Hi Christian,
you are interested in the longitudinal change in your study group. The problem of course is that you are missing a longitudinal control group. Control groups are so important.
With your data you can analyze if there is change in your study group (using the 2 stage model, since time interval seems to be approximately equal across subjects and all have 2 time points). You will find some changes in some areas, but it will be unclear if that is noise (from outliers), regular aging, hydration, motion etc. You'd have to compare to a longitudinal control group to test if the changes in the study group are different from the changes that happen in controls.
Since you are looking for small changes (4 month) cross sectional analysis will probably not help either (because of the inter subject variability). Comparing time point 1 with controls and comparing time point 2 with controls is not great, you could find a difference in one of the time points (or in both) with respect to the controls but maybe that is because of subject selection (gender, age or whatever) (+- noise to move it in and out of significance).
To answer your question, since you already computed the rate (or percent change) maps (i.e. the first stage of the model), you simply run a qdec analysis on that (instead of e.g. thickness) using a cross sectional qdec table with one line per subject (2nd stage). This is described here:
http://surfer.nmr.mgh.harvard.edu/fswiki/LongitudinalTwoStageModel http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/LongitudinalTutorial
Best, Martin
On 06/05/2014 02:49 AM, Rummel, Christian wrote:
Dear FreeSurfers,
we have MR scans from a group of healthy subjects (about 40, equal
number of males and females, age range 25 to 65 years), who were scanned
at two time points (about four months separation). In between they
experienced exposure to a condition that might or might not influence
cortical thickness or other morphometric measures. From other studies we
have a large number of healthy controls in the same age range. The
difference is that they were scanned only once and were not exposed to
any special condition.
We want to find out where cortical thickness changes significantly
stronger in our subjects than expected from the controls for normal
aging. We know how to do that for the averages over cortical
parcellations and sub-cortical segmentations as reported in the
aseg.stats and aparc.stats.
Is it possible to use qdec to do the same thing vertex-wise? The
symmetrized percent changes have been calculated already from
FreeSurfer's longitudinal stream (long_mris_slopes). How can we feed the
information about the locally expected change rate (and maybe even its
variability) into qdec?
Thanks,
Christian Rummel
Christian Rummel (PhD)
Support Center for Advanced Neuroimaging (SCAN)
University Institute for Diagnostic and Interventional Neuroradiology
Inselspital, 3010 Bern, Switzerland
Tel. 0041 31 6328038
Fax 0041 31 6324872
crummel@web.de
gpg encryption preferred
Key-ID 5BEB9DB8
Fingerprint 8A73 6894 5668 49EA 816C DEFA 78D0 BF87 5BEB 9DB8
Mit TouchDown von meinem Android-Telefon gesendet (www.nitrodesk.com)
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer