Re: [Freesurfer] longitudinal QDEC help

Hi Taylor,

You can do the glmfit osgm on whatever you like (rate, percentchange pc1, or symm. percent change spc). I recommend SPC. The 2stage model page shows an example of glmfit with osgm flag and explains how to stack your data first.  Best Martin

Sent via my smartphone, please excuse brevity.

<div>-------- Original message --------</div><div>From: Taylor North taylornorth189@hotmail.ca </div><div>Date:09/04/2014 11:14 PM (GMT-05:00) </div><div>To: Freesurfer support list freesurfer@nmr.mgh.harvard.edu </div><div>Subject: Re: [Freesurfer] longitudinal QDEC help </div><div> </div>Hi Martin

Yes I that definitely clears things up, thanks very much for the help.

As for the QCAHCE we did use the long_mris_slopes described in the tutorial you linked in your email. Just to confirm I believe I should be using the mri_glmfit -osgm model for my one sample group mean analysis and then when I want to compare against my continuous/dependent variables I should be using QDEC? Also it looks like mri_glmfit -osgm calculates % change, so should I be analyzing % change, not symmetrized percent change on my QDEC analysis as well, even though it say symmetrized percent change is the more robust result?

Thanks again for your help

Taylor From: mreuter@nmr.mgh.harvard.edu Date: Thu, 4 Sep 2014 22:44:16 -0400 To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] longitudinal QDEC help

Hi Taylor,

first, in your description you mention 'base and 6 month'. Better call it 'baseline' or 'first and second time points at 0 and 6 months' in order to not confuse the first time point with the 'base' = within-subject template created as the second step. I think you did not confuse it, but just want to make sure.

From your description it is not clear how you computed the symmetrized percent change. You say you ran QCACHE, but how? Did you use long_mris_slopes as described here: http://freesurfer.net/fswiki/LongitudinalTwoStageModel http://freesurfer.net/fswiki/FsTutorial/LongitudinalTutorial

That step computes slopes for each subject and (instead of having 2 thickness maps) reduces the information to a single map (here the symmetrized percent change is simply: 100 * tp2-tp1/(0.5*(tp1+tp2)) ) . These maps are stored in the base directories (one for each subject and of corse each hemisphere that you run).

Now to your questions: 1) I think a 'one sample group mean' analysis is not possible in QDEC. For that you have to use mri_glmfit -osgm …. I wish we had this functionality and I'll see how hard it would be to implement it in future versions of QDEC (this came up frequently and it should not be too hard).

2) a) the longitudinal qdec table use the cross id's in the fsid column for simplicity. When it is used in many programs (e.g. long_mris_slopes, long_stats_slopes and other scripts with the long-qdec option) the scripts knows that it is a longitudinal qdec table (because of the existing fsid-base column) and internally use the longitudinal directories for processing. The cross sectional directories are never used. b) as mentioned above, the results from the first stage analysis (creation of the within subject slopes and percent changes) are stored in the base directories. They are used in this step and found there. (and you are right, the recon-all results from the base should not be analyzed directly, but here the data that is used is derived from the longitudinal directories and just placed in the base so that it can be found for each subject).

Hope that explains it. Best, Martin

On Sep 4, 2014, at 9:33 PM, Taylor North taylornorth189@hotmail.ca wrote:

Hello,

I am having some troubles using QDEC to analyze the cortical thickness changes in my group study and I was hoping someone may be able to give me a hand. The steps I have done are as follows:

I am working on a study looking at 40 subjects scanned at two time points (base and 6 months after). I have followed all the steps for longitudinal processing for each subject (cross,base, and longitudinals) and made edits where they were necessary to the base. I then entered all 40 subjects into a QDEC table as follows

fsid fsid-base time gender age crosstpN1 base1 0 female 20 crosstpN2 base1 6 female 20 .......... continues for rest of the subjects

then I ran the QCHACHE on the data, made a Qdecrc. file in the subject DIR, and then transformed the longitudinal QDEC table into a QDEC table which looks like:

fsid time gender age Base1 3 female 20 .......... continues for the rest of the subjects.

I then followed the QDEC steps outlined In the group tutorial. (import cross sectional table, generate stats data table, import aseg volume, set longitudinal dependent variables and pick continuous variable.

1) the first problem I am having is that when I go to process the longitudinal group results through QDEC it makes me pick a continuous variable when I just want to analyze the dependent variable The symmetrized percent change. So I have tried picking different continuous variables ( such as age or importing rhcortex volume etc...) with choosing symmetrized percent change as the dependent variable. No matter what continuous variable I pick the question: ``does the average symmetrized percent change differ from zero"? always comes up, the problem is depending on what continuous variable I pick the results change. Is there a way to just look at the longitudinal thickness changes across a group alone, because I first want to look at the atrophy across the whole group over 6 months and then compare the differences in atrophy against the diagnosed and non diagnosed group.

2) another thing I don't quite understand is it seems like the longitudinal qdec tables are made with the cross of the subjects under fsid, but why wouldn't you use the longitudinals under the fsid for the longitudinal table instead? Also it seems like in the cross sectional qdec table the base is the subject being analyzed under fsid, which also doesn't make sense to me because in the tutorial it says base should not be used directly for analysis.

thanks for the help!

Taylor

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

--------------------------------- Dr. Martin Reuter Assistant in Neuroscience - Massachusetts General Hospital Instructor in Neurology - Harvard Medical School MGH / HMS / MIT

A.A.Martinos Center for Biomedical Imaging 149 Thirteenth Street, Suite 2301 Charlestown, MA 02129

Phone: +1-617-724-5652 Email: mreuter@nmr.mgh.harvard.edu reuter@mit.edu Web : http://reuter.mit.edu

_______________________________________________ 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.