect sizes, i.e. beta estimates, as you have clearly
interpretable main variables (group and time) with intuitive units that should
be comparable across studies from your field. So maybe this is an argument not
to standardize, and to state the effect size in terms of "change in thickness
per year" etc.?
Best regards,
Kersten
On Do, 2019-11-28 at 11:29 +1100, Bronwyn Overs wrote:
Hi Kersten,
No problem at al and thank you for your reply.
While running my correction for multiple comparisons I store the df associated
with each contrast vector using the following:
[~,~,dflh] = find(F_lhstats.df(2,:));
[~,~,dfrh] = find(F_rhstats.df(2,:));
dfmodelh = {floor(mode(dflh))};
dfmoderh = {floor(mode(dfrh))};
So far the dfmodelh and dfmoderh have been identical. I then calculate the
effect size for each cluster using the following steps:
1. Derive the p-value from the Max threshold ('Max'), where p = 10 to the power
of -abs(Max)
2. Calculate the t-value associated with this p-value, using the relevant df
for this contrast
3. Calculate cohen's d using 't' and 'df', where d = (t*2)/(sqrt(df))
When the Max threshold is listed as -inf I cannot complete these calculations.
Can you suggest an alternate method for generating effect sizes for these
clusters?
Kind regards,
Bronwyn Overs
Research Assistant
[cid:1574977195.32275.38.camel@dzne.de]
Neuroscience Research Australia
Margarete Ainsworth Building
Barker Street Randwick Sydney NSW 2031 Australia
M 0411 308 769 T +61 2 9399 1725
neura.edu.au <http://neura.edu.au/>
Twitter<https://twitter.com/neuraustralia> |
Facebook<https://www.facebook.com/NeuroscienceResearchAustralia> |
Subscribe<http://www.neura.edu.au/help-research/subscribe>
________________________________
From: "Kersten Diers, DZNE" <kersten.di...@dzne.de>
To: "Freesurfer support list" <freesurfer@nmr.mgh.harvard.edu>
Sent: Thursday, November 28, 2019 12:29:18 AM
Subject: Re: [Freesurfer] Max threshold marked as '-inf' in cluster summary file
External Email - Use Caution
Hi Bronwyn, Doug,
sorry - I missed the other mail from two weeks ago (and have been absent for a
few days), but I'm able to take a look now.
You mention that your goal is to calculate effect sizes. Could you briefly let
us know how you are planning to do it - this would help me to better understand
the issue.
Thanks,
Kersten
On Di, 2019-11-26 at 23:14 +0000, Greve, Douglas N.,Ph.D. wrote:
can you send the cluster summary file?
On 11/24/2019 7:09 PM, Bronwyn Overs wrote:
External Email - Use Caution
Hi Kersten,
Following on from Douglas's reply, you may remember my analysis as I forwarded
you some files in October (see
https://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg64148.html). I
followed your suggestions of removing the '--thmax 5' argument form
'mri_surfcluster', and now the labels generated by 'mri_label2label' overlap
perfectly with the significance maps from the LME. However, now my largest
clusters have a maximum threshold of '-inf' in my cluster summary files. Do you
have any idea how to get around this problem so that I can use my max threshold
values to calculate cluster-wise effect sizes?
Kind regards,
Bronwyn Overs
Research Assistant
[cid:1574977195.32275.38.camel@dzne.de]
Neuroscience Research Australia
Margarete Ainsworth Building
Barker Street Randwick Sydney NSW 2031 Australia
M 0411 308 769 T +61 2 9399 1725
neura.edu.au <http://neura.edu.au/>
Twitter<https://twitter.com/neuraustralia> |
Facebook<https://www.facebook.com/NeuroscienceResearchAustralia> |
Subscribe<http://www.neura.edu.au/help-research/subscribe>
________________________________
From: "Greve, Douglas N.,Ph.D."
<dgr...@mgh.harvard.edu><mailto:dgr...@mgh.harvard.edu>
To: "Freesurfer support list"
<freesurfer@nmr.mgh.harvard.edu><mailto:freesurfer@nmr.mgh.harvard.edu>,
"Kersten Diers, DZNE" <kersten.di...@dzne.de><mailto:kersten.di...@dzne.de>
Sent: Thursday, November 14, 2019 4:23:12 AM
Subject: Re: [Freesurfer] Max threshold marked as '-inf' in cluster summary file
I'll have to leave this one for Kersten as it appears that the LME stuff
is generating and inf value
On 11/12/19 7:05 PM, Bronwyn Overs wrote:
>
> External Email - Use Caution
>
> Hi Douglas,
>
> Yes I have looked at it in Freeview and I have attached a screenshot
> for your reference.
>
> The /fsdata/lme/thickness/rh.thickness.B1B7.mgh file was generated
> with a mass-univariate spatiotemporal model using
> 'lme_mass_fit_EMinit' and then FDR correction was applied across both
> hemispheres (lme_mass_FDR). I have provided analysis details and
> syntax below. The B1B7 contrast represents the affect of years in cases.
>
> Sample:
> Our sample includes 112 controls subjects, and 106 cases. All subjects
> are aged between 12 and 30 years. 153 or these subjects have 2
> time-points (77 control, 76 cases), while the remaining 65 individuals
> have only 1 MRI time-point. We also have mixed ethnicites - 165
> Caucasians, 23 Asians, and 30 mixed (Asians-Caucasians).
