Thanks very much Martin and Bruce. I guess I'd misread the Wiki (my own fault, not the text's), and am glad to hear that the longitudinal pipeline is in fact perfectly suitable for my needs here.

Having run the first 2 steps (cross and base), I'm a bit unclear how the output so far has to be manually inspected. It says in the tutorial that you should load each subject's base volume + surfs in freeview and then "move back and forth a few slices". However, even just loading each base in this manner takes ~1 min on my PC (CoreDuo, 4GB, Ubuntu Virtualbox in Windows 7), and then moving with PgUp/PgDn between all coronal slices (starting from the default slice=128, going all the way posterior and then all the way anterior) is excruciatingly slow. All of this would have to be repeated for all my 72 subjects - is there any way to optimise this manual inspection?

Also, if the surfs turn out to not follow the volume correctly, presumably the thing to do is white surface correction + re-running recon. But what should one do if, due to an erroneous averaging between timepoints, you see blurs/ghosts in your base template?

Many thanks!
Tudor

On 9 May 2014 21:33, Martin Reuter <mreuter@nmr.mgh.harvard.edu> wrote:

Hi Tudor,

the longitudinal pipeline in FS is actually one of the best on the planet as far as I know :-). If there is any contradictory information on the wiki, can you point me to that so I can see what causes the misconception. Really: compared to independent processing, it significantly increases sensitivity. Also we have designed it to be unbiased with respect to a single time point or directionality. It is quite mature by now.

You should definitely use the longitudinal pipeline for the analysis of your data. Now to your questions

1. QDEC: I am not too familiar with qdec. You can definitely try the 2-stage approach described on the wiki. There you first compute a measure of change (e.g. hippocampal volume change during your week) and then compare that measure across groups similar to a cross sectional volume/thickness analysis. You can also use our tools to run a linear mixed effects model if you want to do that (it is more involved and requires you to use matlab tools). In your case, you probably have 2 time points for all subjects and the time distance is probably the same for all subjects, so the 2-stage approach should be fine.

2. The image processing is done via the longitudinal pipeline (three steps: cross, base, long), to prepare the data look at the description of the 2-stage model
http://freesurfer.net/fswiki/LongitudinalTwoStageModel
and also the longitudinal tutorial
http://freesurfer.net/fswiki/FsTutorial/LongitudinalTutorial

3. At the recon all level in FS you get (after the 3 steps) measurement for all time points. So you would compare those results across time in the stats.

Hope that helps, Martin

On 05/08/2014 08:14 AM, Tudor Popescu wrote:

Sorry for the repeat, wasn't sure whether this was received the first time.
Tudor

On 6 May 2014 19:55, Tudor Popescu <tudor3@gmail.com> wrote:

Dear FS list,

I have structural data from a learning study (pre&post-training scans, with 3 groups). Although the training was only one week, I'm guessing from an analysis point of view, this still qualifies as longitudinal. I want to check for

• the main within-subjects effect of time point (pre&post)
• the main between-subjects effect of group (treatment A, treatment B, control),
• the time x group interaction

I intend to look at thickness, surface area, volume, and lGI.

I read on the wiki that FS is currently not optimal for longitudinal analyses. I intend my FreeSurfer analysis to supplement a VBM analysis done in FSL. In case it is in fact a good idea to do this, my questions (not covered in the 'longitudinal' wiki page) are:

1) Can QDEC be used for such an analysis, and if so, what would be different as compared to a cross-sectional (no temporal/within factor) study?

2) Also, is the pre-processing stage any different?

3) In FSL, for longitudinal designs you do stats on images obtained as the difference between consecutive time points. Does this have to be done in FreeSurfer as well, and if so, is it done at the recon-all level or only at the stats (QDEC) level?

Thanks!

Tudor

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

-- 
Martin Reuter, Ph.D.

Instructor in Neurology
  Harvard Medical School
Assistant in Neuroscience
  Dept. of Radiology, Massachusetts General Hospital
  Dept. of Neurology, Massachusetts General Hospital
Research Affiliate
  Computer Science and Artificial Intelligence Lab,
  Dept. of Electrical Engineering and Computer Science,
  Massachusetts Institute of Technology

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.