Hi Nadia,

the processing steps look OK.

You have several options to analyze the longitudinal data. We do not have a single tool to do everything automatically but there are many tools in FreeSurfer that can do the individual steps depending on what you want to do (as there are many different statistical analyses that people might want to do).

If you are interested in comparing e.g. the thickness: the thickness files are in register across time within subject (in the tp.long.base/surf directories). For 2 time points, you could compute the difference (mris_calc ) or percent change and divide by the distance in time (if it is different across subjects). For your case (>2 time points) one option is to compute a single statistic per subject in a first step (e.g. fit a slope through the 7 time points) and then compare the slopes across groups etc. (basically run a glm for each subject and then use mris_prepoc to register the results to some template, stack them and run a final glm to compare groups).

A better way to deal with longitudinal data is to use Linear Mixed Models, but for this you are on your own. You could load stuff into matlab and setup your mixed models there or convert the data into tables and load it into some other statistics package.

Hope that helps getting you started, best, Martin

On Wed, 2011-02-23 at 15:53 +0100, nadia merhoum wrote:

Hello

I have just started to use FreeSurfer (version 4.5) and have some questions on the longitudinal stream:

I have started to process cross-sectionally all time points (7 tp) of my patients with the following script:

recon-all–i tpidN.nii –subjid tpidN –all

Then, created an unbiased template from all time points for each subject and process it with recon-all with the following script:

recon-all -base <templateid> -tp <tp1id> -tp <tp2id> ... -all

and finally did the longitudinal process to all time points with the following script:

recon-all -long <tpNid> <templateid> -all

I would like to know if what I have been doing till this final step is correct or not?

Iwould like also to know how we could calculate the differences between tpid1.long.template and tpid2.long.template for example? And is there any automatic way to do so?

Thankyou for your help

Sincerely

Nadia

---

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