>
> The QDEC file contains the following 6 variables:
> 1. Y (years between scans)
> 2. A (baseline age)
> 3. G (group, 1=case, 0=control)
> 4. S (sex, 1=female, 0=male)
> 5. E1 (Ethnicity 1, 1=asian, 0=other)
> 6. E2 (Ethnicity 1, 1=mixed-asian-caucasian, 0=other)
>
> Design matrix: [ones(length(M),1) M M(:,1).*M(:,3)]
> i.e. main effects for each of the qdec variables + an
> interaction term for years X group
>
> DVs: Cortical thickness, area and volume
>
> Model: Mass-univariate spatiotemporal model using
> 'lme_mass_fit_EMinit'
> CODE:
> % Read in surface files
> [Y,mri] = fs_read_Y(mgh);
> % Read in qdec file
> Qdec = fReadQdec(qdec);
> % Remove fsid from qdec
> Qdec = rmQdecCol(Qdec,1);
> % Store col 1 (fsid-base) in sID variable
> sID = Qdec(2:end,1);
> % Remove col 1 (fsid-base) from Qdec array
> Qdec = rmQdecCol(Qdec,1);
> % Convert Qdec to numeric matrix M
> M = Qdec2num(Qdec);
> % Sort data and evaluate design matrix
> [M,Y,ni] = sortData(M,1,Y,sID);
> X = eval([ones(length(M),1) M M(:,1).*M(:,3)])
> % Compute vertex-wise temporal covariance estimates.
> [Th0, Re] = lme_mass_fit_EMinit(X,[1],Y,ni,cortex,3);
> %Segmentation and model fitting.
> [Rgs, RgMeans, stats] = fit(Th0, Re, [1], sphere, cortex, X, Y,
> ni);
> %Check surfaces.
> surfcomp(Th0, RgMeans, sphere, fig1, fig2)
>
> Correction for multiple comparisons: FDR across both hemispheres
> CODE:
> P = [ F_lhstats.pval(lhcortex) F_rhstats.pval(rhcortex) ]; G = [
> F_lhstats.sgn(lhcortex)
> F_rhstats.sgn(rhcortex) ];
> [detvtx, sided_pval, pth] = lme_mass_FDR2(P,G,[],0.05,0);
> altfdr(r,2) = num2cell(abs(log10(lme_mass_FDR(P,0.05))));
> pcor = -log10(pth);
> [~,~,dflh] = find(F_lhstats.df(2,:));
> [~,~,dfrh] = find(F_rhstats.df(2,:));
> dfmodelh(r,2) = {floor(mode(dflh))};
> dfmoderh(r,2) = {floor(mode(dfrh))};
> thrlh(r,2) = {pcor};
> thrrh(r,2) = {pcor};
> [~,dc] = size(detvtx);
> dvtx(r,2) = {dc};
>
> Kind regards,
>
>
> Bronwyn Overs
>
> Research Assistant
>
>
> Neuroscience Research Australia
> Margarete Ainsworth Building
> Barker Street Randwick Sydney NSW 2031 Australia
> *M* 0411 308 769 *T* +61 2 9399 1725
>
>
> neura.edu.au <http://neura.edu.au/><http://neura.edu.au/>
>
> Twitter
> <https://twitter.com/neuraustralia><https://twitter.com/neuraustralia> |
> Facebook
> <https://www.facebook.com/NeuroscienceResearchAustralia><https://www.facebook.com/NeuroscienceResearchAustralia>
> | Subscribe
> <http://www.neura.edu.au/help-research/subscribe><http://www.neura.edu.au/help-research/subscribe>
>
>
>
>
> ------------------------------------------------------------------------
> *From: *"Greve, Douglas N.,Ph.D."
> <dgr...@mgh.harvard.edu><mailto:dgr...@mgh.harvard.edu>
> *To: *"Freesurfer support list"
> <freesurfer@nmr.mgh.harvard.edu><mailto:freesurfer@nmr.mgh.harvard.edu>
> *Sent: *Wednesday, November 13, 2019 2:44:25 AM
> *Subject: *Re: [Freesurfer] Max threshold marked as '-inf' in cluster
> summary file
>
> It looks like you have one gigantic cluster with a value of -inf at
> vertex 41. How did you generate
> /fsdata/lme/thickness/rh.thickness.B1B7.mgh ? Have you looked at it in
> freeview?
>
> On 11/11/2019 7:39 PM, Bronwyn Overs wrote:
>
> External Email - Use Caution
>
> Dear Freesurfer Mailing List,
>
> I have completed an LME analysis in matlab and have generated a
> number of different cluster summary files for my various
> contrasts. In many of resulting cluster summary files the 'Max'
> value for the largest cluster which encompasses most of the brain
> surface is reported as infinity ('-inf'). Please find an example
> file attached. As I was hoping to use these values to generate
> effect size estimates, can you please tell me what alternate value
> I could substitute here for 'Max'?
>
> Kind regards,
>
>
> Bronwyn Overs
>
> Research Assistant
>
>
> Neuroscience Research Australia
> Margarete Ainsworth Building
> Barker Street Randwick Sydney NSW 2031 Australia
> *M* 0411 308 769 *T* +61 2 9399 1725
>
>
> neura.edu.au <http://neura.edu.au/><http://neura.edu.au/>
>
> Twitter
> <https://twitter.com/neuraustralia><https://twitter.com/neuraustralia> |
> Facebook
>
> <https://www.facebook.com/NeuroscienceResearchAustralia><https://www.facebook.com/NeuroscienceResearchAustralia>
> |
> Subscribe
> <http://www.neura.edu.au/help-research/subscribe><http://www.neura.edu.au/help-research/subscribe>
>
>
>
>
> _______________________________________________
> Freesurfer mailing list
> Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu>
> <mailto:Freesurfer@nmr.mgh.harvard.edu><mailto:Freesurfer@nmr.mgh.harvard.edu>
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
>
>
>
> _______________________________________________
> Freesurfer mailing list
> Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu>
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
>
> _______________________________________________
> Freesurfer mailing list
> Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu>
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu>
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu>
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu<mailto: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
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
